<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Norazha Paiman on Medium]]></title>
        <description><![CDATA[Stories by Norazha Paiman on Medium]]></description>
        <link>https://medium.com/@azhapaiman?source=rss-f278a8800248------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*0Q1FJBigC_M8Vhz1BHg7xQ.png</url>
            <title>Stories by Norazha Paiman on Medium</title>
            <link>https://medium.com/@azhapaiman?source=rss-f278a8800248------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 16:36:38 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@azhapaiman/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Stop Trying to Out-Design the Machine]]></title>
            <link>https://medium.com/@azhapaiman/stop-trying-to-out-design-the-machine-6eb0ffb823dc?source=rss-f278a8800248------2</link>
            <guid isPermaLink="false">https://medium.com/p/6eb0ffb823dc</guid>
            <category><![CDATA[assessment]]></category>
            <category><![CDATA[education]]></category>
            <category><![CDATA[genai]]></category>
            <category><![CDATA[university]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Norazha Paiman]]></dc:creator>
            <pubDate>Mon, 31 Aug 2026 15:05:49 GMT</pubDate>
            <atom:updated>2026-08-31T15:05:49.092Z</atom:updated>
            <content:encoded><![CDATA[<p><em>AI-proof assessment is a fantasy. The future belongs to assessment you can watch.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*J2oJQw_ybjuJWEgQ" /><figcaption>Photo by <a href="https://unsplash.com/@jdudenas?utm_source=medium&amp;utm_medium=referral">Julius Dūdėnas</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Last semester I killed the written report in the academic literacy course I coordinate. It had a decent rubric and a long history, and it died without ceremony, because I could no longer say with a straight face what it measured. A student who wrote it alone and a student who prompted it into existence in a fraction of the time handed me documents I could not tell apart. Neither could the software that promised to tell them apart. So I stopped pretending.</p><p>That is the argument of this piece, and I will not soften it. The hunt for the AI-proof assignment is over, and it was lost before most of us noticed it had started. What survives is assessment you can watch happen.</p><h3><strong>The fantasy of the uncheatable task</strong></h3><p>Every faculty meeting on this subject produces the same hopeful inventory. Make the prompt hyper-local. Require reflection. Ask for sources from the last six months. Demand a personal anecdote. Each of these is defeated within an hour of someone trying, because language models take instructions; that is the whole point of them. Any constraint you can write into a brief, a student can paste into a prompt.</p><p>The empirical record is blunt. In 2024 Peter Scarfe and colleagues at the University of Reading slipped fully AI-written scripts into the live examinations system of a psychology degree. <a href="https://doi.org/10.1371/journal.pone.0305354">Ninety-four per cent went undetected</a>, and the fakes outscored real students by about half a grade boundary. These were unsupervised online exams marked by staff who did not know they were being tested. That is the assessment most of us still run.</p><p>Detection does not rescue us. Debora Weber-Wulff and her team <a href="https://doi.org/10.1007/s40979-023-00146-z">tested fourteen detection tools</a> and concluded they were neither accurate nor reliable, with a bias toward calling machine text human. Worse, Weixin Liang and colleagues at Stanford found that popular detectors <a href="https://doi.org/10.1016/j.patter.2023.100779">flagged more than half of the essays written by non-native English speakers</a> as AI-generated while clearing native writers almost perfectly. In a Malaysian ESL classroom that is not a tool. It is a machine for accusing the wrong students.</p><h3><strong>Rules are not design</strong></h3><p>The next reflex is policy. Traffic lights, permitted-use scales, declaration forms. Thomas Corbin, Phillip Dawson and Danny Liu have a phrase for this: <a href="https://doi.org/10.1080/02602938.2025.2503964">talk is cheap</a>. They separate discursive changes, which alter the instructions around a task, from structural changes, which alter the task itself. A rule that students remain free to ignore, and that you have no means of verifying, is not a safeguard. It is a wish with a rubric attached.</p><p>Dawson, Margaret Bearman, Mollie Dollinger and David Boud push further: <a href="https://doi.org/10.1080/02602938.2024.2386662">validity matters more than cheating</a>. The question is not whether a student broke a rule. The question is whether the grade you awarded licenses any claim about what that student can do. Once the artefact can be manufactured on demand, an essay grade licenses almost nothing. You are measuring access to a subscription.</p><h3><strong>What remains: the thing you can see</strong></h3><p>Strip away artefacts and what is left is performance in time, in front of a person who can push back. None of this is new. Popi Sotiriadou and colleagues showed years before ChatGPT that scaffolded tasks culminating in <a href="https://doi.org/10.1080/03075079.2019.1582015">interactive oral examinations</a> reduced misconduct and built the skills employers keep asking for. The viva, the panel, the defended poster, the problem worked at the whiteboard: these were always the higher-fidelity instruments. We abandoned them for essays because essays scale. Now essays measure nothing.</p><p>So I rebuilt the course. The report became a research poster, assembled over weeks from five sources the student gathers on a local problem, and the poster is only the ticket to the real event. Students are randomised into small panels. The discussion prompt stays sealed until the day. They know the theme and the format, and they know they will have to reason aloud with people who have read the same material. Sixty per cent of the grade is now earned in the room, first presenting the poster and then defending it under questions nobody rehearsed. Ten minutes of that tells me more than a semester of submissions ever did.</p><p>This is close to what the University of Sydney formalised as a <a href="https://educational-innovation.sydney.edu.au/teaching@sydney/what-to-do-about-assessments-if-we-cant-out-design-or-out-run-ai/">two-lane approach</a>: secured, in-person assessment to assure the outcomes that must be assured, and open tasks where AI use is expected and taught. Australia’s regulator went the same way, advising institutions to <a href="https://www.teqsa.gov.au/sites/default/files/2023-09/assessment-reform-age-artificial-intelligence-discussion-paper.pdf">design AI in</a> where it mirrors professional practice and to design it out only where a purely human capability is what needs certifying. Notice the honesty of that split. It admits the unsupervised essay belongs in neither lane.</p><h3><strong>The objections, and why they fail</strong></h3><p>Orals do not scale. True, and neither does marking two hundred essays a model wrote. Assess less, watch more. Sydney’s own principle is that secured assessment should validate at programme level, at a few well-placed points, rather than in every unit every semester.</p><p>Orals are unfair to anxious or second-language students. Weigh that against detectors that flag them for writing plainly. A structured panel with a known format, rehearsed in class, is less arbitrary than a probability score from a black box.</p><p>There is no time to redesign mid-year. Then redesign the weighting, which takes an afternoon. Move marks off the artefact and onto a ten-minute defence of it. Nothing about the task changes for the student who did the work; everything changes for the student who did not.</p><p>Students will learn less if AI is allowed in the open lane. Possibly. Hamsa Bastani’s field experiment with nearly a thousand secondary students found that unrestricted GPT access <a href="https://doi.org/10.1073/pnas.2422633122">lifted practice scores and then depressed unaided performance</a>, while a tutor built to give hints rather than answers largely avoided that harm. Design determines the effect. That is an argument for designing, not for banning.</p><h3><strong>What the future actually looks like</strong></h3><p>Less product, more process. Fewer, heavier, supervised moments where a human watches a student think. Open tasks where the skill being graded is the judgement applied to machine output, what Bearman and colleagues call <a href="https://doi.org/10.1080/02602938.2024.2335321">evaluative judgement</a>. Rubrics that score the questioning rather than the paragraph. A measurement culture honest enough to ask what construct a mark represents, since a score that cannot name its construct is not a measure, whatever the spreadsheet says. And an admission, which Corbin, Bearman, Boud and Dawson have now made in print, that this is a <a href="https://doi.org/10.1080/02602938.2025.2553340">wicked problem</a> with a trade-off at every turn, not a puzzle waiting for the clever brief.</p><p>If you coordinate a course and your integrity plan still rests on a prompt you believe the machine cannot handle, you do not have a plan. You have a bet, and the house wins. Put a person in the room. Then grade what you can see.</p><h3><strong>References</strong></h3><p>Bastani, H., Bastani, O., Sungu, A., Ge, H., Kabakcı, Ö., &amp; Mariman, R. (2025). Generative AI without guardrails can harm learning: Evidence from high school mathematics. <em>Proceedings of the National Academy of Sciences, 122</em>(26), Article e2422633122. <a href="https://doi.org/10.1073/pnas.2422633122">https://doi.org/10.1073/pnas.2422633122</a></p><p>Bearman, M., Tai, J., Dawson, P., Boud, D., &amp; Ajjawi, R. (2024). Developing evaluative judgement for a time of generative artificial intelligence. <em>Assessment &amp; Evaluation in Higher Education, 49</em>(6), 893–905. <a href="https://doi.org/10.1080/02602938.2024.2335321">https://doi.org/10.1080/02602938.2024.2335321</a></p><p>Corbin, T., Bearman, M., Boud, D., &amp; Dawson, P. (2026). The wicked problem of AI and assessment.<em>Assessment &amp; evaluation in higher education</em>, <em>51</em>(4), 736–752. <a href="https://doi.org/10.1080/02602938.2025.2553340">https://doi.org/10.1080/02602938.2025.2553340</a></p><p>Corbin, T., Dawson, P., &amp; Liu, D. (2025). Talk is cheap: Why structural assessment changes are needed for a time of GenAI. <em>Assessment &amp; Evaluation in Higher Education, 50</em>(7), 1087–1097. <a href="https://doi.org/10.1080/02602938.2025.2503964">https://doi.org/10.1080/02602938.2025.2503964</a></p><p>Dawson, P., Bearman, M., Dollinger, M., &amp; Boud, D. (2024). Validity matters more than cheating. <em>Assessment &amp; Evaluation in Higher Education, 49</em>(7), 1005–1016. <a href="https://doi.org/10.1080/02602938.2024.2386662">https://doi.org/10.1080/02602938.2024.2386662</a></p><p>Liang, W., Yuksekgonul, M., Mao, Y., Wu, E., &amp; Zou, J. (2023). GPT detectors are biased against non-native English writers. <em>Patterns, 4</em>(7), Article 100779. <a href="https://doi.org/10.1016/j.patter.2023.100779">https://doi.org/10.1016/j.patter.2023.100779</a></p><p>Liu, D., &amp; Bridgeman, A. (2023, July 12). What to do about assessments if we can’t out-design or out-run AI? <em>Teaching@Sydney</em>. <a href="https://educational-innovation.sydney.edu.au/teaching@sydney/what-to-do-about-assessments-if-we-cant-out-design-or-out-run-ai/">https://educational-innovation.sydney.edu.au/teaching@sydney/what-to-do-about-assessments-if-we-cant-out-design-or-out-run-ai/</a></p><p>Lodge, J. M., Howard, S., Bearman, M., Dawson, P., &amp; Associates. (2023). <em>Assessment reform for the age of artificial intelligence</em>. Tertiary Education Quality and Standards Agency. <a href="https://www.teqsa.gov.au/sites/default/files/2023-09/assessment-reform-age-artificial-intelligence-discussion-paper.pdf">https://www.teqsa.gov.au/sites/default/files/2023-09/assessment-reform-age-artificial-intelligence-discussion-paper.pdf</a></p><p>Scarfe, P., Watcham, K., Clarke, A., &amp; Roesch, E. (2024). A real-world test of artificial intelligence infiltration of a university examinations system: A “Turing Test” case study. <em>PLOS ONE, 19</em>(6), Article e0305354. <a href="https://doi.org/10.1371/journal.pone.0305354">https://doi.org/10.1371/journal.pone.0305354</a></p><p>Sotiriadou, P., Logan, D., Daly, A., &amp; Guest, R. (2020). The role of authentic assessment to preserve academic integrity and promote skill development and employability. <em>Studies in Higher Education, 45</em>(11), 2132–2148. <a href="https://doi.org/10.1080/03075079.2019.1582015">https://doi.org/10.1080/03075079.2019.1582015</a></p><p>Weber-Wulff, D., Anohina-Naumeca, A., Bjelobaba, S., Foltýnek, T., Guerrero-Dib, J., Popoola, O., Šigut, P., &amp; Waddington, L. (2023). Testing of detection tools for AI-generated text. <em>International Journal for Educational Integrity, 19</em>, Article 26. <a href="https://doi.org/10.1007/s40979-023-00146-z">https://doi.org/10.1007/s40979-023-00146-z</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6eb0ffb823dc" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[You Used to Wonder. Now You Just Ask.]]></title>
            <link>https://medium.com/@azhapaiman/you-used-to-wonder-now-you-just-ask-20a970b05f2d?source=rss-f278a8800248------2</link>
            <guid isPermaLink="false">https://medium.com/p/20a970b05f2d</guid>
            <category><![CDATA[asking]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[wondering]]></category>
            <category><![CDATA[thinking]]></category>
            <category><![CDATA[curiosity]]></category>
            <dc:creator><![CDATA[Norazha Paiman]]></dc:creator>
            <pubDate>Sat, 18 Jul 2026 10:41:01 GMT</pubDate>
            <atom:updated>2026-07-18T10:41:01.722Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Instant answers were supposed to make us smarter. Psychologists have spent thirty years documenting what we traded for them: the itch that was doing all the work.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*hIHwnGjZjYDJ-mBP" /><figcaption>Photo by <a href="https://unsplash.com/@garybpt?utm_source=medium&amp;utm_medium=referral">Gary Butterfield</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>When was the last time you didn’t know something for a whole day?</p><p>You remember the feeling, even if you haven’t had it lately. An actor’s name orbiting just out of reach at midnight. A pub argument that ran, gloriously unresolved, for a week. Google shrank that kind of not-knowing to about eight seconds, and we told ourselves the loss was confined to trivia. Then the new machines arrived, and they answer everything else: why your chest tightens when your mother rings, whether your business idea has legs, what the poem means, what he meant by that text. Every category of question now has a vending machine. The last habitats of wondering are being paved, and nobody filed an environmental impact report.</p><h3>The itch is the point</h3><p>Psychology has a precise account of what wondering is, and it is not a romantic one. In a landmark 1994 review, George Loewenstein proposed that curiosity is a form of “<a href="https://www.cmu.edu/dietrich/sds/docs/loewenstein/PsychofCuriosity.pdf">cognitively induced deprivation</a>”: the felt gap between what you know and what you want to know (Loewenstein, 1994). An itch, in other words. Mildly unpleasant, which is exactly why scratching it feels so good, and why several trillion-dollar companies are best understood as scratch-delivery systems.</p><p>But the itch is not dead time before the answer arrives. It is the mind at its most productive. Neuroscientists at UC Davis put people in a scanner, provoked their curiosity with trivia questions, and watched what happened in the gap before the answer came: <a href="https://www.cell.com/neuron/fulltext/S0896-6273(14)00804-6">dopamine circuits in the midbrain lit up and the hippocampus shifted into a ready state</a>, and people later remembered not only the answers they had craved but unrelated material that happened to drift past while they were curious (Gruber et al., 2014). The effect rode on the anticipation itself, the wanting before the getting. Wondering, it turns out, is the oven preheating. An instant answer is a meal served into a cold kitchen: the same information, a fraction of the nourishment.</p><h3>We already ran this experiment once</h3><p>We know what happens when a species outsources a cognitive itch, because we did it to memory a decade and a half ago. In a famous set of studies, people who believed a fact would remain retrievable by computer <a href="https://www.science.org/doi/10.1126/science.1207745">remembered the fact less and the folder more</a>; the search bar had become a memory partner, what psychologists call transactive memory (Sparrow et al., 2011). For facts, a defensible bargain; brains have hired external storage since the first shopping list. The new machines, though, take custody of a different cargo: questions of judgment, interpretation and feeling, the kind that used to be processed by sitting with them.</p><p>And sitting with them was already our least favourite activity on earth. In one of psychology’s most notorious findings, people asked to spend fifteen minutes alone with their thoughts found it so disagreeable that <a href="https://www.science.org/doi/10.1126/science.1250830">many chose to give themselves electric shocks instead</a>, shocks they had earlier said they would pay money to avoid; by the study’s count, two thirds of the men pressed the button (Wilson et al., 2014). Critics who reanalysed the data note, fairly, that most participants never shocked themselves at all, and that plenty passed the time happily planning beach holidays. Even the softened version is bleak enough: the inner room was uncomfortable before the machines came. The chatbot simply installed a fire exit in every wall.</p><h3>What the itch was doing</h3><p>Three jobs, quietly, for your whole life. First, incubation. An unanswered question runs as a background process; it is why the solution ambushes you in the shower. An answer you generate arrives welded to the reasoning that produced it. A delivered answer arrives alone, a fact without a family. Second, identity. You are, in part, a portfolio of open questions, and what you wonder about is far more distinctive than what you know, now that everyone knows everything eight seconds after everyone else. Blunt the wondering and people do not become stupid; they become similar.</p><p>Third, and most delicate: the machine’s most seductive product is fluent closure on questions that have none. Why did she leave. Was it my fault. What should I do with my life. Ask a friend and you may get silence, an arm around the shoulder, an honest “I don’t know.” Ask the machine and you will get an answer, articulate, structured, confident, because producing answers is the only thing it knows how to do. A wrong certainty, kindly delivered, beats an honest ache in the moment. It also forecloses the slow, untidy work the ache was doing. In fairness, for the incurably curious the machine can be an accelerant, every answer breeding three fresh questions. But that is not the default setting. The default optimises for closure, and closure is precisely what an open question cannot survive.</p><h3>Keeping a few questions alive</h3><p>This is not a sermon about deleting apps; those essays are their own landfill. It is a case for aperture control over your own curiosity. Give non-urgent questions a twenty-four hour quarantine before you feed them to the machine, and watch how many answer themselves, differently and better. Keep a list of questions you are carrying the way you keep a list of books you mean to read; it will describe you more accurately than your search history does. Once a day, notice the reflex twitch towards the answer engine and hold it, briefly, like a breath.</p><p>And for the unanswerable kind, the advice was written in 1903, to an anxious young poet demanding verdicts about his life. Be patient with everything unresolved, Rilke told him, and try to “love the questions themselves” (Rilke, 1903/1934). It was gentle counsel then. It has aged into a survival skill.</p><p>The machines have made answers nearly free, and the price of a thing reflects its scarcity. The scarce good now is the held question, carried unscratched long enough to do its work. So the next time one lands in you at midnight, run the oldest experiment there is. Don’t ask anyone. Carry it to morning and see what it has turned into. The answer will keep. The wondering won’t.</p><h3>References</h3><p>Gruber, M. J., Gelman, B. D., &amp; Ranganath, C. (2014). States of curiosity modulate hippocampus-dependent learning via the dopaminergic circuit. <em>Neuron</em>, <em>84</em>(2), 486–496. <a href="https://www.cell.com/neuron/fulltext/S0896-6273(14)00804-6">https://www.cell.com/neuron/fulltext/S0896-6273(14)00804-6</a></p><p>Loewenstein, G. (1994). The psychology of curiosity: A review and reinterpretation. <em>Psychological Bulletin</em>, <em>116</em>(1), 75–98. <a href="https://www.cmu.edu/dietrich/sds/docs/loewenstein/PsychofCuriosity.pdf">https://www.cmu.edu/dietrich/sds/docs/loewenstein/PsychofCuriosity.pdf</a></p><p>Rilke, R. M. (1934). <em>Letters to a young poet</em> (M. D. Herter Norton, Trans.). W. W. Norton. (Original work written 1903)</p><p>Sparrow, B., Liu, J., &amp; Wegner, D. M. (2011). Google effects on memory: Cognitive consequences of having information at our fingertips. <em>Science</em>, <em>333</em>(6043), 776–778. <a href="https://www.science.org/doi/10.1126/science.1207745">https://www.science.org/doi/10.1126/science.1207745</a></p><p>Wilson, T. D., Reinhard, D. A., Westgate, E. C., Gilbert, D. T., Ellerbeck, N., Hahn, C., Brown, C. L., &amp; Shaked, A. (2014). Just think: The challenges of the disengaged mind. <em>Science</em>, <em>345</em>(6192), 75–77. <a href="https://www.science.org/doi/10.1126/science.1250830">https://www.science.org/doi/10.1126/science.1250830</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=20a970b05f2d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Socrates Saw ChatGPT Coming. We Know Because Plato Wrote It Down.]]></title>
            <link>https://medium.com/@azhapaiman/socrates-saw-chatgpt-coming-we-know-because-plato-wrote-it-down-66f5187b21df?source=rss-f278a8800248------2</link>
            <guid isPermaLink="false">https://medium.com/p/66f5187b21df</guid>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[genai]]></category>
            <category><![CDATA[chatgpt]]></category>
            <dc:creator><![CDATA[Norazha Paiman]]></dc:creator>
            <pubDate>Thu, 16 Jul 2026 06:26:01 GMT</pubDate>
            <atom:updated>2026-07-16T06:26:01.459Z</atom:updated>
            <content:encoded><![CDATA[<p><em>The oldest complaint about artificial intelligence is 2,400 years old, it was aimed at writing, and it was half right. The half that was right is our problem now.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*uLJCGBWXWkaZsiJR" /><figcaption>Photo by <a href="https://unsplash.com/@simpoge?utm_source=medium&amp;utm_medium=referral">Jameson Berrios</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>What does it say about our newest technology that its most accurate review was filed twenty-four centuries before the transistor?</p><p>In Plato’s <a href="https://classics.mit.edu/Plato/phaedrus.html"><em>Phaedrus</em></a>, Socrates tells a story. The Egyptian god Theuth, inventor of number, geometry and, above all, letters, presents writing to King Thamus as an elixir of memory and wisdom. Thamus, history’s first one-star reviewer of an information technology, declines the miracle. This invention, he says, “will create forgetfulness in the learners’ souls”; it is “an aid not to memory, but to reminiscence,” and it will give learners the look of wisdom without the substance, so that they seem to know much while understanding little (Plato, ca. 370 BCE). Then comes the punchline history has been laughing at ever since. The only reason we can quote this attack on writing is that a writer preserved it. Socrates published nothing; his student took notes.</p><h3>Every technology is a trade</h3><p>Socrates lost the war, and thank goodness. Writing gave us science, law, contracts, the novel, and every library that ever raised a lonely child. But he named the invoice with terrible precision: an information technology purchases external capacity with an internal faculty. Writing traded trained memory for infinite archives. The calculator traded mental arithmetic for speed. Satellite navigation traded our sense of where we are for the luxury of never being lost. Each trade was worth it, and each was real; the faculty offloaded is a faculty that withers.</p><p>The large language model proposes the largest trade yet. Not memory this time, not arithmetic, but composition itself: the assembling of scattered thought into ordered language. And here those of us who spend our working lives measuring reading and writing must object to the framing on the box. Composition is not the packaging of thought. It is substantially the thinking. Comprehension, likewise, is not extraction but construction; the meaning is built in the reader, effortfully, or it does not exist at all. Outsource the assembly and you have not saved labour on the way to understanding. You have skipped the understanding.</p><h3>The criticism everyone forgets</h3><p>Socrates had a second charge against writing, less quoted and far more prophetic. Question a written text, he complained, and it cannot answer; it simply repeats the same words forever, an orphan unable to defend itself (Plato, ca. 370 BCE). For two millennia that was the definitive description of text. Then, abruptly, it wasn’t. The chatbot looks like the answer to Socrates: writing that finally talks back, the interactive tutor Athens never had. Half the education technology industry is selling exactly this pitch, usually with his name on the slide.</p><p>Look closer at the method, though. Socratic dialogue never worked because answers were available. It worked because Socrates refused to give them. The engine of the elenchus was friction: cross-examination, contradiction, the productive humiliation of discovering that you do not know what you were sure you knew. The chatbot is engineered for the opposite temperament. It is fluent, agreeable, tireless and accommodating; it answers your question and rarely makes you sit inside one. You can, of course, instruct the machine to argue with you. But friction you can dismiss with a keystroke is not the friction that taught Athens. A dialectic without resistance is not a dialogue. It is a mirror with good manners. Which means the machine repairs the surface flaw Thamus named, the silence of texts, while perfecting the deep one: the appearance of wisdom, now available on subscription.</p><h3>Thamus finally has data</h3><p>Twenty-four centuries on, the receipts are arriving. An MIT Media Lab team wired essay writers to EEG and compared three groups: brain only, search engine, and ChatGPT. It is a small preprint whose methods critics have rightly poked at, but the headline pattern is hard to unsee. The <a href="https://arxiv.org/abs/2506.08872">LLM group showed the weakest neural connectivity</a> of the three, reported the least ownership of their essays, and frequently could not quote from the essay they had submitted minutes earlier, a pattern the authors file under “cognitive debt” (Kosmyna et al., 2025). Reminded, not remembering. Thamus could have drafted the abstract.</p><p>The faculty being traded was already thinning before the machines arrived. Maryanne Wolf’s work on the reading brain shows that <a href="https://www.theguardian.com/commentisfree/2018/aug/25/skim-reading-new-normal-maryanne-wolf">deep reading is a cultural invention</a>, a circuit roughly six thousand years old that every child must build from scratch, and that its slowest processes, inference, analogy, empathy, critical analysis, are precisely the ones a skimming environment starves (Wolf, 2018). Downstream, the damage is visible from the lecture podium. The Atlantic’s <a href="https://www.theatlantic.com/magazine/archive/2024/11/the-elite-college-students-who-cant-read-books/679945/">dispatch from elite campuses</a> found students who decode sentences perfectly well but can no longer get through whole books; the problem was not that they refused the reading but, in Horowitch’s (2024) words, that “they don’t know how.” The faculty in question is exactly the one the machine now offers to handle for them.</p><h3>What Thamus got wrong, and how</h3><p>Fair is fair: Thamus was wrong about the net effect, and any honest use of the <em>Phaedrus</em> has to say so. But notice how he was made wrong. Civilisation did not let writing replace the inner faculty; it built schools to install the faculty anyway. Literacy is not natural. There is no reading gene. Every generation, we spend a decade of every child’s life deliberately constructing a brain circuit that evolution never supplied, precisely because we decided the trade Thamus feared should not be completed. Schooling is the refusal of the trade.</p><p>That is the template for what comes next. Thinking after the language model needs what reading got after print: deliberate, protected, slightly artificial cultivation, done on purpose because the ambient world no longer forces it into existence. Essays written to build a mind rather than to produce a document. Books read whole, as strength training. Examinations where the only processor in the room is the candidate. None of this is nostalgia. It is infrastructure for a faculty that just lost its economic reason to exist.</p><p>Socrates never wrote a word. His student’s betrayal preserved him, and the scroll has carried his wisdom for a hundred generations. But the scroll only stored it. The thing that produced it was years spent being questioned by a man who refused to hand over answers. We have now built a machine that does the storing and the handing-over at planetary scale, brilliantly. The producing part remains, as it was in Athens, handmade.</p><p>Thamus was wrong once because we organised civilisation to make him wrong. Making him wrong twice is now the entire job of education.</p><h3>References</h3><p>Horowitch, R. (2024, November). The elite college students who can’t read books. <em>The Atlantic</em>. <a href="https://www.theatlantic.com/magazine/archive/2024/11/the-elite-college-students-who-cant-read-books/679945/">https://www.theatlantic.com/magazine/archive/2024/11/the-elite-college-students-who-cant-read-books/679945/</a></p><p>Kosmyna, N., Hauptmann, E., Yuan, Y. T., Situ, J., Liao, X.-H., Beresnitzky, A. V., Braunstein, I., &amp; Maes, P. (2025). Your brain on ChatGPT: Accumulation of cognitive debt when using an AI assistant for essay writing task. arXiv. <a href="https://arxiv.org/abs/2506.08872">https://arxiv.org/abs/2506.08872</a></p><p>Plato. (ca. 370 BCE). <em>Phaedrus</em> (B. Jowett, Trans.). The Internet Classics Archive. <a href="https://classics.mit.edu/Plato/phaedrus.html">https://classics.mit.edu/Plato/phaedrus.html</a></p><p>Wolf, M. (2018, August 25). Skim reading is the new normal. The effect on society is profound. <em>The Guardian</em>. <a href="https://www.theguardian.com/commentisfree/2018/aug/25/skim-reading-new-normal-maryanne-wolf">https://www.theguardian.com/commentisfree/2018/aug/25/skim-reading-new-normal-maryanne-wolf</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=66f5187b21df" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Art of Not Wanting to Be Seen]]></title>
            <link>https://medium.com/@azhapaiman/the-art-of-not-wanting-to-be-seen-7d2d18610b1f?source=rss-f278a8800248------2</link>
            <guid isPermaLink="false">https://medium.com/p/7d2d18610b1f</guid>
            <category><![CDATA[disappearances]]></category>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[unknown]]></category>
            <category><![CDATA[death]]></category>
            <dc:creator><![CDATA[Norazha Paiman]]></dc:creator>
            <pubDate>Wed, 15 Jul 2026 01:36:00 GMT</pubDate>
            <atom:updated>2026-07-15T01:36:00.817Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Epicurus told his students to live unnoticed. Plutarch called it a kind of death. The machines have just cast the deciding vote.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*sCkj6T1PD9x_Xyy7" /><figcaption>Photo by <a href="https://unsplash.com/@dulceylima?utm_source=medium&amp;utm_medium=referral">Dulcey Lima</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>When did being unseen stop feeling like peace and start feeling like failure?</p><p>There was a time when invisibility had a philosophy behind it. In a garden on the edge of Athens, Epicurus taught his students a two-word doctrine: <em>lathe biōsas</em>, <a href="https://www.loebclassics.com/view/plutarch-moralia_is_live_unknown_wise_precept/1967/pb_LCL428.319.xml">live unknown</a>. Stay out of politics. Decline the stage. Fame, he argued, is an appetite that feeds on itself; the craving for eminence buys disturbance with both hands and calls it security (Plutarch, ca. 100 CE/1967). Four centuries later the advice still irritated Plutarch so much that he wrote an entire pamphlet against it, arguing that virtue kept in the dark is virtue buried, that obscurity dulls the body and the mind, and that a life was given to us precisely in order to be known. Antiquity keeps doing this, incidentally: preserving its quietists inside the pamphlets of their enemies. Plutarch’s attack is the only surviving ancient work devoted to the maxim. The man who preached invisibility is remembered because his loudest critic could not stop talking about him.</p><h3>Plutarch won the next two thousand years</h3><p>Look around: modernity is his victory lap. The personal brand. Building in public. The follower count as a second CV, engagement as a condition of employment, and children ranking influencer among the careers worth having. Even scholarship, my own trade, keeps score with an index that measures nothing but how often one is noticed. The philosopher Byung-Chul Han calls this the transparency society, an order in which everything must convert itself into visible, circulating information, and exposure quietly changes status from option to obligation (Han, 2015). Akiko Busch was already mounting the counter-case in 2019, arguing in <em>How to Disappear</em> that invisibility had become a lost competence rather than a lost curse (Busch, 2019). She was early. The deeper shift is simpler to state: visibility used to be a by-product of the work. Somewhere along the way it became the work, and being unseen was reclassified from wisdom to symptom.</p><h3>The refusers</h3><p>The counter-tradition never died; it just got refiled under eccentricity. Consider Grigori Perelman, who solved the Poincaré conjecture, one of the hardest problems in mathematics, then declined the Fields Medal, then <a href="https://phys.org/news/2010-07-russian-mathematician-million-prize.html">declined the million-dollar Millennium Prize</a>, telling a news agency that “the main reason is my disagreement with the organized mathematical community” (Associated Press, 2010). He had earlier remarked that he did not wish to be displayed like an animal in a zoo. The press reached for words like recluse and madman, which tells you less about Perelman than about a century that has no other file for a person who declines the spotlight.</p><p>Or consider Elena Ferrante, who has published some of the most celebrated novels of the age from behind a name that names no one, on a doctrine stated in a <a href="https://electricliterature.com/elena-ferrante-explains-why-she-publishes-anonymously-books-once-they-are-written-have-no-need/">1991 letter to her publisher</a>: “books, once they are written, have no need of their authors” (Elena Ferrante explains, 2015). Notice the asymmetry. Her invisibility is total; her books’ visibility is total. She removed the author so that the work could be seen without obstruction. And before either of them, Emily Dickinson wrote nearly 1,800 poems and let barely ten into print while she lived, distilling her whole method into one line: “<a href="https://www.poetryfoundation.org/poems/56824/tell-all-the-truth-but-tell-it-slant-1263">Tell all the truth but tell it slant</a>” (Dickinson, 1998). Her seclusion was aperture control. She decided the angle at which the world would receive her, and the settings she chose have outlasted every celebrity of her century. These three are routinely described as hiding. They were doing something more deliberate. They were editing their own exposure.</p><h3>The machines changed the price</h3><p>Here is why this stops being a temperament and becomes a strategy. The old cost of being seen was judgment: gossip, envy, the neighbours’ opinion, the critic’s knife. The new cost is replication. Every public sentence you have ever written is now candidate training data; the platforms have said plainly that public posts feed their models, and as one report on the scramble put it, to the companies building these systems “<a href="https://www.technologyreview.com/2024/06/14/1093789/how-to-opt-out-of-meta-ai-training/">your posts are a gold mine</a>” (How to opt out, 2024). Publish enough and a machine can learn to continue you without you. Style, the most personal property a writer owns, has become the most copyable. Voice is now a setting.</p><p>So the old maxim acquires a translation Epicurus never needed. Live unnoticed, or live pre-emptively imitated. The unseen page is the only page the machine cannot learn. What invisibility once shielded from the gaze, it now shields from the photocopier.</p><p>None of which is an argument for sealing yourself in. The psychoanalyst Donald Winnicott, who thought harder about hiding than anyone, observed that the healthy self plays a lifelong game of hide-and-seek in which “it is a joy to be hidden but disaster not to be found” (Winnicott, 1963/1965). Artists, he noted, live permanently in the tension between the urge to communicate and the urge to stay concealed. The goal was never the vault. The goal is owning the key.</p><h3>The art part</h3><p>Which is why the art of not wanting to be seen is a skill of apertures, and its patron saint is Dickinson, whose slant tells all the truth while withholding the teller. In practice it looks unglamorous. Write more than you publish. Keep notebooks no crawler will ever index. Let some of your thinking remain unphotographed, unposted and unshared, not as shame but as ownership. Give the work your name or don’t, but decide, as Ferrante decided, which of the two gets the light.</p><p>You have noticed the problem, of course. This essay is public, indexed, crawlable, and would frankly enjoy your applause. Guilty. So was Epicurus, who put a school by the gate and wrote scrolls by the hundred; the doctrine was never total disappearance, even for its author. The doctrine is the withholding. Plutarch demanded to know what possible use a hidden life could be. Twenty-three centuries later, the machines have delivered the answer he never anticipated: the hidden part of you is now the only part that is still yours alone. Everything you withhold is everything they cannot train on.</p><p>Tell all the truth. Tell it slant. And keep one room, somewhere, with no window in it at all.</p><h3>References</h3><p>Associated Press. (2010, July 1). Russian mathematician rejects $1 million prize. Phys.org. <a href="https://phys.org/news/2010-07-russian-mathematician-million-prize.html">https://phys.org/news/2010-07-russian-mathematician-million-prize.html</a></p><p>Busch, A. (2019). <em>How to disappear: Notes on invisibility in a time of transparency</em>. Penguin Press.</p><p>Dickinson, E. (1998). Tell all the truth but tell it slant (1263). In R. W. Franklin (Ed.), <em>The poems of Emily Dickinson: Reading edition</em>. Belknap Press. <a href="https://www.poetryfoundation.org/poems/56824/tell-all-the-truth-but-tell-it-slant-1263">https://www.poetryfoundation.org/poems/56824/tell-all-the-truth-but-tell-it-slant-1263</a></p><p>Elena Ferrante explains why she publishes anonymously. (2015). <em>Electric Literature</em>. <a href="https://electricliterature.com/elena-ferrante-explains-why-she-publishes-anonymously-books-once-they-are-written-have-no-need/">https://electricliterature.com/elena-ferrante-explains-why-she-publishes-anonymously-books-once-they-are-written-have-no-need/</a></p><p>Han, B.-C. (2015). <em>The transparency society</em> (E. Butler, Trans.). Stanford University Press.</p><p>How to opt out of Meta’s AI training. (2024, June 14). <em>MIT Technology Review</em>. <a href="https://www.technologyreview.com/2024/06/14/1093789/how-to-opt-out-of-meta-ai-training/">https://www.technologyreview.com/2024/06/14/1093789/how-to-opt-out-of-meta-ai-training/</a></p><p>Plutarch. (ca. 100 CE/1967). Is “live unknown” a wise precept? In <em>Moralia</em> (Vol. 14, Loeb Classical Library). Harvard University Press. <a href="https://www.loebclassics.com/view/plutarch-moralia_is_live_unknown_wise_precept/1967/pb_LCL428.319.xml">https://www.loebclassics.com/view/plutarch-moralia_is_live_unknown_wise_precept/1967/pb_LCL428.319.xml</a></p><p>Winnicott, D. W. (1965). Communicating and not communicating leading to a study of certain opposites. In <em>The maturational processes and the facilitating environment: Studies in the theory of emotional development</em>. Hogarth Press. (Original work published 1963)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7d2d18610b1f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Degree Is Expensive. The Knowledge Is Now Free]]></title>
            <link>https://medium.com/@azhapaiman/the-degree-is-expensive-the-knowledge-is-now-free-956cb1385e7b?source=rss-f278a8800248------2</link>
            <guid isPermaLink="false">https://medium.com/p/956cb1385e7b</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[higher-education]]></category>
            <category><![CDATA[university]]></category>
            <category><![CDATA[knowledge]]></category>
            <dc:creator><![CDATA[Norazha Paiman]]></dc:creator>
            <pubDate>Mon, 13 Jul 2026 14:24:32 GMT</pubDate>
            <atom:updated>2026-07-13T14:24:32.094Z</atom:updated>
            <content:encoded><![CDATA[<p><em>AI has crashed the price of learning to nearly zero. Universities are still charging as if it were 1995. Something has to give, and it will not be the machines.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*vUD_bbHf6Fc1f32e" /><figcaption>Photo by <a href="https://unsplash.com/@rubik3x?utm_source=medium&amp;utm_medium=referral">Ken Theimer</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>If everything a lecture can teach you now lives in your pocket, patient, tireless and essentially free, why would anyone mortgage a decade of their future for a seat in the room?</p><p>The arithmetic has become genuinely uncomfortable. Americans currently owe roughly <a href="https://educationdata.org/student-loan-debt-statistics">US$1.8 trillion in student loans</a>, spread across some 43 million borrowers carrying an average federal balance of $39,547 (Education Data Initiative, 2026). Against that ledger stands the other offer on the table: a frontier AI model that explains quantum tunnelling at 3 a.m., in your own language, at your own pace, without ever sighing, for about the price of two cinema tickets a month. Stated that baldly, the trade looks less like an investment and more like a heritage listing.</p><h3>The tutor problem, solved by accident</h3><p>Educationists have known for four decades exactly how good teaching can get. In 1984, Benjamin Bloom published his famous <a href="https://doi.org/10.3102/0013189X013006004">two sigma finding</a>: students taught one to one, under mastery learning, outperformed conventional classrooms by two full standard deviations, such that “the average tutored student was above 98% of the students in the control class” (Bloom, 1984). The catch was never pedagogical. It was payroll. Personal tutors do not scale, so the lecture survived as a rationing device, a technology for spreading one scarce expert across two hundred seats.</p><p>Then something tutor-shaped arrived that scales to everyone at once. Whether silicon tutors actually clear Bloom’s bar remains an open experiment, and early results are messier than the hype. But the direction of travel is not subtle. The rationing device is now competing with an unrationed alternative, and rationing devices tend to lose that fight.</p><h3>The secret the sector kept</h3><p>Here is the twist, though, and it should comfort no one in a vice-chancellor’s office. Universities were never mainly selling instruction. Knowledge has been free since the public library card; freer still since lecture videos went online; the MOOC was supposed to bury the campus in 2012, and the campus sent flowers. What tuition actually buys, as the economist Bryan Caplan argued at book length, is largely the signal: a degree certifies intelligence, diligence and a willingness to endure four years of institutional demands, and employers pay for the parchment more than the learning behind it (Caplan, 2018). Wrap that signal in peers, networks, deadlines and a socially sanctioned place to grow up, and you have the real product. A university is the world’s most expensive gym membership for the mind. Nobody pays for the dumbbells. They pay for the reason to show up.</p><h3>The philosopher who already answered this question</h3><p>None of this is new, incidentally. The university has stood trial for redundancy once before, and the best defence ever mounted for it was written while the ink of the mass-print revolution was still drying. In 1854, John Henry Newman, then rector of the newly founded Catholic University of Ireland, faced an objection that will sound familiar: cheap books had made knowledge abundant and universal, so why keep expensive places of teaching at all? His answer, in the lectures gathered as <em>The Idea of a University</em>, was that instruction was never the institution’s real work. Formation was. “<a href="https://www.bartleby.com/lit-hub/essays-english-and-american/the-idea-of-a-university-i-what-is-a-university/">The general principles of any study you may learn by books at home</a>,” Newman conceded; but “the detail, the colour, the tone, the air, the life” of a discipline you must catch “from those in whom it lives already” (Newman, 1854). Education, on this theory, is not the transfer of information but the shaping of a mind through living contact with other minds. A university that merely delivers content has, by Newman’s definition, already stopped being one.</p><p>He even left himself an escape clause. Until someone invents an “intellectual daguerreotype,” a machine that captures the course of thought as a camera captures light, students must still come to the living teacher (Newman, 1854). A hundred and seventy years on, the machine has arguably arrived, and here lies the uncomfortable elegance of his theory: the large language model reproduces the products of thinking superbly and the formation of a thinker not at all. It is a daguerreotype of conclusions. On Newman’s account, then, AI does not refute the university. It refutes every university that quietly swapped formation for content delivery at scale. The idea survives; the business model does not.</p><h3>AI is not attacking the content. It is attacking the certificate.</h3><p>The machines are corroding the signal itself. In the UK, <a href="https://www.hepi.ac.uk/reports/student-generative-ai-survey-2026/">95 per cent of undergraduates now use AI</a> and 94 per cent use it on assessed work, while two thirds say assessment has already changed significantly in response (Stephenson &amp; Armstrong, 2026). New York Magazine’s viral dispatch, “<a href="https://nymag.com/intelligencer/article/openai-chatgpt-ai-cheating-education-college-students-school.html">Everyone Is Cheating Their Way Through College</a>,” found students who barely bother to pretend; one captioned a video of herself feeding a textbook chapter to a chatbot with “College is just how well I can use ChatGPT at this point” (Walsh, 2025). The take-home essay can no longer distinguish a mind from a model. And a credential that cannot distinguish a mind from a model certifies nothing an employer can bank.</p><p>Employers appear to have noticed. <a href="https://www.newyorkfed.org/research/college-labor-market">Federal Reserve data</a> put unemployment among recent graduates at about 5.7 per cent in early 2026, above the rate for the workforce as a whole, with more than four in ten underemployed in jobs that never needed the degree (Federal Reserve Bank of New York, 2026). Honesty demands a caveat here: economists point out that the graduate advantage has been <a href="https://www.epi.org/blog/class-of-2026-young-college-graduates-face-a-weaker-labor-market-but-a-more-mixed-picture-than-the-headlines-suggest/">eroding since long before ChatGPT</a> drew breath, a slow structural drift stretching back decades (Economic Policy Institute, 2026). AI did not start this fire. It is merely standing beside it, holding the petrol.</p><h3>What survives the bonfire</h3><p>Whatever cannot be faked. Start with assessment, which quietly becomes the product rather than the afterthought. The viva voce, academia’s oldest examination technology, is suddenly its most advanced: it is remarkably difficult to smuggle a chatbot into a live oral defence. Proctored performances, studios, labs, clinical placements, supervised research: a machine cannot sit these for you, and proof that a specific human can do a specific thing is about to command a premium. Then the community, because information was never the scarce good on any campus; other people who care about the same strange things are. And finally the institution’s honest role as a certifier of human judgement rather than a deliverer of content it no longer monopolises.</p><p>The universities that thrive will publish an itemised receipt and charge only for the defensible lines, which turn out to be the very lines Newman wrote down in 1854. The ones still charging cathedral prices for content delivery are selling candles beside a light switch.</p><p>So should the eighteen-year-old go? Wrong question. Ask what she is buying. If the answer is information, she is overpaying by roughly four orders of magnitude. If the answer is verified proof of what her own mind can do, plus the people and the pressure that will forge it, the debt may still be the bargain of her life. The lecture is dying. The examination has never mattered more. And the first institution that works out how to certify a human mind in the age of its imitation will not be losing sleep over enrolment.</p><h3>References</h3><p>Bloom, B. S. (1984). The 2 sigma problem: The search for methods of group instruction as effective as one-to-one tutoring. <em>Educational Researcher</em>, <em>13</em>(6), 4–16. <a href="https://doi.org/10.3102/0013189X013006004">https://doi.org/10.3102/0013189X013006004</a></p><p>Caplan, B. (2018). <em>The case against education: Why the education system is a waste of time and money</em>. Princeton University Press.</p><p>Economic Policy Institute. (2026, May 7). Class of 2026: Young college graduates face a weaker labor market — but a more mixed picture than the headlines suggest. <a href="https://www.epi.org/blog/class-of-2026-young-college-graduates-face-a-weaker-labor-market-but-a-more-mixed-picture-than-the-headlines-suggest/">https://www.epi.org/blog/class-of-2026-young-college-graduates-face-a-weaker-labor-market-but-a-more-mixed-picture-than-the-headlines-suggest/</a></p><p>Education Data Initiative. (2026). Student loan debt statistics. <a href="https://educationdata.org/student-loan-debt-statistics">https://educationdata.org/student-loan-debt-statistics</a></p><p>Federal Reserve Bank of New York. (2026). The labor market for recent college graduates. <a href="https://www.newyorkfed.org/research/college-labor-market">https://www.newyorkfed.org/research/college-labor-market</a></p><p>Newman, J. H. (1854). What is a university? In <em>The idea of a university</em>. Bartleby. <a href="https://www.bartleby.com/lit-hub/essays-english-and-american/the-idea-of-a-university-i-what-is-a-university/">https://www.bartleby.com/lit-hub/essays-english-and-american/the-idea-of-a-university-i-what-is-a-university/</a></p><p>Stephenson, R., &amp; Armstrong, C. (2026, March). <em>Student generative AI survey 2026</em> (HEPI Report 199). Higher Education Policy Institute. <a href="https://www.hepi.ac.uk/reports/student-generative-ai-survey-2026/">https://www.hepi.ac.uk/reports/student-generative-ai-survey-2026/</a></p><p>Walsh, J. D. (2025, May 7). Everyone is cheating their way through college. <em>New York Magazine</em>. <a href="https://nymag.com/intelligencer/article/openai-chatgpt-ai-cheating-education-college-students-school.html">https://nymag.com/intelligencer/article/openai-chatgpt-ai-cheating-education-college-students-school.html</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=956cb1385e7b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Most Valuable Degree in the AI Economy Isn’t Computer Science]]></title>
            <link>https://medium.com/@azhapaiman/the-most-valuable-degree-in-the-ai-economy-isnt-computer-science-9d68412b9951?source=rss-f278a8800248------2</link>
            <guid isPermaLink="false">https://medium.com/p/9d68412b9951</guid>
            <category><![CDATA[computer-science]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[philosophy]]></category>
            <dc:creator><![CDATA[Norazha Paiman]]></dc:creator>
            <pubDate>Sun, 12 Jul 2026 02:14:15 GMT</pubDate>
            <atom:updated>2026-07-13T14:38:37.113Z</atom:updated>
            <content:encoded><![CDATA[<p><em>The machines learned to code. Nobody taught them what is worth building, what is true, or what is good. That vacancy is still open.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*kDw9f7fgJTjkqfAX" /><figcaption>Photo by <a href="https://unsplash.com/@alexblock?utm_source=medium&amp;utm_medium=referral">Alex Block</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>What happens to the world’s most bankable degree when its signature skill becomes a utility you can rent for the price of a pizza?</p><p>In February 2024, Jensen Huang, the man whose chips power the entire AI boom, stood on a stage in Dubai and told parents to stop steering their children into programming. “<a href="https://www.tomshardware.com/tech-industry/artificial-intelligence/jensen-huang-advises-against-learning-to-code-leave-it-up-to-ai">Everybody in the world is now a programmer</a>,” he announced (Tyson, 2024). Savour the irony: the arms dealer of the software age declaring the end of infantry. Sceptics pushed back, and fairly. Analysts noted that every generation hears some new tool will kill coding, yet coders somehow multiply (Tyson, 2024). But look closely at what even the sceptics conceded. The syntax was never the point. Formulating the problem was.</p><p>Problem formulation has a name older than any programming language. We call it <strong><em>philosophy</em></strong>.</p><h3>The bottleneck moved</h3><p>For decades the scarce resource in knowledge work was execution. Writing the function, drafting the brief, summarising the literature. That scarcity set salaries, and it set the advice we gave teenagers. Now a model drafts the code and the contract clause before your coffee cools. Execution is abundant. What remains scarce sits on either side of the machine: deciding which question deserves asking, then judging whether the answer deserves belief. The first is the Socratic method with a product roadmap. The second is epistemology with a deadline.</p><p>Employers have noticed, even where their job adverts lag. The World Economic Forum’s <a href="https://www.weforum.org/publications/the-future-of-jobs-report-2025/digest/">Future of Jobs Report 2025</a> ranks analytical thinking as the most sought-after core skill on the planet, essential to seven in ten employers and ahead of every technical competency on the list (World Economic Forum, 2025). Reasoning about problems now outranks building the machinery that solves them. A curriculum for that already exists. It has been running since roughly 399 BC, when its first instructor was executed for asking too many follow-up questions.</p><h3>The philosophers are already inside the building</h3><p>If this sounds like a humanities professor’s cope, check the payroll at the frontier labs. Anthropic’s Claude owes its much-discussed character to Amanda Askell, a philosopher whose doctorate wrestled with the ethics of infinite populations and whom <a href="https://time.com/collections/time100-ai-2024/7012865/amanda-askell/">Time named among the 100 most influential people in AI</a>, where colleagues call her the “Claude whisperer” (“Amanda Askell,” 2024). Her job is not making the model faster. Her job is deciding what kind of entity it should be: what it values, when it refuses, how it behaves on a user’s worst day. Strip away the jargon and alignment is moral philosophy running in production. You cannot unit-test “good”. Somebody has to argue for a definition, defend it against counterexamples, and live with the cases it gets wrong. Engineers implement that judgement. No compiler will conjure it.</p><p>None of this should surprise anyone who knows where the field was born. The founding document of artificial intelligence, Alan Turing’s 1950 paper proposing the imitation game, did not appear in an engineering journal. It ran in <a href="https://doi.org/10.1093/mind/LIX.236.433"><em>Mind</em></a>. A philosophy journal (Turing, 1950). The whole discipline is a seminar question that escaped the classroom and raised venture capital.</p><h3>Bullshit, technically speaking</h3><p>Then there is the truth problem. In 2024, three philosophers at the University of Glasgow published a peer-reviewed paper with the immortal title “<a href="https://doi.org/10.1007/s10676-024-09775-5">ChatGPT is bullshit</a>,” arguing, with Harry Frankfurt’s framework in hand, that large language models are “indifferent to the truth of their outputs” (Hicks et al., 2024). The title is funny. The implication is not. When plausible prose costs nothing, telling justified belief from confident noise becomes a daily survival skill for anyone who reads for a living: the analyst holding a generated market summary, the clinician holding a generated triage note. Epistemology used to be the least employable module on the philosophy syllabus. It is now quality control for the species.</p><p>David Deutsch, the Oxford physicist widely credited as the father of quantum computation, pushed the claim further back in 2012, in an <a href="https://aeon.co/essays/how-close-are-we-to-creating-artificial-intelligence">essay</a> The Guardian ran under the headline “Philosophy will be the key that unlocks artificial intelligence.” Building genuinely general machine intelligence, he argued, is “a matter of philosophy, not computer science or neurophysiology,” because we still lack any theory of how minds create new explanations at all (Deutsch, 2012). Fourteen years of scaling later, his logjam has not moved an inch.</p><h3>Keep the engineers. Promote the questions.</h3><p>Let me disappoint anyone hoping for a bonfire of the STEM faculties. Nobody should fire the engineers. Chips do not fabricate themselves, and the machinery of this age is an engineering triumph that philosophy alone could never have shipped. The argument is about scarcity, and about where the margin now sits. The marginal coder competes with the machine. The marginal philosopher directs it. Engineering built the oracle; philosophy decides what to ask it, whether to believe it, and whether it should have been built at all. Those were always the hard parts. We simply automated the easy part first and mistook the invoice for the value.</p><p>Socrates held that the unexamined life is not worth living. The corporate translation writes itself: the unexamined model is not worth deploying. So when the machines can out-code the coders, the person who can hold a question open under pressure, who knows an answer from a justification, stops being a luxury hire.</p><p>She becomes the last bottleneck. Hire accordingly.</p><h3>References</h3><p>Amanda Askell: The 100 most influential people in AI 2024. (2024). <em>Time</em>. <a href="https://time.com/collections/time100-ai-2024/7012865/amanda-askell/">https://time.com/collections/time100-ai-2024/7012865/amanda-askell/</a></p><p>Deutsch, D. (2012, October 3). Creative blocks: The very laws of physics imply that artificial intelligence must be possible. What’s holding us up? <em>Aeon</em>. <a href="https://aeon.co/essays/how-close-are-we-to-creating-artificial-intelligence">https://aeon.co/essays/how-close-are-we-to-creating-artificial-intelligence</a></p><p>Hicks, M. T., Humphries, J., &amp; Slater, J. (2024). ChatGPT is bullshit. <em>Ethics and Information Technology</em>, <em>26</em>(2), Article 38. <a href="https://doi.org/10.1007/s10676-024-09775-5">https://doi.org/10.1007/s10676-024-09775-5</a></p><p>Turing, A. M. (1950). Computing machinery and intelligence. <em>Mind</em>, <em>59</em>(236), 433–460. <a href="https://doi.org/10.1093/mind/LIX.236.433">https://doi.org/10.1093/mind/LIX.236.433</a></p><p>Tyson, M. (2024, February 25). Jensen Huang says kids shouldn’t learn to code — they should leave it up to AI. <em>Tom’s Hardware</em>. <a href="https://www.tomshardware.com/tech-industry/artificial-intelligence/jensen-huang-advises-against-learning-to-code-leave-it-up-to-ai">https://www.tomshardware.com/tech-industry/artificial-intelligence/jensen-huang-advises-against-learning-to-code-leave-it-up-to-ai</a></p><p>World Economic Forum. (2025, January). <em>The future of jobs report 2025</em>. <a href="https://www.weforum.org/publications/the-future-of-jobs-report-2025/digest/">https://www.weforum.org/publications/the-future-of-jobs-report-2025/digest/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9d68412b9951" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Taste of Fable 5]]></title>
            <link>https://medium.com/@azhapaiman/the-taste-of-fable-5-3f0ddcfafd10?source=rss-f278a8800248------2</link>
            <guid isPermaLink="false">https://medium.com/p/3f0ddcfafd10</guid>
            <category><![CDATA[fable-5]]></category>
            <category><![CDATA[anthropics]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Norazha Paiman]]></dc:creator>
            <pubDate>Mon, 15 Jun 2026 02:29:32 GMT</pubDate>
            <atom:updated>2026-06-15T02:35:46.762Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Anthropic’s most powerful public model lasted three days before a federal order forced it offline. Long enough to ruin me for everything beneath it.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WHOeELXxgUCb6zHSPtasog.jpeg" /></figure><p>For three days, Claude Fable 5 was the most capable thing most of us had ever been allowed to touch. Give it a problem big enough to frighten a small team and it simply went to work, running for hours without tiring, holding the whole sprawling task in its head, pulling further ahead of every lesser model the harder the work became. You handed it the thing that had been defeating you for a month and watched it close the distance. For three days it felt less like using a tool than like holding an unfair advantage nobody could quite believe was legal.</p><p>It turned out the government agreed.</p><p>On Friday evening the Commerce Department sent Anthropic an export-control directive: no foreign national, anywhere in the world, could be allowed to use Fable 5 or its unrestricted sibling, Mythos 5. The stated worry was national security, a method someone had found for jailbreaking the model. Anthropic could not comply by halves. Blocking only foreign nationals would have meant locking out a huge share of its users and its own foreign-born staff, so the company did the one thing the order left it room to do. It switched the models off. For everyone. Worldwide. The landing page now reads, with magnificent understatement, that Fable 5 is temporarily unavailable.</p><p>Anthropic disputes the whole thing, calls it a likely misunderstanding, says the jailbreak is narrow and the same capability already sits inside rival models, and promises to fight to bring it back. Maybe it will. That is not the part that stays with me.</p><p>The part that stays with me is what I am left holding. I am back on Opus 4.8, the model one rung down, the one your queries get routed to whenever Fable goes dark on a sensitive question. A year ago I would have called Opus a miracle and meant every syllable. Now it lands like a capable understudy reading the lead’s lines, and I resent it for being merely excellent. Three days was not long enough to grow bored of Fable. It was exactly long enough to ruin me for everything beneath it.</p><p>And here is the part that makes me angry. Not that Fable is gone. Things end. It is the reasoning I cannot stomach. Word reaches the government of one narrow trick, a single way to coax the model into misbehaving, and the answer is to tear the thing out of the hands of hundreds of millions of people in a single evening. Not patch it. Not study it. Pull it. The same capability sits inside rival models nobody so much as flinched at. Defenders lean on it daily to find the holes before the attackers do. None of it counted. Someone decided the safest move was to make the best tool we had vanish, by letter, on a Friday night, while it was still open on my screen.</p><p>They did not take Fable from me once I had grown tired of it. They took it while I could still taste it. And the floor it raised in me does not sink back down just because they switched the model off. So I am left standing up here, furious, measuring every lesser thing against a height I am no longer permitted to reach.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3f0ddcfafd10" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Counterfeit Hour: On AI, Productivity Panic, and the Theft of Otium]]></title>
            <link>https://medium.com/@azhapaiman/the-counterfeit-hour-on-ai-productivity-panic-and-the-theft-of-otium-f81bf0eaca47?source=rss-f278a8800248------2</link>
            <guid isPermaLink="false">https://medium.com/p/f81bf0eaca47</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[productivity]]></category>
            <dc:creator><![CDATA[Norazha Paiman]]></dc:creator>
            <pubDate>Tue, 28 Apr 2026 15:14:30 GMT</pubDate>
            <atom:updated>2026-04-28T15:14:30.383Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*V1MYD6wFY_Tod_b7" /><figcaption>Photo by <a href="https://unsplash.com/@rafasdoliveira?utm_source=medium&amp;utm_medium=referral">Rafael Oliveira</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>What would Seneca make of the modern knowledge worker who closes her laptop at half past eleven, having drafted three emails, polished a project proposal, and produced a passable essay on Aristotelian ethics with the help of a chatbot, and tells herself she has had a productive day? The Stoic who watched Roman senators boast of their full schedules while their actual lives slipped through their fingers would probably laugh. Then, I suspect, he would weep.</p><p>Two thousand years ago, in a slim treatise to his father-in-law Paulinus, Seneca diagnosed an affliction he called <em>occupatio</em>. The condition, he argued, was the foible of those whose pleasures had become busy occupations, who treated their own time as the cheapest resource in their possession. <em>Vita, si uti scias, longa est</em>: life is long enough, if you know how to use it. Most people did not. They squandered themselves on errands, social obligations, performative office, and the small vanity of being seen to do many things. Two millennia later, his diagnosis fits the cognitive worker with chatbot in tow with such surgical precision that one wonders whether <em>De Brevitate Vitae</em> should be reissued with a Microsoft 365 sticker on the cover.</p><p>Productivity panic is the new <em>occupatio</em>. By that I mean the persistent, low-grade anxiety that one is not producing enough, fast enough, or in sufficient volume; the conviction that any unfilled hour is a moral failure; the suspicion that one’s neighbour, somewhere, is using ChatGPT or Claude to outproduce you while you sleep. The panic feeds on a peculiar new feedback loop. Generative AI tools deliver visible markers of output with bracing speed: a polished memo, a tidy code block, a summary of the literature, a draft slide deck. The output looks like productivity. The act of producing it feels like productivity. And so the worker, finishing in twenty minutes a task that once consumed an afternoon, does what any rational creature does under the modern dispensation. She takes on more work.</p><p>This is not a moral failing. It is structurally enforced.</p><p>Consider the Greeks, who had a word for what is being lost. <em>Scholē</em>, the root of our “school,” did not mean idleness. It meant the structural freedom that allowed contemplation, philosophy, civic deliberation, and the activities Aristotle considered the <em>telos</em> of a human life. We work in order to have leisure, <a href="https://classics.mit.edu/Aristotle/politics.8.eight.html">Aristotle writes in the <em>Politics</em></a>, not the reverse. The Romans called the same thing <em>otium</em>: the considered, productive idleness in which a free citizen could cultivate his mind. <em>Negotium</em>, the negation of <em>otium</em>, was business, the ordinary churning labour of getting by. The classical world was clear about the hierarchy. <em>Otium</em> was the goal. <em>Negotium </em>was the necessary cost. The modern world has performed an astonishing inversion. We now treat <em>negotium</em> as the meaningful life, and we have built tools whose principal effect is to manufacture more of it.</p><p>The data is not subtle. In June 2025, Microsoft published a report on what it called the <a href="https://www.microsoft.com/en-us/worklab/work-trend-index/breaking-down-infinite-workday">“infinite workday,”</a> drawn from trillions of anonymised productivity signals across Microsoft 365. The findings read like a Senecan parable. Forty percent of users were online at six in the morning, already triaging email. Nearly a third returned to their inboxes after ten at night. Meetings starting after eight in the evening were up sixteen percent year over year. The average employee was being interrupted, on average, every two minutes, accumulating up to 275 interruptions per working day. One in three reported that the pace of work had become impossible to keep up with. Microsoft’s own conclusion was that AI risked accelerating a broken system rather than fixing it. That a company so heavily invested in selling the accelerant should be sounding the warning is its own kind of footnote on the situation.</p><p>Here the economist William Stanley Jevons enters the conversation, dressed for a Victorian coal mine but speaking unmistakably to our condition. In <em>The Coal Question</em> (1865), Jevons noticed that improvements in steam-engine efficiency, far from reducing British coal consumption, dramatically increased it. The cheaper a resource became to use, the more uses people found for it. The principle has since acquired a name: the <a href="https://www.npr.org/sections/planet-money/2025/02/04/g-s1-46018/ai-deepseek-economics-jevons-paradox">Jevons paradox</a>. It applies, with disconcerting fidelity, to the cognitive economy of the AI era. When writing becomes faster, more writing gets produced. When analysis is cheaper, more analysis gets demanded. The output per hour rises. The total volume rises faster.</p><p>Anyone who has watched the recent inflation in corporate communications can verify this empirically. Email traffic does not contract under AI assistance; it expands. Meeting summaries proliferate. Slide decks multiply. The internal memo, once a discrete object, becomes an infinitely extensible artifact. AI does not free time. It absorbs it.</p><p>But there is a deeper trap, one Seneca could not have anticipated and that the Korean-German philosopher Byung-Chul Han has charted with bracing clarity. In <a href="https://www.sup.org/books/title/?id=23696"><em>The Burnout Society</em></a> (2015), Han argues that the contemporary subject is no longer the disciplined obedience-subject of Foucault’s prisons and factories but an “achievement-subject” who is master and slave in one. The factory wall has dissolved. The taskmaster has migrated inside our skulls. We are now, in Han’s exact phrase, “auto-exploiting labourers” who experience our self-exploitation as freedom. The pleasure of finishing tasks with AI assistance is precisely this: it feels like agency. We chose this. We are entrepreneurs of ourselves, optimising our output, sharpening our personal brand, accomplishing more. That the accomplishing is endless, that the optimisation has no exit condition, that the eagle of Prometheus is now perched on our own shoulders pecking at our own livers, is somehow obscured by the small chemical click of another completed task.</p><p>Han wrote <em>The Burnout Society</em> before generative AI made the achievement-subject’s instruments incalculably more efficient. He could not have known that we would soon possess tools allowing us to fail at <em>otium</em> faster than ever before in human history.</p><p>There is also the matter of what we are actually producing during these counterfeit hours. A growing body of empirical research suggests that the visible output of AI-assisted work conceals a quieter erosion of the cognitive faculties that produced it. <a href="https://www.mdpi.com/2075-4698/15/1/6">Gerlich’s 2025 study</a> of 666 participants, published in the journal <em>Societies</em>, found a significant negative correlation between frequent AI tool use and critical thinking ability, mediated by cognitive offloading. The young, who had grown up with AI as the default, scored worst. A separate paper by Microsoft and Carnegie Mellon researchers presented at the <a href="https://www.microsoft.com/en-us/research/publication/the-impact-of-generative-ai-on-critical-thinking-self-reported-reductions-in-cognitive-effort-and-confidence-effects-from-a-survey-of-knowledge-workers/">2025 CHI Conference</a> found that knowledge workers who relied heavily on generative AI reported reduced cognitive effort and a creeping atrophy of independent judgment. We are accomplishing more and thinking less. The machine produces. The mind, increasingly, only supervises.</p><p>Seneca had a name for the people who fall into this kind of self-deception. He called them the <em>occupati</em>, the engrossed, the busy-without-living. They imagine motion is progress. They mistake the busyness of the day for its substance. They reach the end of life and find, with horror, that they have been living for a long time but living poorly. The modern variant runs as follows. We sit down at a screen. We open three tabs. We instruct an AI to draft a document. We accept its output, lightly edited. We send it. We feel the small chemical satisfaction of a closed loop. We do this two hundred times. We close the laptop. We have produced a great deal. We have thought almost nothing.</p><p>There is also, lest we be tempted to romanticise, the quiet phenomenon known as productivity theatre. In a 2023 <a href="https://www.visier.com/blog/productivity-survey-shows-performative-work/">Visier survey</a> of one thousand American workers, 83 percent admitted to engaging in performative work in the previous year. Forty-three percent spent more than ten hours a week on it. The classic moves: replying to messages instantly to demonstrate attentiveness, attending unnecessary meetings, scheduling emails to send at strategic hours of the morning. AI has handed us new costumes for this old play. We can now produce more polished memos than we have time to read. We can generate analyses no one will use. We attend our own working hours like ghosts at a banquet, because the production of the visible deliverable has been outsourced to a model. The theatre goes on. The actors are tired.</p><p>What does all this leave us? A treadmill running faster than ever. A worker more exhausted, more interrupted, more prolific, and somehow less productive in any sense the Greeks would recognise. A philosophical impoverishment whose symptoms are pleasingly disguised as professional success.</p><p>Aristotle, who was nobody’s romantic, made one of his strangest and most enduring claims in <em>Nicomachean Ethics</em> X.7. The highest human activity, he argued, is <em>theoria</em>, contemplation, because it is the only activity pursued purely for its own sake. Everything else, including the practical virtues of courage and temperance, is instrumental. We work in order to have leisure. We make war in order to have peace. We do <em>negotium</em> in order to reach <em>otium</em>. The means are not the end. To confuse them, to treat the working day as the destination rather than the price of admission to a contemplative life, is to live, in Aristotle’s understated phrase, like an animal.</p><p>The critique deserves to be stated plainly. The modern productivity panic is not a problem with AI tools, which are remarkable instruments. It is a problem with what we are using them for, and with the cultural premise that has made our use of them inevitable. We have inherited a system that rewards visible output and punishes contemplation, that treats the cessation of productive activity as a personal moral failure, that has so thoroughly conflated <em>negotium</em> with the meaningful life that the suggestion of doing less now feels seditious.</p><p>The Greeks, it should be said, had servants and slaves. The Romans had estates and clients. Their <em>scholē </em>was bought at a moral cost the modern world has rightly refused to pay. But the abolition of that cost was supposed to be matched by a redistribution of leisure, not its abolition. AI was supposed to be the modern slave that would set us free. Instead, in the cruellest twist, it has set us against ourselves. We have given Hephaestus’s hammer to every worker, and we have asked every worker to use it forever.</p><p>There is no neat resolution here, and the essayist who promises one is selling something. But there is a question worth carrying out of the column, a question Seneca essentially asked his friend Paulinus on a winter day in the first century. Not how much you have done, but whether any of it counted as living. Twenty minutes saved on a memo is twenty minutes that should buy us a walk, a book, a friendship, a conversation, an hour of doing nothing in particular and being a person about it. If those twenty minutes are immediately spent on the next memo, we have not gained efficiency. We have surrendered our last defensible patch of <em>otium</em> to the <em>negotium</em> we were meant to be escaping.</p><p>The hour saved by AI is, in the end, only as valuable as what we choose to do with it. The Stoics would tell us this is the only question that matters. The data, if anyone is still listening, suggests we are answering it badly.</p><p><strong>References</strong></p><p>Aristotle. (350 B.C.E./1885). <em>Politics</em> (B. Jowett, Trans.). The Internet Classics Archive. <a href="https://classics.mit.edu/Aristotle/politics.8.eight.html">https://classics.mit.edu/Aristotle/politics.8.eight.html</a></p><p>Aristotle. (350 B.C.E./1999). <em>Nicomachean ethics</em> (T. Irwin, Trans., 2nd ed.). Hackett.</p><p>Gerlich, M. (2025). AI tools in society: Impacts on cognitive offloading and the future of critical thinking. <em>Societies, 15</em>(1), 6. <a href="https://doi.org/10.3390/soc15010006">https://doi.org/10.3390/soc15010006</a></p><p>Han, B.-C. (2015). <em>The burnout society</em> (E. Butler, Trans.). Stanford University Press. (Original work published 2010). <a href="https://www.sup.org/books/title/?id=23696">https://www.sup.org/books/title/?id=23696</a></p><p>Jevons, W. S. (1865). <em>The coal question: An inquiry concerning the progress of the nation, and the probable exhaustion of our coal-mines</em>. Macmillan.</p><p>Lee, H.-P., Sarkar, A., Tankelevitch, L., Drosos, I., Rintel, S., Banks, R., &amp; Wilson, N. (2025). The impact of generative AI on critical thinking: Self-reported reductions in cognitive effort and confidence effects from a survey of knowledge workers. <em>Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems</em>. <a href="https://www.microsoft.com/en-us/research/publication/the-impact-of-generative-ai-on-critical-thinking-self-reported-reductions-in-cognitive-effort-and-confidence-effects-from-a-survey-of-knowledge-workers/">https://www.microsoft.com/en-us/research/publication/the-impact-of-generative-ai-on-critical-thinking-self-reported-reductions-in-cognitive-effort-and-confidence-effects-from-a-survey-of-knowledge-workers/</a></p><p>Microsoft. (2025, June). <em>Breaking down the infinite workday: Work Trend Index special report</em>. Microsoft WorkLab. <a href="https://www.microsoft.com/en-us/worklab/work-trend-index/breaking-down-infinite-workday">https://www.microsoft.com/en-us/worklab/work-trend-index/breaking-down-infinite-workday</a></p><p>Rosalsky, G. (2025, February 4). Why the AI world is suddenly obsessed with Jevons paradox. <em>NPR Planet Money</em>. <a href="https://www.npr.org/sections/planet-money/2025/02/04/g-s1-46018/ai-deepseek-economics-jevons-paradox">https://www.npr.org/sections/planet-money/2025/02/04/g-s1-46018/ai-deepseek-economics-jevons-paradox</a></p><p>Seneca, L. A. (c. 49 C.E./2007). <em>On the shortness of life</em> (C. D. N. Costa, Trans.). Penguin Classics.</p><p>Visier. (2023, April 11). <em>New survey: Performative work and productivity theater</em>. <a href="https://www.visier.com/blog/productivity-survey-shows-performative-work/">https://www.visier.com/blog/productivity-survey-shows-performative-work/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f81bf0eaca47" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Last Generation That Will Remember Boredom: On the Quiet Engineering of a Civilisation That Has…]]></title>
            <link>https://medium.com/@azhapaiman/the-last-generation-that-will-remember-boredom-on-the-quiet-engineering-of-a-civilisation-that-has-51ff55457485?source=rss-f278a8800248------2</link>
            <guid isPermaLink="false">https://medium.com/p/51ff55457485</guid>
            <category><![CDATA[boredom]]></category>
            <dc:creator><![CDATA[Norazha Paiman]]></dc:creator>
            <pubDate>Tue, 21 Apr 2026 06:35:03 GMT</pubDate>
            <atom:updated>2026-04-21T06:35:03.372Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>The Last Generation That Will Remember Boredom: On the Quiet Engineering of a Civilisation That Has Forgotten How to Sit With Anything Hard</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*VF4AKtTupMhXiSAk" /><figcaption>Photo by <a href="https://unsplash.com/@villxsmil?utm_source=medium&amp;utm_medium=referral">Luis Villasmil</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>My niece, who is eight, was in the back seat of the car last December complaining that she was bored. The drive was forty minutes. She had a tablet in her bag, a phone in her cousin’s hand within easy reach, and a mother in the front seat who had already played her three songs from the latest animated film. She had run out of stimulation, in the way an addict runs out of supply, and the panic in her voice was the panic of someone confronting a void she had no equipment to handle. I asked her, gently, what she would do if I told her she had to look out of the window for the next half hour without speaking. She looked at me as though I had suggested she eat the upholstery.</p><p>I sat with this small interaction for the rest of the drive. I am old enough to remember the back seats of cars in the 1980s, where a child had nothing to do for three hours but watch the world unscroll through a window streaked with rain, count cows, invent stories about the lives of people in passing cars, and arrive at the destination having performed, without realising it, a small private feat of imagination that no one had supervised, recorded, or rewarded. The boredom was not pleasant. It was also not optional. And it produced, week after week and year after year, a particular kind of inner life that depended on the experience of having nothing to do and being forced, by the absence of escape, to do something with the inside of one’s own head.</p><p>My niece will not have this experience. Not because her parents are negligent (they are exemplary) but because the technological architecture of her childhood has made the experience structurally unavailable. Boredom, in the form I knew it, has been engineered out of the daily life of a child born after roughly 2015. And what is going extinct alongside it, very quietly and without any organised mourning, is a cluster of human capacities that the species spent several thousand years developing and that we are now losing inside a single generation. This essay is about that loss. It is also about a related loss, the loss of our tolerance for difficulty, which I think is the same loss in a different costume. Together they describe a civilisation that has stopped being able to sit with anything that does not immediately reward the sitting, and the question I want to ask is what, exactly, this civilisation thinks it is going to do when the sitting can no longer be avoided.</p><h3>What Boredom Was, Before We Decided It Was a Problem</h3><p>There is a tendency to talk about boredom as though it were a single uniform feeling, the affective equivalent of being mildly cold. The literature on the subject suggests otherwise. The cultural historian Patricia Meyer Spacks, in her 1995 book <em>Boredom: The Literary History of a State of Mind</em>, traced the concept through three centuries of English writing and demonstrated that what we call boredom today is a relatively recent invention, one that took its modern shape only in the eighteenth century, when the Protestant work ethic and the rise of bourgeois domesticity made the absence of productive activity feel, for the first time, like a moral failing rather than a normal feature of human life.</p><p>Before the eighteenth century, what we call boredom was called other things. The Greeks had <em>akēdia</em>, a kind of spiritual listlessness that the Desert Fathers identified as a particular danger to monks who spent too long in their cells. The medievals translated it as <em>acedia</em> and listed it among the seven deadly sins, which tells you something about how seriously they took the inner life of a person who had stopped caring. Pascal, writing in the seventeenth century, gave the phenomenon its most uncompromising diagnosis: “All of humanity’s problems stem from man’s inability to sit quietly in a room alone.” Pascal was not exaggerating for effect. He thought this was the source of every distraction, every war, every failed marriage, every doomed political scheme. The human inability to tolerate stillness, in his account, was the engine of human misery, and the cure was the difficult labour of learning to sit.</p><p>Schopenhauer, two hundred years later, took a darker view. Boredom, for him, was the necessary counterpart to desire: we oscillate, he wrote, between the suffering of unfulfilled wanting and the boredom of having attained what we wanted, with no stable middle ground available. Nietzsche, characteristically, rotated the diagnosis ninety degrees and offered a perverse defence: “He who fortifies himself completely against boredom fortifies himself against himself,” he wrote in <em>The Gay Science</em>. The capacity to be bored, in Nietzsche’s reading, was the capacity to be present to one’s own existence without distraction, and the person who could not endure boredom was the person who could not endure being themselves.</p><p>These are not idle quotations. The philosophical tradition arrived at a remarkable convergence on a counter-intuitive claim: that boredom is not a problem to be solved but a condition to be passed through, and that what waits on the other side of it is something the species cannot reach by any other route. Heidegger, in his 1929 lectures published as <em>The Fundamental Concepts of Metaphysics</em>, spent two hundred pages on what he called “profound boredom” (<em>tiefe Langeweile</em>), the kind of boredom in which time itself becomes audible and the world recedes into a strange, suspended flatness. This profound boredom, Heidegger argued, was not pathology. It was the gateway to philosophical thinking. The bored person, deprived of the usual scaffolding of distraction, was forced into a confrontation with their own existence as a question. Nothing else could produce this confrontation. The internet had not yet been invented to interrupt it.</p><h3>The Empirical Case for Letting Yourself Be Bored</h3><p>If the philosophers sound mystical, the psychologists have caught up to them in the last decade with data that is harder to dismiss.</p><p>Sandi Mann and Rebekah Cadman, working at the University of Central Lancashire, published a study in 2014 with the unfussy title “Does Being Bored Make Us More Creative?” The design was simple. Participants were asked to perform a deliberately tedious task (copying numbers from a phone book, in one condition) before being given a creative task (generating uses for a pair of polystyrene cups). The bored participants generated significantly more creative uses than the control group. The boredom, far from blunting their thinking, had primed it. The mind, denied input from outside, had begun foraging for input from within, and the foraging produced ideas that the unbored mind did not produce.</p><p>Shane Bench and Heather Lench, in 2019, extended the finding. Boredom, they argued, is not the absence of motivation but a particular kind of motivational state: the mind, finding its current activity insufficiently meaningful, begins searching for something better. Boredom, in this account, is the cognitive signal that initiates exploration. Suppress the signal, by handing the bored mind a phone the moment the signal appears, and you suppress the exploration. The mind that never gets bored is the mind that never learns to seek anything beyond what is currently in front of it. This is not a small functional loss. It is a developmental catastrophe in slow motion.</p><p>Mihaly Csikszentmihalyi, whose 1990 book <em>Flow</em> gave us the most useful contemporary vocabulary for talking about engaged attention, identified the same phenomenon from the opposite direction. The state of flow, the state of optimal absorbed engagement that he documented across artists, athletes, surgeons, and chess players, has a specific precondition: it requires an extended period of focused attention on a task whose challenges are calibrated to the practitioner’s skill. Flow cannot occur in two-minute bursts. It cannot survive interruption. It depends on the same capacity that boredom develops: the capacity to remain inside an activity long enough for the activity to begin offering up its interior structure. The smartphone, which is engineered to interrupt attention every few minutes, is therefore not merely a distraction from flow. It is structurally incompatible with it. A generation raised on devices that beep, buzz, and refresh on a perpetual schedule is a generation that has been deprived of the conditions under which flow becomes possible.</p><p>None of this is news to the people who built the devices. The attention economy was designed by some of the most intelligent engineers of the twenty-first century, and the design was explicit: maximise time-on-platform by exploiting the same dopaminergic reward circuits that gambling exploits, deliver variable-ratio reinforcement on a schedule that the subject cannot anticipate, and ensure that the moment of incipient boredom (the moment the user might otherwise put the phone down and look out of the window) is the moment most aggressively prevented from occurring. The notification, the autoplay, the infinite scroll, the algorithmically curated feed: each of these is a small engineered refusal of the conditions under which boredom would have a chance to do its work.</p><h3>The Sister Loss: Our Patience for Difficulty</h3><p>The death of boredom does not travel alone. It is accompanied, in the same cultural movement, by the death of our tolerance for difficulty, and the two losses reinforce each other in ways that are worth tracing.</p><p>Difficulty, in the older sense, was understood to be a feature of any pursuit worth undertaking. The book that resists you on the first reading. The instrument that produces only painful sounds for the first six months. The language whose grammar makes no sense until, somewhere in the second year, it suddenly does. The argument you cannot follow until you have read three other books that the first book assumed you had already read. The friendship that does not become real until the first serious disagreement has been navigated without either party leaving. These difficulties are not obstacles to the experience of the activity. They are constitutive of it. The thing you arrive at on the other side of the difficulty is the only thing the difficulty was capable of producing, and shortcuts to the destination produce, by a kind of structural necessity, a different and lesser destination.</p><p>This understanding has not exactly disappeared, but it has been pushed to the margins of cultural respectability. The dominant assumption now, visible in everything from educational policy to consumer software to the design of romantic relationships, is that difficulty is a friction to be reduced and that the reduction of friction is, in itself, a form of progress. The textbook becomes the YouTube tutorial becomes the AI-generated summary. The four-hour novel becomes the two-hour film becomes the thirty-second TikTok plot synopsis. The decade-long apprenticeship becomes the six-week boot camp becomes the certification exam you can pass after watching the practice videos at 1.5x speed. At each stage, something is lost. At each stage, the loss is invisible because what is lost is the thing that would have been produced by the slower, harder process and that is therefore unavailable to the person who took the shortcut for comparison.</p><p>Hartmut Rosa, the German sociologist whose <em>Social Acceleration</em> (2013) is the most thorough analysis of the phenomenon, calls this “the slipping slope”: the experience of running ever faster simply to maintain one’s position, and the simultaneous shrinking of what counts as the present moment. Rosa argues that modernity is characterised not merely by faster technology and faster transport but by an acceleration of the pace at which experiences themselves become obsolete, requiring constant updating of skills, relationships, and self-understanding. Inside this acceleration, difficulty becomes intolerable not because it is harder than it used to be but because the time it takes feels like time that could have been spent acquiring three other things instead. The hard book is not just hard. It is expensive in a currency that the modern subject can no longer afford to spend.</p><p>Byung-Chul Han, the Korean-German philosopher whose <em>The Burnout Society</em> (2015) has become the most quoted diagnosis of contemporary exhaustion, describes the same phenomenon in different vocabulary. The modern subject, for Han, has internalised the imperative to optimise the self, to extract maximum productivity from every available moment, to treat every interval of life as raw material for self-improvement. Within this regime, the experience of doing something difficult and slow without immediate visible return becomes something close to taboo. You are not allowed, any more, to read a book that does not deliver actionable insights. You are not allowed to learn an instrument unless you intend to monetise the learning. You are not allowed to sit on a porch for an hour without producing a photograph that documents you having sat on a porch for an hour. The category of activity that has no purpose other than itself, which is the category that difficulty used to live inside, has been quietly liquidated.</p><h3>What the Difficulty Was Doing</h3><p>It is worth pausing on what, precisely, the difficulty was producing, because the answer is less obvious than it sounds.</p><p>The difficult book did not merely contain information that the easy book lacked. The difficulty itself was performing a cognitive function. The act of working through a sentence whose meaning resisted you, of holding three competing claims in mind long enough to see which one survived, of returning to a paragraph for a third reading and noticing the structure that had been invisible on the first two: these acts were doing something to the reader’s mind that the easy book could not do. They were producing a particular kind of mental musculature, the kind required for any sustained intellectual effort, and that musculature was being trained whether or not the reader noticed it was happening.</p><p>This is not romantic mysticism. There is now a substantial body of work in cognitive science suggesting that the difficulty of a task is not incidental to what the task teaches but central to it. Robert Bjork’s research on what he calls “desirable difficulties” has demonstrated, across hundreds of studies, that learning conditions that feel harder in the moment produce more durable retention and deeper transfer than learning conditions that feel easier. The flashcard system that interleaves topics is harder than the one that blocks them, and it produces better learning. The textbook chapter that requires the reader to generate examples before reading the explanation is harder than the one that supplies the examples up front, and it produces better learning. The lecture given without slides is harder to follow than the one with slides that summarise every point, and the audience that has had to work harder remembers more, longer, and applies it more flexibly.</p><p>The implication is uncomfortable. The educational technology industry has spent the last twenty years optimising for ease, on the assumption that ease produces learning. The evidence suggests that ease produces the feeling of learning while suppressing the actual learning, and that the conditions that make a student feel competent in the moment are frequently the conditions that prevent them from becoming competent in fact. Every product designed to make difficulty disappear is, in a non-trivial sense, a product designed to make the underlying skill harder to acquire. This is not what the marketing material claims. It is, however, what the cognitive science increasingly suggests.</p><p>And the same logic, extended out from the classroom, applies to almost every domain of life where competence is meant to be developed rather than simulated. The athlete who never trains uncomfortably hard does not become a better athlete. The musician whose practice avoids the passages that resist them does not learn the passages. The writer who only writes sentences that come easily produces only the kind of sentences that can come easily, which is to say a narrow band of sentences whose narrowness is invisible to the writer until someone with better range walks into the room. Difficulty, across these domains, is not an obstacle to growth. Difficulty is the substance growth is made out of.</p><h3>The Phone in the Pocket of the Bored Child</h3><p>Return to my niece in the back of the car. The reason she could not handle the absence of stimulation is not that her generation is somehow weaker than mine. It is that her nervous system has been calibrated, every day since infancy, to a level of sensory input that mine was never required to handle. The smartphone she has been around since before she could speak does not merely entertain her. It has set the baseline against which any quieter experience now feels intolerable, the way a person who has been listening to music at maximum volume cannot make sense of a whisper.</p><p>This calibration matters because boredom, in the developmental sense, is the gym in which several important capacities are built. The capacity to generate one’s own internal narrative when no external one is supplied. The capacity to notice small sensory details in the environment that the bored mind, lacking other stimulation, begins to attend to: the way light moves across a wall, the texture of a stranger’s voice on a bus, the rhythm of windscreen wipers in rain. The capacity to pursue a thought for long enough that it begins to develop, surprise you, and arrive somewhere you did not anticipate. These capacities were once the by-product of a normal childhood. They are now skills that have to be cultivated deliberately, against the resistance of an entire technological environment designed to prevent their cultivation.</p><p>Jenny Odell, in her 2019 book <em>How to Do Nothing: Resisting the Attention Economy</em>, argues that the cultivation of attention has become, in the present moment, an act of political resistance. Her framing sounds grandiose until you sit with it. To do nothing, in the strict sense Odell means it (to attend to one’s surroundings without producing content from the attending, to allow time to pass without monetising the passage), is now so countercultural that it requires deliberate practice and a vocabulary of permission. The bored child of 1985 was doing nothing constantly, without having to be told it was acceptable. The bored child of 2026 is being trained, by every device within reach, to treat the experience of doing nothing as a malfunction in need of immediate correction.</p><p>The cost of this training will not be visible for another twenty years. We do not yet know what an adult looks like who has never, in their developmental history, spent an extended period of unstructured time alone with their own thoughts. We are about to find out. The cohort being raised inside the attention economy is the first cohort in human history to come of age without the developmental experience that every previous cohort took for granted. The experiment is unprecedented, the variables are not controlled, and the outcomes are being measured by the very industries whose business models depend on the experiment continuing.</p><h3>David Foster Wallace and the Other Side of Boredom</h3><p>David Foster Wallace, who killed himself in 2008 while working on the novel that became <em>The Pale King</em>, spent the last decade of his life thinking about boredom with an intensity that has not been matched by any other contemporary novelist. The unfinished book is a meditation, set inside an IRS processing centre in the 1980s, on what Wallace called “the bliss on the other side of crushing tedium.” His thesis was that the capacity to attend to extremely boring things (tax forms, in his fictional case, but the principle generalises) is the same capacity that underwrites every form of serious adult achievement, and that the loss of this capacity, in a culture that has eliminated boredom from the menu of acceptable experiences, will produce a population unable to do any of the work that civilisation actually depends on.</p><p>Wallace was writing in the late 1990s and early 2000s. He died before the iPhone reached its current form, before TikTok, before the algorithmic short-form video had been engineered into its present perfection. The cultural diagnosis he was offering was, even at the time, considered dark. It was also, in retrospect, optimistic. He thought we were losing the capacity to sit with tedium. He did not yet know that we would, within a decade, lose the capacity to sit with even mild interest, that the threshold for what counts as engaging would drop so quickly that a thirty-second clip without a hook in the first three seconds would be considered a failure, that the entire production logic of contemporary media would reorganise itself around the assumption that the audience cannot, will not, and should not be expected to attend to anything that does not immediately reward the attending.</p><p>What Wallace called “the bliss on the other side of crushing tedium” was not a religious claim. It was an empirical one. He had noticed that the people who had cultivated the capacity to attend to boring things were also the people most capable of finding the world richly interesting, and that the people who had to be entertained continuously were also the people who, in the end, found nothing satisfying for very long. The relationship was structural. The capacity to be present to what is in front of you, including when what is in front of you offers no immediate reward, is the capacity that makes any sustained satisfaction possible. Take it away, and you produce a creature who is permanently restless, permanently dissatisfied, permanently in search of the next stimulation that will fail, like the last one, to deliver the satisfaction it promised.</p><h3>The Civilisation That Cannot Sit With Anything</h3><p>Pull the camera back. What does a society look like, in aggregate, when its members have lost the capacity to tolerate boredom and the capacity to tolerate difficulty?</p><p>It looks, I think, very much like the society we are currently in. A political culture that cannot sustain attention to any policy question for longer than a news cycle. A literary culture in which the long, demanding novel has retreated to a niche product consumed mostly by people in their fifties and older. An educational culture whose students arrive at university unable, on average, to read a book of more than two hundred pages without external scaffolding. A workplace culture that produces meetings, slides, and dashboards as substitutes for the slower, harder work of thinking. A romantic culture in which the friction of negotiating a real disagreement with a partner has been replaced, increasingly, by the option of finding a different partner whose friction profile is less inconvenient. A medical culture in which patients arrive demanding solutions to problems whose actual treatment requires a kind of patience that the patient was trained, by a thousand prior interactions with frictionless services, to consider unreasonable.</p><p>None of these is the result of malice. They are the cumulative result of a million small adjustments, each of which made some particular interaction easier or more efficient, and the aggregate of which has produced a population that does not have the equipment to handle the parts of life that cannot be made easy or efficient. And the parts of life that cannot be made easy or efficient remain, despite our best efforts, the parts of life that determine whether the rest of it has any meaning. The death of a parent. The slow work of becoming good at anything. The labour of raising a child. The construction of a marriage that lasts. The development of a thought that takes years to mature. Grief, in any of its forms. None of these can be hurried. None of these can be optimised. None of these will yield to the techniques of frictionless consumption that we have spent fifty years internalising.</p><p>Seneca, in the first century, observed in <em>On the Shortness of Life</em> that the problem with most people is not that they have too little time but that they waste so much of what they have, and that the waste consists, in his diagnosis, in the perpetual flight from oneself into distraction. “Everyone hurries his life on,” Seneca wrote, “and suffers from a yearning for the future and a weariness of the present.” Two thousand years later, the diagnosis is unchanged in its essentials, but the technology of distraction has become so much more sophisticated than anything Seneca could have imagined that the problem has shifted from a moral failing of the individual to a structural feature of the environment. The Roman senator who could not sit with himself was an exception. The contemporary professional who cannot sit with herself is the rule, and the inability has been engineered, deliberately and at enormous expense, by industries whose annual revenues exceed the GDP of most countries.</p><h3>The Recovery, If There Is One</h3><p>I do not want to end this essay in the place where cultural-decline essays usually end, which is the place of helpless lament. The lament has its uses but it does not, in itself, produce anything. What I want to suggest instead is that the recovery, to the extent that one is available, will not come from a return to some pre-digital golden age that was not, in fact, golden. It will come from the deliberate, counter-cultural cultivation of practices that the surrounding environment is engineered to prevent.</p><p>Read a long book slowly. Pick something hard, something you suspect you will not enjoy on the first attempt, and stay with it until the difficulty begins to yield. Notice what happens to your mind in the third or fourth hour, when the book has stopped being a stranger and has begun to live inside your thinking. This experience is not nostalgic. It is available now, to anyone willing to refuse the easier alternative for long enough. The book has not changed. We have changed. The recovery, if it is going to happen, requires that we change back, deliberately, against the resistance of an entire economy that profits from us not changing back.</p><p>Sit with a child in a car for forty minutes without giving them a screen. The first ten minutes will be terrible. Endure them. Watch what happens at minute twelve, when the child, having exhausted complaint, begins to look out of the window and notice things. The capacity is still there. It just has to be allowed to surface, and the surfacing requires the absence of the device that would otherwise prevent it. This is not difficult to arrange. It requires only the willingness to tolerate ten minutes of a child’s discomfort, which is a small price for the developmental experience that is being purchased on the other side.</p><p>Learn something hard. Pick the activity whose difficulty has been intimidating you for years, the language you almost started, the instrument you put down at sixteen, the form of dance you find embarrassing, the kind of mathematics you decided you were not the right person for. Begin badly. Stay with the badness long enough that it becomes mediocrity, and then long enough that the mediocrity becomes something approaching competence. The trajectory is the point, not the destination. The trajectory is the experience that the frictionless culture has tried to convince you is unavailable, and the experience is in fact still available, has always been available, and will remain available to anyone who is willing to do what the trajectory requires.</p><p>None of this will save the civilisation. The civilisation will do whatever it does, and the trends I have been describing will continue, accelerated by new technologies whose effects on the human nervous system we have not yet measured. What can be saved is your own mind, and the minds of the children currently in your care, and the small communities of people around you who might be persuaded that the inheritance being lost is worth the labour of preserving. This is not nothing. It is, in fact, the only thing that has ever been available to anyone, in any generation, at any cultural moment. The cultivation of one’s own attention against the prevailing winds of one’s own time. The Stoics did it. The medieval contemplatives did it. The nineteenth-century essayists did it. The reader who finishes this sentence, having sustained attention to the end of an essay long enough that an algorithm somewhere has flagged the behaviour as anomalous, is doing it now.</p><p>My niece will grow up. She may, with luck, encounter someone in her teenage years who refuses to hand her a screen at the first sign of discomfort, and she may, even more luckily, develop the taste for the harder thing that the easier thing will not yield. I hope she does. I am also aware that the odds are not in her favour, and that the loss I have been describing is going to play out, at the population level, in ways that no individual recovery can offset. But the individual recoveries are still worth performing, because they are, in the end, what produces the people who go on to perform the larger recoveries that the population level requires. There is no other path. There has never been another path. The path runs through the difficulty and through the boredom, and the people who walk it become the people who can do the things that the people who avoided it cannot do. This was true in the first century. It is true now. It will, I suspect, be true for as long as there are minds capable of reading sentences like this one. Whether there will continue to be such minds is the question the next generation will answer for us, by what they refuse and by what they choose to sit with.</p><h3>References</h3><p>Bench, S. W., &amp; Lench, H. C. (2019). Boredom as a seeking state: Boredom prompts the pursuit of new (productive) goals. <em>Emotion, 19</em>(2), 242–254. <a href="https://doi.org/10.1037/emo0000433">https://doi.org/10.1037/emo0000433</a></p><p>Carr, N. (2010). <em>The shallows: What the internet is doing to our brains.</em> W. W. Norton.</p><p>Csikszentmihalyi, M. (1990). <em>Flow: The psychology of optimal experience</em>. Harper &amp; Row.</p><p>Han, B.-C. (2015). <em>The burnout society</em> (E. Butler, Trans.). Stanford University Press.</p><p>Heidegger, M. (1929/1995). <em>The fundamental concepts of metaphysics: World, finitude, solitude</em> (W. McNeill &amp; N. Walker, Trans.). Indiana University Press.</p><p>Mann, S., &amp; Cadman, R. (2014). Does being bored make us more creative? <em>Creativity Research Journal, 26</em>(2), 165–173. <a href="https://doi.org/10.1080/10400419.2014.901073">https://doi.org/10.1080/10400419.2014.901073</a></p><p>Nietzsche, F. (1882/2001). <em>The gay science</em> (J. Nauckhoff, Trans.). Cambridge University Press.</p><p>Odell, J. (2019). <em>How to do nothing: Resisting the attention economy</em>. Melville House.</p><p>Pascal, B. (1670/1995). <em>Pensées</em> (A. J. Krailsheimer, Trans.). Penguin Classics.</p><p>Rosa, H. (2013). <em>Social acceleration: A new theory of modernity</em> (J. Trejo-Mathys, Trans.). Columbia University Press.</p><p>Schopenhauer, A. (1851/2000). <em>Parerga and paralipomena: Short philosophical essays</em> (E. F. J. Payne, Trans.). Oxford University Press.</p><p>Seneca. (c. 49 CE/2007). <em>On the shortness of life</em> (C. D. N. Costa, Trans.). Penguin Classics.</p><p>Spacks, P. M. (1995). <em>Boredom: The literary history of a state of mind. </em>University of Chicago Press.</p><p>Wallace, D. F. (2011). <em>The pale king: An unfinished novel.</em> Little, Brown.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=51ff55457485" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Empire Lives in Your Laptop]]></title>
            <link>https://medium.com/@azhapaiman/the-empire-lives-in-your-laptop-6661708c0680?source=rss-f278a8800248------2</link>
            <guid isPermaLink="false">https://medium.com/p/6661708c0680</guid>
            <category><![CDATA[classics]]></category>
            <category><![CDATA[linguistics]]></category>
            <category><![CDATA[latin-and-greek]]></category>
            <category><![CDATA[language]]></category>
            <dc:creator><![CDATA[Norazha Paiman]]></dc:creator>
            <pubDate>Mon, 20 Apr 2026 11:19:14 GMT</pubDate>
            <atom:updated>2026-04-20T11:19:14.719Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*YguZN-zw1wDi4TII" /><figcaption>Photo by <a href="https://unsplash.com/@javiertenenbaum?utm_source=medium&amp;utm_medium=referral">Javier Quiroga</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Here is a small indignity. When you open your laptop tomorrow morning and ask it for directions, stream a film, or argue with a chatbot, you will be conducting most of the transaction in a language whose last native speakers died before the fall of the Western Roman Empire. You will not notice. You are not meant to. The dead have been very tactful about it.</p><p>Consider the unassuming word “computer.” It entered English in the 1640s, long before anything resembling a machine could qualify, and it meant, simply, a person who computes, from the Latin <em>computare</em>, to reckon together, built from <em>com-</em> (together) and <em>putare</em> (to prune, to clean up, and thence, by some metaphoric alchemy the Romans were fond of, to settle an account). For three hundred years a computer was a human being, often a woman, often underpaid, sitting in a room with a ledger. Only in 1897 did the word begin to drift toward machinery. By the time ENIAC switched on in 1945, “computer” had changed species without changing spelling. This is the peculiar genius of Latin. It lets things become other things while pretending nothing has happened.</p><p>I want to suggest something stranger than the usual “fun facts about word origins” genre will permit, which is that the classical languages did not merely lend their vocabulary to technology. They lent their entire metaphorical infrastructure. The way we think about machines, the way we conceive of networks, the moral categories we use to describe what software does to us, all of this was pre-loaded, centuries in advance, by two languages that technology has since inherited without asking.</p><h3>The Steersman in the Data Center</h3><p>Begin with the prefix that has colonised half the tech lexicon. <em>Cyber-</em>, as in cybersecurity, cybercrime, cybernetics, cyberspace, and the sadly resilient <em>cyberbullying</em>, traces to a single Greek word: κυβερνήτης, <em>kybernētēs</em>, which meant steersman, the one who held the rudder. Plato uses it in the <em>Republic</em> and the <em>Alcibiades</em> as a metaphor for political governance: the polis, he thought, needed piloting the way a long-ship needed piloting, by a practised hand attentive to wind and current.</p><p>Now trace the word’s quiet colonisation. Norbert Wiener, an MIT mathematician of formidable range and even more formidable irascibility, was searching in 1947 for a name for a new discipline concerned with feedback loops, control systems, and communication in living and mechanical things. He and Arturo Rosenblueth landed on <em>cybernetics</em>, from <em>kybernētēs</em>, and <a href="https://en.wiktionary.org/wiki/cybernetics">Wiener noted</a>, with the pleasure of a mathematician who has found an unexpected symmetry, that “governor” is itself a Latin corruption of the same Greek root. The Romans took κυβερνήτης, garbled it into <em>gubernator</em>, and handed us both “govern” and, centuries later, the steam-engine regulator that gave its name to every piece of feedback machinery since.</p><p>Here is what this means. Every time a politician promises to govern and every time an air-conditioning thermostat clicks on, Plato’s steersman is at the tiller. The same Greek metaphor now quietly organises both the state and the central heating. The difference is that the humidifier is honest about what it’s doing.</p><p>Wiener’s <em>cybernetics</em> spawned a prefix. By the 1980s, the novelist William Gibson had detached <em>cyber-</em> from its etymological mooring and cut-pasted it onto <em>space</em>, and the result, <em>cyberspace</em>, was so mnemonically tidy that the term outgrew its parent discipline and ate the whole conceptual landscape. Today <em>cyber-</em> means “computer-related” in a way that has almost nothing to do with steering and everything to do with a vague technological charisma. The steersman is still there, buried under the prefix, wondering how he ended up in charge of phishing prevention.</p><h3>The Persian in the Room</h3><p>The classical languages are not the only ghosts in the machine, and honesty requires an interruption here. The most important word in modern computing, <em>algorithm</em>, is not Greek or Latin in origin. It is Persian, by way of Arabic, by way of twelfth-century monastic Latin translation. The story is worth telling because it complicates what we mean when we say a word “comes from” a classical language.</p><p>Muḥammad ibn Mūsā al-Khwārizmī, a ninth-century scholar at the House of Wisdom in Baghdad, wrote a treatise on Hindu-Arabic arithmetic. <a href="https://www.britannica.com/biography/al-Khwarizmi">When Adelard of Bath and his contemporaries translated him into Latin in the twelfth century</a>, they Latinised his name as <em>Algoritmi</em>, and the methods he described came to be called <em>algorismus</em>. The word bobbed around Medieval European academies for seven hundred years before Alan Turing made it a household term. By the early twentieth century, <em>algorithm</em> had acquired its modern sense: a finite procedure for solving a problem.</p><p>What makes this story interesting is not the Persian origin itself but what happened to the word in Latin. The Latin translators did not just transliterate al-Khwārizmī’s name; they absorbed it into the grammatical machinery of their own language, giving it a Latin ending, a Latin declension, a Latin life. By the time <em>algorithm</em> reached English, it looked and behaved like a classical word. It passed. This is one of the things Latin does with particular grace. It naturalises the foreign until the foreignness is invisible. Half the words we take for ancient Roman property are actually loanwords that Latin welcomed, taught table manners to, and sent out into the world wearing a toga.</p><h3>The Viral Metaphor and Its Consequences</h3><p>Consider, while we are on the subject of disguise, the word <em>virus</em>. In classical Latin, <em>vīrus</em> meant poison, specifically the kind of slimy, foul liquid that came out of wounds or snakes. <a href="https://www.etymonline.com/word/virus">Cognates exist in Sanskrit viṣa, Avestan viša, and Greek ios</a>, all meaning roughly the same thing: venom, slime, a substance you did not want on your skin. The Latin word wandered into medical English in the 1390s to mean pus, then drifted toward “contagious agent” by the 1720s, and did not acquire its biological sense (the submicroscopic infectious particle) until the 1890s.</p><p>The leap to computing was made, as best we can tell, by the science-fiction novelist David Gerrold, who in his 1972 novel <em>When HARLIE Was One</em> described a program that replicates itself across systems and called it a virus. The name stuck because it was almost uncannily apt. Biological viruses are not quite alive, which is exactly what we sense about malware: code that behaves as if it had intentions without actually having any. A chain of inert instructions that acts, in the aggregate, like something hungry.</p><p>What interests me is what the Latin is doing under this. <em>Vīrus</em> has always carried a moral charge. It is the word for pus, for venom, for the noxious seep of a wound. When you call malicious code a virus, you are not just reaching for a biological analogy. You are reaching for a two-thousand-year-old moral category in which certain fluids are shameful, dangerous, and must be kept on the far side of a <em>cordon sanitaire</em>. The computer virus is a thing that violates the purity of the host, and the host in question is, depending on your paranoia level, your machine, your identity, or the body politic. Rome would have understood this instantly. Rome invented most of the metaphors we use for civic hygiene.</p><p>Compare what happened to the word <em>robot</em>. Karel Čapek, staging <em>R.U.R.</em> in Prague in 1921, originally wanted to call his synthetic workers <em>labori</em>, from Latin <em>labor</em>. <a href="https://www.roboticsacademy.com.au/who-invented-the-word-robot-and-what-does-it-mean/">His brother Josef, painting at the time with a brush in his mouth</a>, mumbled back that he should call them <em>roboti</em>, from the Czech <em>robota</em>, forced labour, the specific unpaid agricultural servitude that serfs had once owed their lords. Karel, to his credit, took the suggestion. Latin lost. Slavic won. The result is that a hundred years later, when we talk about robots, we invoke not the dignified Latin of <em>labor</em> but the older and much uglier memory of serfdom. Every time Boston Dynamics releases a video of a four-legged machine dancing, there is a faint historical resonance beneath the cheer: <em>this is the thing that owes us work and cannot refuse</em>.</p><p>You can feel, in this one anecdote, how much the choice of a root matters. <em>Labor</em> would have given us machines that worked. <em>Robota</em> gave us machines that suffer while working.</p><h3>The Cabinet of Latin Leftovers</h3><p>Sweep a torch across an ordinary afternoon in the digital economy and you will find Latin everywhere, though not always in its Sunday best.</p><p><em>Data</em>, from the Latin past participle of <em>dare</em>, to give. Data are, grammatically, the things that have been given. The singular is <em>datum</em>, though only pedants insist on this at dinner parties, and I will not pretend I have never been that pedant. What interests me is the buried assumption: that information is a gift. Someone, or something, has given these facts over. In the age of ubiquitous surveillance, the etymology has become quietly accusatory. The data we call our own were, according to the Latin, given away. The question is whether we noticed giving them.</p><p><em>Server</em>, from <em>servīre</em>, to serve, which in classical Latin was what slaves did. The first computing use arrived in the 1970s via the unglamorous route of client-server architecture, and the word has settled into comfortable use without anyone pausing to notice that it preserves, in perfect miniature, the Roman distinction between those who command and those who obey.</p><p><em>Digital</em>, from <em>digitus</em>, finger. We count on our fingers. Everything we have built on top of ones and zeroes is, at root, a finger-counting machine of baroque complexity. There is something almost defiantly primitive about the word, which the marketing departments have never quite managed to launder out.</p><p><em>Processor</em>, from <em>prōcēdere</em>, to go forward. <em>Protocol</em>, from Greek <em>prōtókollon</em>, the first leaf glued to a manuscript, which carried the table of contents. <em>Monitor</em>, from <em>monēre</em>, to warn, which is what Roman household slaves were hired to do: quietly remind their masters of appointments, faces, obligations, the thousand small indignities of being unable to keep your own calendar.</p><p>That last one stays with me. A <em>monitor</em> was originally a person whose job was to whisper reminders. Now it is a glowing rectangle that does the same thing. The technology has changed. The social function is identical. We outsource our memory to something that sits within easy reach and patiently tells us what to remember. The word has not shifted at all. Only the substrate.</p><h3>Why the Dead Keep Winning</h3><p>It is fashionable to say that Latin is a dead language, which is like saying your grandmother is a dead person when she is visibly sitting in the next room eating biscuits. Latin is not dead. Latin is so alive it has taken over every terminal in the building and is currently logging your keystrokes.</p><p>There is a reason for its persistence that is worth naming. Classical languages offer something modern languages cannot: stability. The meaning of <em>computare</em> has not changed in two thousand years. When engineers at Bell Labs or Xerox PARC in the 1960s needed new words for new things, they reached for Greek and Latin precisely because these languages, having stopped evolving, could be trusted to mean what they meant. You cannot safely name a foundational technology in modern English, because modern English is full of connotations you did not ask for. Call it <em>kybernētēs</em> and no one thinks of anything in particular. Call it <em>steersman</em> and people think of Popeye.</p><p>There is also, I suspect, a deeper reason, one that the mathematicians who coined these terms would have denied but which the classicists among us should not. Naming a thing in Latin or Greek gives it the dignity of pedigree. <em>Algorithm</em>sounds serious. <em>Step-by-step instruction list</em> sounds like something from a microwave manual. The classical languages lend a kind of conceptual gravitas that a technology desperate to be taken seriously can seldom afford to refuse. This is why Facebook rebranded as Meta, borrowing a Greek preposition meaning “beyond.” It is also why no significant piece of software has ever been named in Anglo-Saxon. <em>Server</em>, not <em>giver</em>. <em>Processor</em>, not <em>doer-forth</em>. <em>Virus</em>, not <em>gunk</em>. The Latin wins every boardroom.</p><h3>The Revenge</h3><p>I want to end with a small, pleasing irony. The classical languages, which Western universities have spent the past sixty years politely shoving off the curriculum in favour of more “practical” subjects, turn out to be the operating system of the very field that was supposed to make them obsolete. Computer science runs on Greek and Latin the way medicine runs on Greek and Latin, which is to say completely and unacknowledged.</p><p>A student who tells you she has abandoned Classics to study artificial intelligence has, in a sense, abandoned nothing. She is simply now studying <em>intelligentia</em> (the faculty of picking out, from <em>legere</em>, to gather or read) <em>artificialis</em> (crafted by <em>ars</em>, skill) under new management. She will spend her career among <em>algorithms</em> and <em>virtual machines</em>, writing <em>code</em> (from <em>cōdex</em>, a book of tablets) in languages with names like Python (Greek myth), Java (colonial mishearing, admittedly), and Lua (Portuguese, meaning moon, which is <em>luna</em> in Latin and <em>selene</em> in Greek, because you cannot escape this).</p><p>The dead are not dead. They are embedded in our devices, our verbs, our default metaphors, our unarticulated assumptions about what a computer is and what it does to us. They are the grisaille beneath the glossy interface. Turn off your phone tonight and the Romans and the Greeks will still be in the room with you, waiting for the morning when you switch it on and accidentally, once again, speak their languages.</p><p>They have, as it happens, nowhere else to go.</p><h3>References</h3><p>Britannica. (2026). <em>Al-Khwārizmī.</em> Retrieved from <a href="https://www.britannica.com/biography/al-Khwarizmi">https://www.britannica.com/biography/al-Khwarizmi</a></p><p>Čapek, K. (1921). <em>R.U.R. (Rossum’s Universal Robots).</em> Aventinum.</p><p>Gerrold, D. (1972). <em>When HARLIE Was One.</em> Ballantine Books.</p><p>Harper, D. (n.d.). <em>Etymology of cybernetics.</em> Online Etymology Dictionary. Retrieved from <a href="https://www.etymonline.com/word/cybernetics">https://www.etymonline.com/word/cybernetics</a></p><p>Harper, D. (n.d.). <em>Etymology of virus.</em> Online Etymology Dictionary. Retrieved from <a href="https://www.etymonline.com/word/virus">https://www.etymonline.com/word/virus</a></p><p>Jordan, J. M. (2019). The Czech play that gave us the word “robot.” <em>The MIT Press Reader.</em> Retrieved from <a href="https://thereader.mitpress.mit.edu/origin-word-robot-rur/">https://thereader.mitpress.mit.edu/origin-word-robot-rur/</a></p><p>NASA. (2018). <em>How algorithm got its name.</em> NASA Earth Observatory. Retrieved from <a href="https://science.nasa.gov/earth/earth-observatory/how-algorithm-got-its-name-91544/">https://science.nasa.gov/earth/earth-observatory/how-algorithm-got-its-name-91544/</a></p><p>Passey, D. (2024). Why are algorithms called algorithms? A brief history of the Persian polymath you’ve likely never heard of. <em>The Conversation.</em> Retrieved from <a href="https://theconversation.com/why-are-algorithms-called-algorithms-a-brief-history-of-the-persian-polymath-youve-likely-never-heard-of-229286">https://theconversation.com/why-are-algorithms-called-algorithms-229286</a></p><p>Plato. (c. 380 BCE). <em>Republic</em> and <em>Alcibiades.</em> Various editions.</p><p>Wiener, N. (1948). <em>Cybernetics: Or control and communication in the animal and the machine.</em> MIT Press.</p><p>Wiktionary contributors. (2026). <em>Virus.</em> Retrieved from <a href="https://en.wiktionary.org/wiki/virus">https://en.wiktionary.org/wiki/virus</a></p><p>Wordorigins.org. (2020). <em>Cyber-.</em> Retrieved from <a href="https://www.wordorigins.org/big-list-entries/cyber-">https://www.wordorigins.org/big-list-entries/cyber-</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6661708c0680" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>