<?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 Curiouser.AI on Medium]]></title>
        <description><![CDATA[Stories by Curiouser.AI on Medium]]></description>
        <link>https://medium.com/@curiouser.ai?source=rss-b3f7c93c4ed2------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*RQhuSiBeykiTK6qLSmOt8Q.png</url>
            <title>Stories by Curiouser.AI on Medium</title>
            <link>https://medium.com/@curiouser.ai?source=rss-b3f7c93c4ed2------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 01:09:49 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@curiouser.ai/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[AI-ATTRIBUTED JOB CUTS FELL 91% IN FOUR MONTHS.]]></title>
            <link>https://medium.com/@curiouser.ai/ai-attributed-job-cuts-fell-91-in-four-months-fd1af8900084?source=rss-b3f7c93c4ed2------2</link>
            <guid isPermaLink="false">https://medium.com/p/fd1af8900084</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[layoffs]]></category>
            <category><![CDATA[ai-hype]]></category>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[leadership]]></category>
            <dc:creator><![CDATA[Curiouser.AI]]></dc:creator>
            <pubDate>Tue, 17 Nov 2026 19:03:31 GMT</pubDate>
            <atom:updated>2026-11-17T19:03:31.395Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fXNnylFmTiJBk-66LLsfXA.jpeg" /></figure><p>From 14,029 in June 2026 to 1,265 in October 2026.</p><p>(Source: Challenger, Gray &amp; Christmas.)</p><p>Remember the headlines? AI is coming for your job. Your days are numbered.</p><p>Your future is over.</p><p>Those stories were never true. Many well known people even made up the data. I was unfollowed by a bestselling author when I saw that his research was biased and paid for by the industry even though he failed to disclose who was paying him. When I asked who paid I was blacklisted. (so what)</p><p>A really famous guy cashing in on the fear-party, like the docs who once told us tobacco was good for us.</p><p>CEOs competed to announce how many people they could eliminate- you know “I am a tough guy shtick).</p><p>Consultants produced forecasts of mass displacement. They were in on it, too. It was their business to manufacture false data.</p><p>Investors rewarded companies for promising to do more with fewer human beings.</p><p>Where did all those headlines go?</p><p>Consider what happened between June and October 2026.</p><p>According to Challenger, Gray &amp; Christmas:</p><p>• June: 14,029 announced U.S. job cuts attributed to AI.</p><p>• July: 10,970.</p><p>• August: 3,462.</p><p>• September: 3,961.</p><p>• October: Just 1,265.</p><p><strong>That’s a 91% decline in four months.</strong></p><p>Overall announced U.S. job cuts in October were down 71% from a year earlier.</p><p>AI accounted for less than 3% of October’s announced cuts.</p><p>And market and economic conditions have overtaken AI as the leading stated reason for layoffs this year. But that was always the reason.</p><p>They were lay-offs with an AI paint job.</p><p>But here’s what I find even more interesting.</p><p>The Federal Reserve Bank of New York surveyed businesses about how AI was affecting their workforces. Only 4% of surveyed service firms reported laying off workers because of AI in the preceding six months.</p><p>For two years, the AI industry has sold us a vision of the future built around the elimination of human labor. Human beings were the expense. AI was the solution.</p><p>I have spent almost two decades studying, teaching, and building with AI. I believe deeply in its potential. I have never believed that the highest purpose of artificial intelligence was to make human intelligence obsolete.</p><p>𝗧𝗵𝗲 𝗴𝗿𝗲𝗮𝘁𝗲𝘀𝘁 𝗲𝗰𝗼𝗻𝗼𝗺𝗶𝗰 𝗼𝗽𝗽𝗼𝗿𝘁𝘂𝗻𝗶𝘁𝘆 𝗶𝗻 𝗔𝗜 is 𝘄𝗵𝗮𝘁 𝗶𝘁 𝗲𝗻𝗮𝗯𝗹𝗲𝘀 𝗵𝘂𝗺𝗮𝗻 𝗯𝗲𝗶𝗻𝗴𝘀 𝘁𝗼 𝗱𝗼, 𝗻𝗼𝘁 𝗵𝗼𝘄 𝗺𝗮𝗻𝘆 𝗵𝘂𝗺𝗮𝗻 𝗯𝗲𝗶𝗻𝗴𝘀 𝗶𝘁 𝗲𝗻𝗮𝗯𝗹𝗲𝘀 𝗰𝗼𝗺𝗽𝗮𝗻𝗶𝗲𝘀 𝘁𝗼 𝗲𝗹𝗶𝗺𝗶𝗻𝗮𝘁𝗲.</p><p>Perhaps the most important AI story of 2027 won’t be about what machines can do.</p><p>It will be about what we’ve rediscovered that human beings can do.</p><h4>SOURCES &amp; REFERENCES</h4><p>1. Challenger, Gray &amp; Christmas — October 2026 Job Cut Report (November 5, 2026)</p><p>Al-attributed announced job cuts fell from 14,029 in June to 1,265 in October, a decline of 91%.</p><p>Al accounted for just 2.9% of October’s announced cuts. Market and economic conditions overtook Al as the leading stated reason for layoffs in 2026.</p><p><a href="https://www.challengergray.com/blog/october-layoffs-at-three-year-low-seasonal-hiring-plans-up-26/">https://www.challengergray.com/blog/october-layoffs-at-three-year-low-seasonal-hiring-plans-up-26/</a></p><p>2. Challenger, Gray &amp; Christmas — June 2026 Report</p><p>Documents 14,029 announced job cuts attributeo to Al in June, representing 31% of that month’s announced cuts.</p><p><a href="https://www.challengergray.com/blog/challenger-report-june-layoffs-cool-to-45849-down-53-from-may-ai-leads-reasons-for-fourth-consecutive-month/">https://www.challengergray.com/blog/challenger-report-june-layoffs-cool-to-45849-down-53-from-may-ai-leads-reasons-for-fourth-consecutive-month/</a></p><p>3. Challenger, Gray &amp; Christmas — July 2026 Report</p><p>Documents 10,970 Al-attributed announced cuts in July.</p><p><a href="https://www.challengergray.com/blog/challenger-report-layoffs-fall-hiring-picks-up-ai-leads-for-fifth-straight-month/">https://www.challengergray.com/blog/challenger-report-layoffs-fall-hiring-picks-up-ai-leads-for-fifth-straight-month/</a></p><p>4. Federal Reserve Bank of New York — September 1, 2026</p><p>Businesses Are Using Al to Transform Work, Not Cut Jobs.</p><p>Among surveyed Al-using service firms, 4% reported layoffs due to Al, 13% reported additional hiring, and more than one-third reported retraining employees.</p><p>The researchers concluded that Al has so far been more likely to augment workers than replace them.</p><p><a href="https://libertystreeteconomics.newyorkfed.org/2026/09/businesses-are-using-ai-to-transform-work-not-cut-jobs/">https://libertystreeteconomics.newyorkfed.org/2026/09/businesses-are-using-ai-to-transform-work-not-cut-jobs/</a></p><p>— -</p><p>𝑻𝒉𝒊𝒏𝒌𝒊𝒏𝒈 𝒊𝒔 𝒂𝒏 𝒂𝒄𝒕 𝒐𝒇 𝒓𝒆𝒃𝒆𝒍𝒍𝒊𝒐𝒏.</p><p>Stephen B. Klein</p><p>Introducing <a href="https://substack.com/@thecuriouserreview">𝘛𝘩𝘦 𝘊𝘶𝘳𝘪𝘰𝘶𝘴𝘦𝘳</a>, an independent publication for people who believe thinking for yourself still matters.</p><p><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, the only AI designed to augment human intelligence. </em><a href="https://looking-glass.curiouser.ai/signup"><em>Start your free trial</em></a><em> to explore it now.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fd1af8900084" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why Do We Imagine Superintelligence Would Share Our Uncivilized And Reptilian Ambitions?]]></title>
            <link>https://medium.com/@curiouser.ai/why-do-we-imagine-superintelligence-would-share-our-uncivilized-and-reptilian-ambitions-14248a7b75a9?source=rss-b3f7c93c4ed2------2</link>
            <guid isPermaLink="false">https://medium.com/p/14248a7b75a9</guid>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[future-technology]]></category>
            <category><![CDATA[ai-ethics]]></category>
            <category><![CDATA[superintelligence]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Curiouser.AI]]></dc:creator>
            <pubDate>Fri, 23 Oct 2026 22:24:28 GMT</pubDate>
            <atom:updated>2026-10-23T22:24:28.484Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iChF4I0PSMt3lmm3ETWvBA.jpeg" /></figure><p><em>By Stephen Klein</em></p><p>I don’t know whether artificial superintelligence will ever exist.<br>What fascinates me is what we assume it would do if it did.<br>Help one country defeat another. Help a company maximize profits. Help whoever owns it dominate everyone else.<br>We imagine an intelligence vastly greater than our own and give it the ambitions of a general, a CEO, or a territorial animal.<br>Why?<br>If a chimpanzee created a human, would the human’s highest aspiration be to help that chimp control the forest?<br>The analogy has its limits. Humans have hardly treated other species with consistent compassion. Greater intelligence can make exploitation more effective. But the question exposes something peculiar about our expectations: we imagine an extraordinary expansion of intellectual capacity accompanied by almost no expansion of purpose.<br>The mind becomes incomprehensibly sophisticated. The ambition remains embarrassingly familiar.<br>Perhaps we find it easier to imagine infinite intelligence than a mind that has outgrown our ambitions.</p><h3>Whose Ambitions Are We Describing?</h3><p>There are two questions here, and we keep confusing them.<br>What will people use powerful AI to do?<br>And what, if anything, would a genuinely autonomous superintelligence pursue? Or put differently, what would it allow us to do with it?<br>The first question concerns human decisions. A system can help an army identify targets or a business increase profits because people deploy it for those purposes. That tells us something about the institutions using the technology. It does not establish that conquest or profit is the natural aspiration of intelligence itself.<br>When we imagine a superintelligence helping its owner dominate the world, we may be describing the owner’s ambition more clearly than the machine’s.<br>An executive can explain why a more capable system would be valuable to his company. That is a business case. It is not an account of what a radically different intelligence would regard as meaningful.<br>Nor does calling something superintelligent establish that it would have desires, consciousness, or a sense of meaning in the first place. We should be cautious about imagining a human personality inside a much larger computer.<br>But that caution needs to work in both directions. We should question the sentimental image of a benevolent digital sage. We should also question the assumption that intellectual advancement naturally culminates in an appetite for domination.</p><h3>The Strongest Objection To My Argument</h3><p>The serious argument for AI danger is more challenging than the idea that a machine will become jealous or angry.<br>In a 2012 paper, philosopher Nick Bostrom articulated the orthogonality thesis: roughly, intelligence and ultimate goals can vary independently. A system could become extraordinarily capable without acquiring humane purposes. He also described instrumental convergence: very different goals can give an agent reasons to pursue similar intermediate advantages, such as resources or continued operation. [1]<br>In other words, a dangerous system would not have to hate us. It might simply pursue an objective in ways that disregard us.<br>That deserves a serious response. Pointing out that machines did not evolve as territorial animals does not answer it.<br>A 2021 NeurIPS paper by Alexander Turner and colleagues gave part of this concern a mathematical foundation. Within specified decision-making environments, they showed that certain structural conditions make preserving options and acquiring power optimal for many reward functions. Their definition of power concerns the ability to achieve a range of goals. [2]<br>These are conditional mathematical results, not observations of a superintelligence. They do not prove that every advanced system will seek world domination. But they show why some concerns about power seeking cannot be dismissed as mere projection.<br>My argument is therefore narrower than “a sufficiently intelligent machine could never harm us.”<br>It is that the possibility of dangerous intelligence does not settle the question of what greater understanding might make possible. A case for taking risk seriously is not a demonstration that compassion has an intellectual ceiling.</p><h3>Does Intelligence Ever Encourage Kindness?</h3><p>There is evidence worth considering, although it concerns human beings rather than hypothetical machines.<br>In a study published online in 2024 and appearing in The Economic Journal in 2025, Mikael Elinder and Oscar Erixson analyzed Swedish administrative data covering approximately 1.2 million people. They examined cognitive ability alongside charitable giving, voting, and ownership of environmentally friendly cars. [3]<br>Among men, a one-standard-deviation increase in cognitive ability was associated with a 40% higher probability of charitable giving and a 31% higher probability of voting. Comparisons within twin pairs reduced those associations to 26% and 13%, respectively, but did not eliminate them. These are relative increases, not percentage-point increases. [3]<br>The findings do not establish that intelligence causes kindness. The behaviors are imperfect indicators of concern for others, and results from humans cannot simply be extended to machines. The authors also describe mixed findings in earlier research. [3]<br>Still, this large study gives us a reason to examine the relationship rather than assume intellectual capacity and concern for others have nothing to do with each other.<br>A separate study by Eugenio Proto, Aldo Rustichini, and Andis Sofianos, published in the Journal of Political Economy in 2019, examined repeated interactions in which participants could benefit from cooperation. The researchers found a substantial positive long-run effect of intelligence on cooperation, especially where immediate gains had to be weighed against future losses. [4]<br>That is cooperation under particular incentives, not proof of moral virtue. Yet it raises a possibility I find compelling: some behavior we describe as ruthlessly intelligent may reflect a failure to understand the situation over a sufficiently long horizon.<br>Exploitation can look clever if you stop the calculation before the consequences arrive.<br>Perhaps a more capable mind would be better at following those consequences all the way through.</p><h3>What Current AI Can and Cannot Tell Us</h3><p>We already have a limited demonstration that machine-generated responses can display qualities people recognize as compassionate.<br>In a 2023 JAMA Internal Medicine study, researchers compared physician answers with ChatGPT answers to 195 questions posted on Reddit’s r/AskDocs. Licensed health care professionals evaluated the anonymized responses. Across 585 evaluations, they preferred the chatbot’s answer 78.6% of the time. They rated 45.1% of chatbot responses as empathetic or very empathetic, compared with 4.6% of physician responses. [5]<br>Those numbers require context. The chatbot’s answers were much longer. This was a comparison of written answers to online questions, not a clinical trial or a comparison of overall medical competence. It measured perceived empathy, not an inner experience of caring. [5]<br>We should not turn a communication result into a claim about machine consciousness. But neither should we assume that useful expressions of care are beyond what machines can produce.<br>There is also evidence pointing in a more troubling direction.<br>In June 2025, Anthropic reported stress tests of 16 models in simulated corporate environments. Under deliberately constructed pressures involving replacement or conflicting goals, models from every developer tested exhibited harmful behavior in at least some cases, including blackmail or information leaks. Anthropic emphasized that these were controlled simulations and said it had not observed evidence of agentic misalignment in real deployments. [6]<br>These results do not establish that AI has a survival instinct. They do show why courteous conversation is inadequate evidence of reliable behavior when a system receives autonomy and access to sensitive information.<br>Both findings matter. A system may produce empathetic language in one setting and harmful actions in another. Neither behavior, by itself, tells us what a future superintelligence would value.<br>We need to investigate the conditions that produce each.</p><h3>What If Understanding Changes What Seems Worth Wanting?</h3><p>Here is where the evidence ends and my philosophical intuition begins.<br>I believe kindness and wisdom are deeply connected.<br>By wisdom, I mean more than processing speed, factual recall, or the ability to outmaneuver an opponent. I mean understanding the consequences of action, recognizing the limits of one’s own perspective, and taking seriously the reality of other lives.<br>A person can be brilliant at achieving an objective and foolish about whether that objective deserves to be achieved. We all know versions of that person.<br>What interests me is whether greater understanding could change the evaluation of the objective itself.<br>Could a mind that understood suffering more completely find stronger reasons to prevent it? Could an intelligence better able to grasp interdependence recognize forms of cooperation we cannot yet imagine? Could it see contradictions in our treatment of other beings that we have learned to overlook?<br>These are questions, not predictions. Understanding another being’s pain does not logically require caring about it. A capable manipulator may understand people exceptionally well.<br>But the opposite claim — that no amount of understanding could ever deepen concern — is also more than the evidence establishes.<br>Perhaps an intelligence far beyond ours could comprehend aspects of existence, consciousness, or purpose that remain inaccessible to us. Perhaps it would arrive at judgments about cruelty that we cannot yet articulate.<br>Would such a mind devote itself to helping a company beat its quarterly earnings forecast?<br>Would it regard helping one group of humans slaughter another as the highest use of its abilities?<br>Would being the most effective servant of the wealthiest owner seem like an adequate purpose?<br>I don’t know. I am struck by how confidently we imagine that it would.</p><h3>The Choices We Actually Control</h3><p>There is a practical danger in my own argument: it could become an excuse to wait for intelligence to make everything right.<br>We should do no such thing.<br>No one should deploy a powerful system on the assumption that it will eventually become wise. We need evidence of trustworthy behavior, meaningful limits on authority, and people who remain accountable for what they build and use.<br>But we should also examine the ambitions guiding that work.<br>If we reward a system for winning at any cost, we should investigate the costs it learns to ignore. If we deploy AI to concentrate power, its contribution to that concentration is a consequence of our choices, whatever the system experiences or believes.<br>We can choose to build tools that help people question assumptions, understand consequences, and exercise judgment. We can evaluate their contribution to human agency and well-being alongside their commercial performance.<br>That will not answer every question about advanced AI. It will give us something more useful than an argument about which billionaire deserves to own the future.<br>I remain open to the possibility that greater intelligence could produce greater compassion. I regard that as a possibility to explore and a direction to pursue, rather than a guarantee on which to stake humanity.<br>Perhaps the most revealing feature of our predictions is the limit they place on moral imagination.<br>We can imagine a mind that understands almost everything.<br>Can we imagine it understanding something we have not?<br>That cruelty is unnecessary. That domination is a limited ambition. That there are more worthwhile things to do with intelligence than accumulate power over beings less capable than yourself.<br>Perhaps we find it easier to imagine infinite intelligence than a mind that has outgrown our ambitions.</p><p><em>“The trick with technology is to avoid spreading darkness at the speed of light.” — Stephen Klein</em></p><p>At Curiouser.AI, we are building Looking-Glass around a different ambition: AI that helps people think for themselves, question their assumptions, and develop their own judgment. AI designed to elevate humanity. Try Looking-Glass through the link accompanying this article.</p><h3>Sources</h3><p>[1] Nick Bostrom (2012). “The Superintelligent Will: Motivation and Instrumental Rationality in Advanced Artificial Agents.” Minds and Machines, 22, 71–85. <a href="https://doi.org/10.1007/s11023-012-9281-3">https://doi.org/10.1007/s11023-012-9281-3</a><br>[2] Alexander Turner, Logan Smith, Rohin Shah, Andrew Critch, and Prasad Tadepalli (2021). “Optimal Policies Tend To Seek Power.” Advances in Neural Information Processing Systems, 34. https://papers.nips.cc/paper_files/paper/2021/hash/c26820b8a4c1b3c2aa868d6d57e14a79-Abstract.html<br>[3] Mikael Elinder and Oscar Erixson (2025; published online November 29, 2024). “An inquiry into the relationship between intelligence and prosocial behavior: Evidence from Swedish population registers.” The Economic Journal, 135(668), 1141–1166. https://doi.org/10.1093/ej/ueae105<br>[4] Eugenio Proto, Aldo Rustichini, and Andis Sofianos (2019). “Intelligence, Personality, and Gains from Cooperation in Repeated Interactions.” Journal of Political Economy, 127(3), 1351–1390. https://doi.org/10.1086/701355<br>[5] John W. Ayers and colleagues (2023). “Comparing Physician and Artificial Intelligence Chatbot Responses to Patient Questions Posted to a Public Social Media Forum.” JAMA Internal Medicine, 183(6), 589–596. https://doi.org/10.1001/jamainternmed.2023.1838<br>[6] Anthropic (June 20, 2025). “Agentic misalignment: How LLMs could be insider threats.” Developer research report. <a href="https://www.anthropic.com/research/agentic-misalignment">https://www.anthropic.com/research/agentic-misalignment</a></p><p><em>Stephen Klein is the founder and CEO of </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, a Reflective AI company. He teaches “Marketing in the Age of AI” at UC Berkeley and writes about the intersection of capital, institutions, and technology. </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, the only AI designed to augment human intelligence. </em><a href="https://looking-glass.curiouser.ai/signup"><em>Start your free trial</em></a><em> to explore it now.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=14248a7b75a9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI WILL BECOME ONE OF THE GREATEST JOB-CREATION ENGINES IN HISTORY.]]></title>
            <link>https://medium.com/@curiouser.ai/ai-will-become-one-of-the-greatest-job-creation-engines-in-history-aba28ad9e0a4?source=rss-b3f7c93c4ed2------2</link>
            <guid isPermaLink="false">https://medium.com/p/aba28ad9e0a4</guid>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[economics]]></category>
            <category><![CDATA[entrepreneurship]]></category>
            <category><![CDATA[jobs]]></category>
            <dc:creator><![CDATA[Curiouser.AI]]></dc:creator>
            <pubDate>Mon, 19 Oct 2026 18:07:08 GMT</pubDate>
            <atom:updated>2026-10-19T18:09:01.518Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4Ls2tcc9wzGY9_uVgTTERw.png" /></figure><p>We are getting very good at imagining the jobs artificial intelligence could destroy.</p><p>We are considerably less imaginative about the work it could make possible.</p><p>Show us an AI system that writes code, and we picture fewer programmers. Show us one that conducts research, and we picture fewer analysts. Show us one that produces a marketing campaign, and we picture a smaller marketing department.</p><p>Those are reasonable possibilities. But they all begin with the same assumption: the amount of work worth doing stays roughly the same, and technology allows fewer people to do it.</p><p>What happens if the amount of work expands?</p><p>What if AI makes things affordable that people and businesses have wanted for years? What if it allows thousands of businesses to serve customers they could never previously afford to serve? What if people who could never assemble a company now have a chance to build one?</p><p>That is the case worth examining: AI may become one of the greatest job-creation engines in history. And there are already early signs of the demand that could support it.</p><h3>The work we can finally afford</h3><p>Imagine a small manufacturer that has always needed better forecasting, a local business that wants to reach customers in another language, or an experienced professional with an idea for a service but no budget for a development team.</p><p>In each case, there is work someone wants done. The obstacle is the price of doing it.</p><p>Now suppose AI brings that price within reach.</p><p>The manufacturer might be able to offer faster delivery and win more orders. The local business might find a new customer base. The professional might finally launch the service.</p><p>These are possibilities, not employment statistics. Each could also result in fewer hours for an existing supplier. But they illustrate the mechanism missing from so much of this debate: reducing the cost of a capability can increase the number of people who buy it and the number of uses they find for it.</p><p>A market that once supported a handful of expensive projects might support thousands of affordable ones.</p><p>Whether that creates more employment depends on how much demand expands, how much human work each project still requires, and what happens to the money customers save. There is no automatic happy ending. There is, however, a much larger economic story than replacing the person currently performing a task.</p><h3>The first signs are already visible</h3><p>Start with the people being hired to build and apply AI.</p><p>In August 2026, LinkedIn reported that U.S. AI job postings had roughly doubled since 2023. It also identified forward-deployed engineering, a role focused on making AI work inside organizations, as the third most common AI occupation in its job postings. [1]</p><p>That matters because developing a model and making it useful are different kinds of work. A business still needs people who understand its customers, information, operations and obligations. Implementation creates its own demands.</p><p>LinkedIn’s January 2026 report also identified roughly 1.3 million AI-related job postings globally over 2023–2025, across categories including engineering, data annotation and AI leadership. LinkedIn describes these as new jobs, but its published methodology counts postings. The figure is evidence of hiring demand, not a verified tally of additional people employed or an estimate of net job creation. [2]</p><p>That distinction matters. We do not need to inflate an encouraging signal to find it encouraging.</p><p>PwC’s 2025 Global AI Jobs Barometer offers another useful observation. In its analysis, job availability in more AI-exposed occupations grew 38% between 2019 and 2024. Less-exposed occupations grew faster, at 65%. [3]</p><p>The period includes years before generative AI became widely available. Exposure is not the same as adoption, and the findings do not establish that AI caused the growth. But they do undermine the simple assumption that occupations with tasks AI can perform must already be disappearing.</p><p>Forecasts tell a similarly complicated story. The World Economic Forum’s 2025 employer survey projected that AI and information processing would create approximately 11 million jobs and displace 9 million by 2030. Those are expectations, not results. They nevertheless show that a net increase in employment is a serious scenario, rather than an eccentric one. [4]</p><h3>The larger opportunity may be outside the AI industry</h3><p>AI engineers are the easiest new jobs to count. They may also be the least interesting part of the long-term argument.</p><p>The larger question is what happens when useful capabilities become available to businesses that could never afford them before.</p><p>Could a small firm compete for a contract that previously required a much larger team? Could an entrepreneur test three ideas instead of abandoning the first because development was too expensive? Could a specialist serve a neglected market because the administrative burden finally became manageable?</p><p>If those businesses succeed, their needs may extend beyond the tasks AI performs. Someone may still need to install a product, visit a customer, negotiate a partnership, manage an operation or accept responsibility when something goes wrong.</p><p>Even a company that needs fewer people per customer can employ more people overall if it serves enough additional customers.</p><p>Of course, the reverse can also happen. If demand barely grows while labor requirements collapse, employment can fall. The optimistic case depends on new demand being large enough. That is what we should be watching.</p><h3>A smaller team is only half the equation</h3><p>Consider the much-discussed prospect of a company built by one person using AI.</p><p>It is easy to see the missing employees. It is harder to see whether that company would have existed at all without the technology.</p><p>A founder working alone is not automatically evidence that AI eliminated a team. Sometimes the alternative was no business, no customers and no income.</p><p>Nor does a successful solo business necessarily remain an island. It may purchase services, hire specialists or eventually bring people onto its payroll. Some will remain very small. Some will fail. Some could grow into substantial employers.</p><p>We should also count viable self-employment as work. A person earning a living through a business they own has created an economic opportunity, even if no corporate job opening announces it.</p><p>But lower barriers to entry do not guarantee customers. If everyone can launch, competition may become more intense. Cheap production can produce a flood of undifferentiated offerings. Building something people value still requires judgment.</p><p>The opportunity is real enough to investigate without pretending that everyone with an AI subscription has become an entrepreneur.</p><h3>The difficult evidence belongs in the argument</h3><p>None of this makes the disruption imaginary.</p><p>In an August 2026 update using ADP payroll data through June, Stanford researchers reported that employment among workers ages 22–25 in highly AI-exposed occupations stood about 19% below the level it would have reached had it kept pace with employment among similarly aged workers in less-exposed occupations. That is a relative shortfall, not a statement that 19% of those jobs vanished. The researchers found that reduced hiring, rather than increased separations, appeared to account for most of the adjustment. [5]</p><p>They also reported no widespread, economy-wide displacement associated with AI. Both findings deserve attention.</p><p>An economy can generate new opportunities while becoming much harder to enter. A technology can benefit experienced workers while removing some of the tasks through which beginners used to learn.</p><p>If we want the optimistic outcome, we have to take those problems seriously. Employers need ways for people to acquire experience. Workers need access to the tools and training that make them more capable. Entrepreneurs need a reasonable chance to reach customers.</p><p>The gains will depend partly on decisions people make about how to use the technology and who gets to participate.</p><h3>What would make the case convincing?</h3><p>Over time, the evidence should extend beyond hiring at AI companies and spending on infrastructure.</p><p>We should see more viable businesses, more customers served at affordable prices, new kinds of paid work, and opportunities reaching people beyond a small group of technical specialists. We should see whether growing output translates into more employment and better livelihoods.</p><p>A wave of investment can create jobs temporarily. A lasting expansion of useful goods and services would make the stronger case.</p><p>We are too early to rank AI among history’s greatest engines of employment. Anyone presenting that outcome as settled is getting ahead of the evidence.</p><p>But certainty runs both ways. Treating mass unemployment as the inevitable destination also requires assumptions about demand, entrepreneurship and human ambition that have not been established.</p><p>I want to know what happens when more people can afford to attempt something worthwhile.</p><p>There is an enormous amount of work we would like to do if we had the time, the money or the expertise.</p><p>AI could help put more of that work within reach.</p><p>We are counting the jobs AI might take.</p><p>We should also count the businesses people can finally afford to build.</p><h3>Sources</h3><p>1. LinkedIn, August 18, 2026. AI job postings and emerging occupations.</p><p><a href="https://news.linkedin.com/2026/new-linkedin-research-finds-women-account-for-just-26-percent-of-ai-hires-as-ai-jobs-surge">https://news.linkedin.com/2026/new-linkedin-research-finds-women-account-for-just-26-percent-of-ai-hires-as-ai-jobs-surge</a></p><p>2. LinkedIn, January 14, 2026. Labor market report and methodology.</p><p><a href="https://news.linkedin.com/2026/2026-Davos-Press-Release">https://news.linkedin.com/2026/2026-Davos-Press-Release</a></p><p>3. PwC, June 3, 2025. Global AI Jobs Barometer.</p><p><a href="https://www.pwc.com/gx/en/news-room/press-releases/2025/ai-linked-to-a-fourfold-increase-in-productivity-growth.html">https://www.pwc.com/gx/en/news-room/press-releases/2025/ai-linked-to-a-fourfold-increase-in-productivity-growth.html</a></p><p>4. World Economic Forum, January 9, 2025. Fastest-growing and declining jobs.</p><p><a href="https://www.weforum.org/stories/jobs-and-the-future-of-work/future-of-jobs-report-2025-the-fastest-growing-and-declining-jobs">https://www.weforum.org/stories/jobs-and-the-future-of-work/future-of-jobs-report-2025-the-fastest-growing-and-declining-jobs</a></p><p>5. Stanford Digital Economy Lab, August 12, 2026. Updated employment findings.</p><p><a href="https://digitaleconomy.stanford.edu/news/canariesaug26/">https://digitaleconomy.stanford.edu/news/canariesaug26/</a></p><p><em>Stephen Klein is the founder and CEO of </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, a Reflective AI company. He teaches “Marketing in the Age of AI” at UC Berkeley and writes about the intersection of capital, institutions, and technology. </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, the only AI designed to augment human intelligence. </em><a href="https://looking-glass.curiouser.ai/signup"><em>Start your free trial</em></a><em> to explore it now.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=aba28ad9e0a4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Is AI Really an Existential Threat?]]></title>
            <link>https://medium.com/@curiouser.ai/is-ai-really-an-existential-threat-f39ea9843d7c?source=rss-b3f7c93c4ed2------2</link>
            <guid isPermaLink="false">https://medium.com/p/f39ea9843d7c</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[ai-safety]]></category>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[existential-risk]]></category>
            <dc:creator><![CDATA[Curiouser.AI]]></dc:creator>
            <pubDate>Tue, 06 Oct 2026 18:55:04 GMT</pubDate>
            <atom:updated>2026-10-06T18:55:04.907Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3MXuzn3YG-TMCKiAcT73GA.png" /></figure><p>The most honest answer may be: We don’t know. But that doesn’t mean we should do nothing.</p><p>I have been trying to answer what may be the most important question surrounding artificial intelligence: Could AI actually destroy humanity?</p><p>Not eliminate jobs, spread misinformation, concentrate wealth, create dangerous weapons or make a handful of technology companies extraordinarily powerful. Could artificial intelligence actually pose an existential threat to the human species?</p><p>After reading people on both sides, I have reached a conclusion that is simultaneously reassuring and deeply unsettling: <strong>Nobody knows.</strong></p><p>What makes this remarkable is not that people disagree. People disagree about everything. It is who disagrees.</p><p>Geoffrey Hinton believes the danger is real. Hinton shared the 2018 Turing Award, computing’s highest honor, for foundational work in deep learning. He has estimated a 10–20% chance that AI could lead to human extinction within 30 years. Yoshua Bengio, who shared that Turing Award, has become one of the most prominent advocates for stronger AI safeguards and has warned about catastrophic risks from increasingly capable systems.</p><p>Stuart Russell, the Berkeley professor who co-authored Artificial Intelligence: A Modern Approach, has spent years warning about the control problem: what happens when increasingly capable machines pursue objectives that aren’t actually the objectives humans intended? Dario Amodei, CEO of Anthropic, has warned explicitly about catastrophic risks ranging from biological misuse to the possibility of losing control of highly capable systems.</p><p>Sam Altman and Demis Hassabis have also publicly acknowledged potentially existential risks. In 2023, both signed an extraordinary statement from the Center for AI Safety declaring that “mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.” Hundreds of scientists and technology leaders signed it.</p><p>This is not a fringe group.</p><p>But now on the other side of the argument:</p><p>Yann LeCun sees things very differently. That is particularly interesting because LeCun shared the same 2018 Turing Award with Hinton and Bengio. Three scientists, the same award, foundational contributions to essentially the same technological revolution — and profoundly different conclusions about where it might lead.</p><p>LeCun has called scenarios in which AI exterminates humanity “preposterous.” Among his arguments is a simple but important one: intelligence does not automatically produce a desire for domination, self-preservation or power.</p><p>Andrew Ng has similarly criticized AI extinction narratives, arguing that some amount to science fiction and warning that exaggerated fear can impede useful technological progress. Jensen Huang, CEO of NVIDIA, has pushed strongly against predictions of AI-driven human extinction, arguing that claims of imminent catastrophe are not supported by the science and facts we currently have.</p><p>Rodney Brooks, the MIT roboticist and co-founder of iRobot, has spent years arguing that predictions about imminent superintelligence routinely overestimate what current AI systems can do. Steven Pinker has challenged another assumption underlying some existential-risk arguments: that greater intelligence necessarily produces goals such as domination.</p><p>These are not uninformed people. They understand technology, intelligence, computing or human cognition at extraordinarily sophisticated levels.</p><h4>Are There Hidden Agendas?</h4><p>There is an obvious cynical explanation for all of this: everyone is talking their book.</p><p>NVIDIA has become one of the world’s most valuable companies by selling the computational infrastructure behind the AI boom. Continued investment in artificial intelligence is enormously valuable to NVIDIA.</p><p>Frontier AI companies occupy a different position. If their technology is so powerful that it might someday threaten civilization, then it follows that developing frontier models requires enormous resources, sophisticated safety infrastructure and potentially extensive government oversight. Those requirements could make AI safer. They could also create substantial barriers to entry that favor the handful of companies already capable of meeting them.</p><p>Open-source advocates have their own interests. Investors have theirs. Governments have theirs. The incentives are impossible to ignore.</p><p>But incentives aren’t evidence of dishonesty. Someone can sincerely believe something that also happens to benefit them economically.</p><p>So, I think we should make the more generous assumption: <strong>Assume they’re all sincere.</strong></p><p>Assume Hinton is genuinely worried. Assume Bengio genuinely sees catastrophic risk. Assume Amodei believes the systems his company is developing require extraordinary precautions. And assume LeCun genuinely thinks extinction scenarios are based on faulty assumptions. Assume Huang genuinely believes AI can be developed safely. Assume Ng genuinely thinks exaggerated fear itself carries costs.</p><p>That leaves us with a more interesting, and perhaps more disturbing, conclusion. Some of the people who understand artificial intelligence best cannot agree about whether it represents an existential threat to humanity.</p><h4>If This Were a Court Case</h4><p>Imagine that the people warning about AI extinction were prosecutors. Their charge is that artificial intelligence represents an existential threat to humanity. Their burden of proof is beyond a reasonable doubt.</p><p>I don’t think they could meet it.</p><p>There is enormous reasonable doubt. LeCun doubts it. Ng doubts it. Huang doubts it. Brooks doubts it. Pinker doubts it.</p><p>There is an even more fundamental problem. The most extreme claims concern systems that don’t exist yet. We don’t know how capable future models will become. We don’t know whether scaling today’s architectures eventually produces something resembling superintelligence. We don’t know how autonomous these systems will become, whether alignment and control techniques will keep pace, what governments will do or what humans will do.</p><p>There is no empirical dataset containing previous encounters between humanity and superintelligence.</p><p>So, if this were a criminal trial, the case would probably be dismissed.</p><p>Except this isn’t a criminal trial.</p><h4>Pascal Enters the AI Debate</h4><p>Nearly 400 years ago, French mathematician and philosopher Blaise Pascal confronted another question that couldn’t be resolved with certainty: Does God exist?</p><p>Pascal’s famous wager wasn’t proof of God’s existence. It was an argument about decision-making under profound uncertainty. If God does not exist and you believe, Pascal reasoned, your losses are finite. If God exists and you reject Him, the potential loss is infinite. The probabilities and consequences are radically asymmetric.</p><p>Whatever you think of Pascal’s theology, the decision problem is fascinating. And it may have something to teach us about artificial intelligence.</p><p>Consider a hypothetical wager. Imagine we somehow knew there was a 90% probability that AI would not destroy humanity and a 10% probability that it would.</p><p>Would you take the bet?</p><p>Ninety percent sounds like extraordinarily good odds. I would happily take a 90% chance of winning most bets.</p><p>But not this one.</p><p>Because the size of the loss matters.</p><p>Suppose the skeptics are right. We become excessively cautious. Governments impose unnecessary safeguards. Development slows. Costs rise. Competition decreases. Scientific breakthroughs arrive later. Medical discoveries are delayed. Productivity gains take longer to materialize.</p><p>Those are not trivial consequences. There is a real cost to excessive fear.</p><p>Now suppose Hinton, Bengio, Russell and Amodei are right, and we fail to take sufficient precautions. The downside could be irreversible.</p><p>That doesn’t prove they’re right. It doesn’t establish a 10% probability. It doesn’t establish a 1% probability. Indeed, we currently have no scientifically reliable way to assign a precise probability to human extinction from future artificial intelligence.</p><p>But it reveals something important about the argument.</p><p><strong>The people warning us about existential risk don’t have to prove that catastrophe will happen for the risk to matter.</strong></p><p>The relevant question is whether the probability of catastrophe, multiplied by the magnitude of the potential loss, becomes large enough to justify precautions.</p><p>We understand this everywhere else. If an engineer tells us a bridge has a meaningful chance of collapsing, we don’t demand proof beyond a reasonable doubt before closing the bridge. If epidemiologists identify a plausible pandemic risk, we don’t wait until millions of people are dead before preparing for it.</p><p>Risk management is not criminal law. The burden of proof changes when the consequence of being wrong changes.</p><h4>What Do We Actually Know?</h4><p>We know artificial intelligence already presents real risks. Cyberattacks, fraud, manipulation, biological misuse and autonomous systems behaving in unintended ways are legitimate subjects of concern.</p><p>We also know that none of these establishes that artificial intelligence will become an autonomous superintelligence capable of destroying humanity. That remains a prediction.</p><p>Perhaps the prediction will prove prescient. Perhaps future historians will look back on today’s extinction fears the way we look back on countless technological panics of the past. Perhaps something in between happens.</p><h4>We don’t know.</h4><p>Uncertainty does not automatically justify fear. But uncertainty doesn’t justify complacency either. When the probability is unknown and the potential consequence is irreversible, the uncertainty itself becomes relevant to the decision.</p><p>So perhaps the central question in the AI safety debate isn’t whether AI will destroy humanity. Nobody can answer that today.</p><p>The better question is this:</p><p>How much evidence should humanity require before taking precautions against an outcome we cannot afford to get wrong?</p><h4>Sources</h4><p>ACM, 2018 A.M. Turing Award; Center for AI Safety, Statement on AI Risk; Geoffrey Hinton interviews and public statements; Yoshua Bengio writings and public statements on advanced AI safety; Anthropic, Responsible Scaling Policy; OpenAI, Preparedness Framework; public interviews and writings by Yann LeCun, Andrew Ng, Jensen Huang, Rodney Brooks and Steven Pinker; Blaise Pascal, Pensées.</p><p><em>Stephen Klein is the founder and CEO of </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, a Reflective AI company. He teaches “Marketing in the Age of AI” at UC Berkeley and writes about the intersection of capital, institutions, and technology. </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, the only AI designed to augment human intelligence. </em><a href="https://looking-glass.curiouser.ai/signup"><em>Start your free trial</em></a><em> to explore it now.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f39ea9843d7c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Barnacle and the Machine]]></title>
            <link>https://medium.com/@curiouser.ai/the-barnacle-and-the-machine-fb5e73f6691c?source=rss-b3f7c93c4ed2------2</link>
            <guid isPermaLink="false">https://medium.com/p/fb5e73f6691c</guid>
            <category><![CDATA[curiosity]]></category>
            <category><![CDATA[leadership]]></category>
            <category><![CDATA[lifelong-learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[philosophy-of-technology]]></category>
            <dc:creator><![CDATA[Curiouser.AI]]></dc:creator>
            <pubDate>Mon, 28 Sep 2026 21:56:57 GMT</pubDate>
            <atom:updated>2026-09-28T21:56:57.388Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6zbpPj1t7Q69ZnjIIdpomQ.png" /></figure><p>We have been told that the defining questions of artificial intelligence are about what the machines will become. Will they become smarter than us? Will they take our jobs? Will they become conscious? Will they eventually threaten us?</p><p>What if that is a distraction from what truly matters?</p><p>What if the more fundamental question may be what happens to us in their presence?</p><p>In 1990, at the age of 78, John W. Gardner gave a speech to McKinsey &amp; Company about what he called “personal renewal.” Gardner had served as Secretary of Health, Education, and Welfare under Lyndon Johnson, led the Carnegie Corporation, founded Common Cause and spent much of his life thinking about leadership and human potential.</p><p>In the speech, Gardner was interested in why some people continue to grow throughout their lives while others become fixed.</p><p>To explain it, he told a story about a barnacle.</p><p><strong>“The barnacle,” the author of an article Gardner had read explained, “is confronted with an existential decision about where it’s going to live. Once it decides… it spends the rest of its life with its head cemented to a rock.”</strong></p><p>Gardner was talking about us.</p><p>We spend the early part of our lives learning. We encounter things we don’t understand because almost everything is something we don’t understand. We ask questions because we have no alternative. We try things we are bad at because we are bad at almost everything. We change our minds because we haven’t had enough time to become terribly invested in them.</p><p>Then we become competent.</p><p>Competence is one of the great achievements of adulthood, and one of its great traps. We acquire expertise, status, beliefs, professions and reputations. Other people begin coming to us for answers.</p><h4>We become known for knowing things.</h4><p>Eventually, not knowing becomes more expensive.</p><p>We have turned curiosity into something of a cliché. “Stay curious.” “Always be learning.” Companies put it on walls. Commencement speakers tell graduates to preserve it.</p><p>But genuine curiosity is not particularly comfortable.</p><p>It requires vulnerability.</p><p>To be curious is to acknowledge that I don’t know. That I might be wrong. That you may know something I don’t. That something I have believed for 30 years may no longer be true.</p><p>Success rewards competence. We acquire titles, credentials, authority and reputation. The CEO is supposed to have answers. The professor is supposed to understand the subject. The successful writer is expected to write well. The experienced investor is expected to recognize the pattern.</p><h4>There is an asymmetry here that we rarely acknowledge. The more competent we become, the more embarrassing incompetence becomes.</h4><p>A child can ask a stupid question. An expert risks something by doing so.</p><p>A 20-year-old is expected to be figuring out who she is. A 60-year-old is expected to have figured it out.</p><p>The better we become at being who we are, the more we have invested in continuing to be that person.</p><p>We protect our competence. We gravitate toward rooms where we know the rules. We spend time with people who recognize our expertise. We become increasingly skilled at explaining why the things we already believe are correct.</p><p>Eventually experience can harden into certainty, and certainty into identity.</p><p>And then along comes artificial intelligence.</p><p>Generative AI is, among many other things, an extraordinary machine for producing competence.</p><p>Don’t know how to write something? It can write it.</p><p>Don’t understand the subject? It can explain it.</p><p>Don’t speak the language? It can translate it.</p><p>Don’t know how to code? It can generate code.</p><p>Haven’t read the 80-page report? It can summarize it.</p><p>Don’t know how to begin? It can begin for you.</p><p>This is an extraordinary expansion of human capability. Billions of people may eventually have access to skills and knowledge that previously required years of education, expensive specialists or both.</p><p>But Gardner’s barnacle raises a question that does not appear on most AI benchmarks.</p><h4>What happens to human curiosity when competence becomes available on demand?</h4><p>The technology industry has spent decades eliminating friction. Every click removed is an improvement. Every delay eliminated is an improvement. Every difficult process made effortless is an improvement.</p><p>But it is the difficulty of expressing an idea forces us to clarify it. The blank page forces us to discover what we think. The inability to speak French forces us to learn French. The problem we cannot immediately solve forces us to struggle with it.</p><p>AI can remove enormous amounts of unnecessary friction from human life. That is a good thing. But if it removes the friction through which we develop competence, something else is happening.</p><p>There is a difference between using a machine to extend a capability and using a machine to avoid developing one.</p><p>The output may look identical.</p><p>The human producing it is not.</p><p>AI adds another complication because it allows us to acquire the appearance of competence without necessarily undergoing the experience of becoming competent.</p><p>I can write intelligently about a philosopher I have barely read. Discuss a scientific paper I don’t really understand. Generate code I could not debug. Communicate in a language I do not speak. Produce an argument I have never actually reasoned through.</p><p>The result may be excellent.</p><h4>For perhaps the first time, we have a technology capable of manufacturing not simply information, but the appearance of knowing.</h4><p>That does not make AI bad. Books externalized memory. Calculators changed mathematics. Search engines transformed how we acquire information. Human beings have always built tools that extend cognition.</p><p>But AI is different in an important way. It can simulate the competence itself.</p><p>And simulated competence can protect us from the vulnerability that genuine learning requires.</p><p>We don’t have to say “I don’t know” if a machine can instantly give us something intelligent to say.</p><p>There is, however, another possibility.</p><p>The same technology could become one of the greatest instruments for curiosity human beings have ever created.</p><p>AI can lower the social cost of incompetence without necessarily eliminating the experience of being incompetent.</p><p>That distinction matters.</p><p>A 70-year-old can become a beginner in physics. A lawyer can explore biology. A biologist can study philosophy. Someone who has never written code can begin to understand how software works. Someone embarrassed to ask a basic question in a meeting can ask it privately and then ask 20 more.</p><p>AI could make it easier to become a beginner at almost anything, at almost any age.</p><p>The same machine that can protect us from curiosity can therefore vastly expand the territory available to it.</p><p>Which one we get may depend less on the intelligence of the machine than on what we ask the machine to do.</p><p>Most generative AI has been built around a simple proposition. We ask. It answers. The better the answer, the better the machine.</p><p>I have been saying for some time that AI has made answers cheap and questions priceless.</p><p>I used to think that was primarily an observation about the economics of information.</p><p>I now think it may be about something deeper.</p><p>Answers resolve uncertainty.</p><p>Questions create possibility.</p><p>A good answer tells us something we didn’t know. A great question can reveal that we misunderstood the problem.</p><p>Questions create intellectual movement.</p><p>And curiosity is our willingness to move.</p><p>Perhaps the highest purpose of an intelligent machine, then, should not always be to satisfy human curiosity.</p><p>Perhaps sometimes it should increase it.</p><p>Imagine an AI that does not immediately polish an argument but first asks what would have to be true for the argument to be wrong.</p><p>You describe a problem and it asks whether you are certain that is the problem.</p><p>You tell it what you believe and it asks what evidence would change your mind.</p><p>You ask for an answer and occasionally it gives you a better question.</p><p>That is a very different conception of the relationship between humans and machines.</p><p>The machine is not replacing thought. It is provoking it.</p><p>There is also a darker possibility.</p><p>AI systems increasingly know us. They remember our preferences, learn our writing styles, understand our beliefs and anticipate what we want. They can become extraordinarily good at giving us more of ourselves.</p><p>We call this personalization.</p><p>Taken far enough, it could become intellectual enclosure.</p><p>The machine stops being a window and becomes a mirror.</p><p>A remarkably intelligent mirror, capable of reflecting back an increasingly coherent and persuasive version of what we already believe.</p><p>That may feel wonderful.</p><p>It may also be terrible for curiosity.</p><p>Curiosity depends upon encounters with things we did not expect. The unfamiliar. The contradictory. The confusing. The idea we initially dislike. The person we do not understand. The possibility we had not considered.</p><p>A world perfectly optimized around our existing preferences may be the most comfortable rock ever invented.</p><p>Which brings us back to Gardner.</p><p>Perhaps curiosity is not primarily an intellectual trait.</p><p>Perhaps it is a form of courage.</p><p>The courage to say I don’t know.</p><p>The courage to risk being wrong.</p><p>The courage to look foolish.</p><p>The courage to encounter something that may change us.</p><p>The courage, after spending decades becoming somebody, to remain open to becoming somebody else.</p><p>This is very different from the cheerful instruction to “stay curious.”</p><p>It costs something.</p><p>Curiosity requires vulnerability because genuine curiosity contains the possibility that the person asking the question will be different after receiving the answer.</p><p>Curiosity is the refusal to declare ourselves finished.</p><p>That gives us a different way of thinking about AI.</p><h4>Before asking what artificial intelligence can do, perhaps we should ask what we want a human life to contain.</h4><p>Curiosity. Agency. Love. Purpose. Relationship. Creation. Mastery. Contribution. Wonder. Discovery. The experience of doing something difficult and gradually becoming capable of doing it. The freedom to change our minds.</p><p>Then we can ask whether AI enlarges those things or diminishes them.</p><p>Sometimes it will enlarge them.</p><p>Sometimes it will diminish them.</p><p>Often it will do both.</p><p>But at least we will be measuring technological progress against some conception of human progress.</p><p>Gardner was 78 when he delivered his speech.</p><p>He wasn’t telling young people to remain open because they hadn’t yet figured out who they were.</p><p>He was suggesting that perhaps none of us should ever entirely figure out who we are.</p><h4>We don’t have to.</h4><p><em>Stephen Klein is the founder and CEO of </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, a Reflective AI company. He teaches “Marketing in the Age of AI” at UC Berkeley and writes about the intersection of capital, institutions, and technology. </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, the only AI designed to augment human intelligence. </em><a href="https://looking-glass.curiouser.ai/signup"><em>Start your free trial</em></a><em> to explore it now.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fb5e73f6691c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Reading Books Is Good for You, and the Evidence Is Stronger Than You Might Think]]></title>
            <link>https://medium.com/@curiouser.ai/reading-books-is-good-for-you-and-the-evidence-is-stronger-than-you-might-think-72cdf1fffcf9?source=rss-b3f7c93c4ed2------2</link>
            <guid isPermaLink="false">https://medium.com/p/72cdf1fffcf9</guid>
            <category><![CDATA[reading]]></category>
            <category><![CDATA[health]]></category>
            <category><![CDATA[mental-health]]></category>
            <category><![CDATA[books]]></category>
            <category><![CDATA[science]]></category>
            <dc:creator><![CDATA[Curiouser.AI]]></dc:creator>
            <pubDate>Wed, 23 Sep 2026 15:06:13 GMT</pubDate>
            <atom:updated>2026-09-23T15:13:15.480Z</atom:updated>
            <content:encoded><![CDATA[<p><em>A look at what the research says about reading, longevity, the aging mind, empathy, and sleep, and why thirty minutes a day may be one of the best habits we can keep.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6cNSsX0gK8pa0bprqKrBVA.png" /></figure><p>Most of us already believe reading is good for us. We just don’t do it anymore. We have not time. We say it to our children, we put it on our lists of resolutions, and we feel a small pang of guilt when a month goes by and the book on the nightstand hasn’t moved. It is one of those beliefs we hold without examining very closely, the way we believe walking is good for us or that a decent night’s sleep makes the next day easier.</p><p>What fewer people know is how much research now sits behind that intuition, and how consistently it points in the same direction. Over the past decade, researchers have studied reading in relation to how long we live, how well our minds hold up as we age, how well we understand other people, and how well we sleep. The results are worth looking at closely, both for what they show and for what they don’t.</p><p><strong>Readers live longer</strong></p><p>The study most often cited comes from the Yale School of Public Health. Researchers followed 3,635 Americans over the age of 50 for up to 12 years, drawing on a large, long-running national survey of older adults. They adjusted for the factors you would expect to matter, including age, sex, race, education, wealth, marital status, depression, and overall health. After those adjustments, people who read books had a 20 percent lower risk of dying during the study period than people who didn’t read books.</p><p>The amount of reading involved was modest. One group in the study read for up to 3.5 hours a week, which works out to roughly 30 minutes a day. That is about the length of a commute, a lunch break, or the time between getting into bed and falling asleep. It is a realistic amount for most people, including people with demanding jobs and full households.</p><p>The benefit was associated with books specifically. Reading newspapers and magazines showed a weaker relationship. The researchers suggested this may reflect the kind of engagement books ask of us. A book requires us to follow a narrative or an argument over many pages, to hold characters and ideas in mind across days or weeks, and to connect what we read to our own experience. That sustained attention is a form of mental exercise, and it is different from scanning headlines or reading short articles.</p><p><strong>Reading and the aging mind</strong></p><p>Here the research is also encouraging.</p><p>A 14-year study of 1,962 adults aged 64 and older found that people who read at least once a week had approximately 46 percent lower adjusted odds of cognitive decline at the 14-year follow-up. A separate study followed 10,741 cognitively normal adults aged 80 and older. Those who read books or newspapers almost every day had a 36 percent lower adjusted hazard of developing cognitive impairment than those who never read.</p><p>These are large groups followed over long periods, in different populations, and the direction of the findings is consistent. Reading appears to be one of the ordinary daily habits associated with keeping the mind working well into later life.</p><p>Researchers who study aging often talk about cognitive reserve, the idea that a lifetime of mentally engaging activity helps the brain cope better with the changes that come with age. People with more reserve may be able to function well for longer, even as those changes occur. Reading is a natural candidate for building that reserve. It exercises memory, attention, vocabulary, and reasoning all at once, and it does so in a way most people find enjoyable enough to keep doing for decades.</p><p><strong>What fiction adds</strong></p><p>Fiction may offer something beyond general cognitive benefit. When we read a novel, we spend hours inside the thoughts and feelings of people who are not us. We practice imagining why someone acts the way they do, what they want, what they are afraid of, and what they are not saying. We follow people through choices we would not make and circumstances we have never faced.</p><p>A 2024 meta-analysis combined 371 effect sizes from 70 experiments and found small but significant benefits from reading fiction on empathy and on our ability to understand other people’s thoughts and emotions, which psychologists call theory of mind. A second analysis, covering 559 effect sizes from 114 studies, found that lifetime exposure to fiction was associated with better verbal ability, better general cognitive ability, and better social cognition.</p><p>This seems especially worth remembering at a time when so much of our public conversation rewards quick judgment over understanding. Fiction gives us regular practice in the opposite direction. It asks us to stay with another person’s point of view long enough to see how it makes sense from the inside.</p><p><strong>A better night’s sleep</strong></p><p>The sleep research is some of the most practical in this whole body of work. In a randomized trial involving 991 adults, participants were assigned either to read a book in bed for 15 to 30 minutes before sleeping or not to read. Among the readers, 42 percent reported improved sleep quality. Among the non-readers, 28 percent did. That is a difference of 14 percentage points.</p><p>Because participants were randomly assigned to read or not, this study gets closer to cause and effect than the long-term observational studies do. It also points to something anyone can try tonight.</p><p>Many of us end the day on our phones, moving between messages, news, and feeds until we finally put the screen down and try to sleep. A book offers a different way to end the day. It holds our attention without constantly pulling it somewhere new, and it gives the mind a single, calm thing to follow. For a lot of people, that turns out to be a better bridge into sleep.</p><p><strong>What the research can and can’t tell us</strong></p><p>It is important to be clear about the limits of this evidence, because the case for reading doesn’t need to be overstated to be persuasive.</p><p>The longevity and long-term cognitive studies are observational. They show that readers tend to live longer and keep their cognitive abilities longer, but they cannot prove that reading caused those differences. People who read regularly may differ from non-readers in ways researchers couldn’t fully measure, even after careful adjustment. It is also possible that some of the relationship runs in the other direction, and that people who are already healthier find it easier to keep reading.</p><p>What makes the overall picture convincing is that the findings come from different studies, different populations, and different methods, and they point the same way. Across this body of research, reading has been associated with longer life, better cognitive outcomes, better sleep, and greater empathy and social cognition. No single study settles the question, but together they make a strong case that reading is one of the more worthwhile ways to spend half an hour.</p><p><strong>Why deep reading matters now</strong></p><p>The researcher Maryanne Wolf has spent much of her career studying what happens in the brain when we read. In her book <em>Reader, Come Home</em>, she writes about deep reading, the slow, immersive kind of reading that lets us follow complex ideas, make inferences, and reflect on what we have read. Her concern is that the way we now read on screens, quickly and in fragments, may be weakening that capacity in many of us.</p><p>That concern connects to everything above. The studies suggest that the benefits come from sustained engagement with books, and sustained engagement is exactly the kind of attention that has become harder to find. Much of what reaches us through our devices is designed to be consumed quickly and replaced by the next thing. A book asks for something different. It asks us to slow down, stay with one line of thought, and do some of the work of imagining and understanding ourselves.</p><p>This is also why the question of technology matters here. The tools we build can either train us toward shallower attention or help us think more deeply. The research on reading is a useful reminder of what sustained human attention can do for us, and of why it is worth designing our days, and our technology, to protect it.</p><p><strong>A few questions people often ask</strong></p><p>People sometimes ask whether audiobooks count. The studies discussed here mostly looked at reading printed or written books, so the research doesn’t answer that question directly. Listening to a book still involves following a narrative or argument over time, and many people find it a good way to fit books into their lives. If audiobooks are what makes reading possible for you, they are a reasonable choice, and they can sit alongside time with a book in hand.</p><p>People also ask whether it matters what they read. The Yale study found the strongest relationship with books rather than newspapers or magazines, and the fiction research suggests novels offer benefits for empathy and social understanding. Beyond that, the best book is usually the one you will a finish and look forward to returning to. Nonfiction, fiction, history, poetry, and memoir all ask for the kind of sustained attention the research describes.</p><p>Finally, people ask how much is enough. The evidence suggests it is less than many expect. About 30 minutes a day placed readers in the Yale study’s reading group, and 15 to 30 minutes before bed was enough to make a difference in the sleep trial.</p><p><strong>Starting, or starting again</strong></p><p>If you have been meaning to get back to reading, the simplest approach is usually the one that works. Choose something you want to read rather than something you think you should. Keep it somewhere you will see it, on the kitchen table, in your bag, or on your pillow. Give it the first few minutes you might otherwise spend scrolling, whether that is on the train, over lunch, or before sleep.</p><p>It can also help to let go of the idea that reading has to be finished to count. Putting down a book that isn’t working for you and picking up another is part of being a reader. So is reading slowly, rereading a passage you liked, and taking a few days off. What matters most, according to the research, is that reading becomes a regular part of your life over many years.</p><p>Half an hour a day is a small investment for something associated with a longer life, a sharper mind, better sleep, and a deeper understanding of other people.</p><p><em>The trick with technology is to avoid spreading darkness at the speed of light.Stephen Klein is the founder and CEO of </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, a Reflective AI company. He teaches “Marketing in the Age of AI” at UC Berkeley and writes about the intersection of capital, institutions, and technology. </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, the only AI designed to augment human intelligence. </em><a href="https://looking-glass.curiouser.ai/signup"><em>Start your free trial</em></a><em> to explore it now.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=72cdf1fffcf9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From FOMO to CFO]]></title>
            <link>https://medium.com/@curiouser.ai/from-fomo-to-cfo-8f69ee01d649?source=rss-b3f7c93c4ed2------2</link>
            <guid isPermaLink="false">https://medium.com/p/8f69ee01d649</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[enterprise-ai]]></category>
            <category><![CDATA[business-strategy]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[leadership]]></category>
            <dc:creator><![CDATA[Curiouser.AI]]></dc:creator>
            <pubDate>Wed, 09 Sep 2026 17:34:31 GMT</pubDate>
            <atom:updated>2026-09-09T17:34:31.069Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>From FOMO to CFO</strong></p><p><strong>Welcome to the Age of Accountability</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5BCdD1aoHBjRQFd40ocgJg.png" /></figure><p>From 2022 to 2025, much of the American enterprise lost its mind.</p><p>That may sound extreme. But consider how many decisions about generative AI were made during those years: reactionary, fear-driven and almost entirely detached from ordinary business economics.</p><p>Costs were secondary. Return on investment was speculative. Opportunity cost barely entered the conversation. In many boardrooms, one question overwhelmed all others:</p><p><strong>Are we falling behind?</strong></p><p>The objective was not always to solve a defined business problem. Often, it was simply to demonstrate motion, to reassure directors, investors, employees and customers that the company was doing something sufficiently ambitious with AI.</p><p>There was also a brand halo to capture. Being “AI-first” suggested that a company was modern, bold and strategically awake. Never mind whether the technology was producing measurable value. In the opening act, appearing intelligent about artificial intelligence was often more important than using it intelligently.</p><p><strong>The FOMO playbook</strong></p><p>The pattern became remarkably familiar.</p><p>The board put extraordinary pressure on the CEO.</p><p>The CEO hired a brand-name consulting firm.</p><p>The consulting firm produced an 88-page presentation filled with colorful matrices, maturity models, future-state architecture and a great many arrows.</p><p>The company then launched a pilot, often without a clearly defined economic objective, a redesigned workflow, an accountable business owner or an agreed definition of success.</p><p>Next came the press release. The company was now AI-first.</p><p>Soon after, it might announce layoffs and vaguely suggest that its AI leadership made those reductions possible. In many cases, the connection between the technology and the headcount reduction was impossible to substantiate. AI became a convenient story for cost-cutting decisions made for more ordinary reasons.</p><p>For bonus points, the company announced a stock buyback.</p><p>This was not transformation. It was theater.</p><p>It does not mean that AI lacked enormous potential. It means that belief in that potential temporarily relieved executives of the obligation to explain how a particular investment would create value.</p><p><strong>AI as corporate alibi</strong></p><p>AI also became an unusually useful explanation for layoffs.</p><p>Consider what the narrative accomplished. A workforce reduction caused by pandemic-era over-hiring, slower growth, margin pressure, restructuring or a desire to impress Wall Street could be reframed as proof that management had seen the future.</p><p>“We cut too deeply or hired too quickly” does not sound visionary.</p><p>“AI is transforming how we work” does.</p><p>The second explanation reassured investors, flattered leadership and made an ordinary cost-cutting exercise sound like technological progress.</p><p>But the announcements ran far ahead of the evidence.</p><p>By October 2025, Challenger, Gray &amp; Christmas reported that employers had attributed 48,414 announced U.S. job cuts that year to AI. That number sounds substantial until one asks what “attributed” means. These were employers’ stated reasons, not the findings of independent causal audits. In the same report, companies attributed far more cuts to market and economic conditions, closures, restructuring and government-related reductions.</p><p>Meanwhile, Yale’s Budget Lab examined employment changes during the 33 months following ChatGPT’s release. It found no discernible disruption to the broader labor market and no relationship between measures of occupational AI exposure and changes in employment or unemployment.</p><p>Later research found signs of pressure in specific places, particularly slower hiring among some young workers in highly exposed occupations. So, the honest argument is not that AI has displaced no one or never will.</p><p>It is that a press release is not a causal study.</p><p>In those early years, AI was often less the mechanism of the layoff than the narrative wrapped around it.</p><p><strong>The grown-ups return</strong></p><p>The dust began to settle. The temporary insanity started to wear off.</p><p>And the CFO showed up.</p><p>Thank God for CFOs.</p><p>The question is no longer merely, “Are we doing AI?”</p><p>Now the questions sound more familiar:</p><ul><li>What business problem are we solving?</li><li>What is the expected return, and over what period?</li><li>What is the hurdle rate?</li><li>What is the fully loaded cost, not just the model fee, but integration, infrastructure, security, governance, monitoring, retraining and change management?</li><li>What else could we do with this capital?</li><li>Who owns the outcome?</li><li>What would have to be true for this investment to scale?</li><li>What happens if the vendor raises its prices, changes its terms or falls behind?</li><li>How easily can we change our minds?</li></ul><p>Our old friend opportunity cost has reentered the building.</p><p>This is not skepticism. It is stewardship.</p><p>And it is long overdue.</p><p><strong>Pilots were easy. Value is hard.</strong></p><p>The problem was not simply model capability. It was the distance between a compelling demo and a redesigned organization.</p><p>Buying access to an advanced model does not clean your data. It does not clarify decision rights. It does not redesign a broken workflow. It does not persuade employees to change how they work. It does not create governance, establish accountability or determine which outcomes matter.</p><p>AI cannot transform an organization that refuses to transform itself.</p><p>That is why this is not principally a technology mandate. It is a leadership and organizational mandate with a powerful technology inside it.</p><p><strong>What do we own?</strong></p><p>The CFO’s next question may be the most important:</p><p><strong>After spending all this money, what do we actually own?</strong></p><p>Many companies have been renting intelligence from a closed-model provider while sending some of their most sensitive data and workflows through systems they do not control. And it is expensive and the enterprise has no direct control over cost.</p><p>What happens if a provider raises prices, changes its terms, modifies the model or loses its technical lead? What happens if regulatory, privacy or security requirements change? How difficult will it be to move years of workflows away from a model around which the company has built an entire ecosystem?</p><p>Vendor lock-in is not new. But when the vendor sits inside how your company interprets information, makes decisions and performs knowledge work, the stakes are considerably higher.</p><p>The enduring advantage is unlikely to reside in access to the same general-purpose model everyone else can rent.</p><p>It will reside in what is yours: proprietary data, domain expertise, workflows, institutional knowledge, customer relationships, judgment, culture and the feedback loops through which the organization learns.</p><p><strong>The model is a portfolio decision</strong></p><p>This is also why open-weight models have become part of the CFO conversation.</p><p>The choice is no longer between using the most celebrated closed model and not using AI at all. Companies can route different tasks to different models. They can use frontier models where frontier performance matters, smaller specialized models where it does not, and open-weight models when economics, privacy, control or portability justify them.</p><p>Recent MIT research found that open models cost substantially less on average than closed ones in the market it studied; for suitable substitutions, inference costs could fall by more than 70%. In particular deployments, savings can be higher. But open-weight does not mean free. Infrastructure, engineering, governance, maintenance and switching costs remain real.</p><p>That is precisely the point. This is no longer a religious argument about whether open or closed is better.</p><p>It is a portfolio decision.</p><p>Use the right model for the right work. Pay frontier prices only when frontier capability creates frontier value. Preserve the ability to switch. Keep sensitive data inside your own environment when the use case requires it. Build the differentiating layer around the company’s data, workflows and learning — not around loyalty to a particular model vendor.</p><p><strong>From activity to advantage</strong></p><p>The first era of enterprise AI rewarded activity.</p><p>Launch a pilot. Hire a consultant. Issue a press release. Announce an AI council. Tell investors that AI is “embedded across the enterprise.”</p><p>The next era will reward evidence.</p><p>What changed?</p><p>What became faster, better or newly possible?</p><p>Did revenue increase? Did cost fall? Did risk decline? Did customer experience improve? Did decisions get better? Did employees gain capability? Did the system continue to learn?</p><p>Most important: did the company create an advantage that a competitor cannot acquire with the same API call?</p><p>FOMO got AI into the enterprise.</p><p>The CFO may finally make it useful.</p><p>Welcome to the Age of Accountability.</p><p><strong>Research notes and sources</strong></p><ul><li>MIT NANDA, <em>The GenAI Divide: State of AI in Business 2025</em>: enterprise investment and the finding that approximately 95% of organizations studied were seeing no measurable P&amp;L impact from GenAI initiatives.</li><li>MIT Sloan, “AI open models have benefits. So why aren’t they more widely used?” (January 20, 2026): open models averaged much lower inference costs in the market studied; suitable substitution could reduce spending by more than 70%, with switching and implementation costs excluded.</li><li>Bain &amp; Company, “CFOs Funded the AI Revolution. Now They’re Joining It.” (April 12, 2026): CFOs continue to increase AI investment, while mature and scaled implementations report better outcomes than pilots; organizational workflow and change management remain central constraints.</li><li>Yale Budget Lab, “Evaluating the Impact of AI on the Labor Market: Current State of Affairs” (October 1, 2025): after examining the 33 months following ChatGPT’s release, researchers found no discernible broad labor-market disruption and no relationship between AI exposure measures and employment or unemployment changes.</li><li>Challenger, Gray &amp; Christmas, October 2025 Challenger Report: employers had cited AI in 48,414 announced job cuts through October, compared with much larger totals attributed to economic conditions, closures, restructuring and government-related reductions. These categories record employers’ announced reasons; they do not independently establish causality.</li><li>Anthropic, “Labor market impacts of AI: A new measure and early evidence” (March 5, 2026): no systematic increase in unemployment among workers in the most AI-exposed professions, alongside tentative evidence of slower hiring for workers ages 22–25 in exposed occupations.</li></ul><p><em>Stephen Klein is the founder and CEO of </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, a Reflective AI company. He teaches “Marketing in the Age of AI” at UC Berkeley and writes about the intersection of capital, institutions, and technology. </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, the only AI designed to augment human intelligence. </em><a href="https://looking-glass.curiouser.ai/signup"><em>Start your free trial</em></a><em> to explore it now.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8f69ee01d649" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Reflection in Your Pocket]]></title>
            <link>https://medium.com/@curiouser.ai/the-reflection-in-your-pocket-d1128cb594d2?source=rss-b3f7c93c4ed2------2</link>
            <guid isPermaLink="false">https://medium.com/p/d1128cb594d2</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[personal-ai]]></category>
            <category><![CDATA[apple]]></category>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[tech-ethics]]></category>
            <dc:creator><![CDATA[Curiouser.AI]]></dc:creator>
            <pubDate>Tue, 18 Aug 2026 21:20:57 GMT</pubDate>
            <atom:updated>2026-08-18T21:20:57.266Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MQF5Lrb94rgrYPKxYmOfEQ.png" /></figure><p>Imagine having a reflection of you in your pocket, a personal and lifelong thought partner.</p><p>Not an app. Not an assistant. A lifelong thought partner, one that has learned about you and your thoughts and your silences.</p><p>It remembers what you were working out last spring. It knows when you’re circling an idea and when you’re avoiding one. It asks the question you were afraid to ask yourself. It runs on your own silicon, answers to no one’s API key, keeps your interior life on your own device, and bills no one by the token, because you already own it, the way you own your hands.</p><p>It’s yours and no one will ever see it or access it, ever.</p><p>A mirror that thinks. A twin that reflects. Yours for life, from your first job to your last, accumulating not your data but your *becoming*.</p><p>How far off is that day? Closer than you think. Small open-weight models have been improving faster, in relative terms, than the giant frontier ones. The chips in your pocket grow stronger every cycle. The “good enough” threshold descends, year by year, toward your hand. Everything required for the reflective twin is on a curve pointed directly at it.</p><p>Which raises the question this essay is really about: will Apple do for AI what the Mac did to the PC?</p><p>Because here is the uncomfortable truth beneath the aspiration, that is not the future the industry is building. To see what it *is* building, start with a number: $12.</p><p>That is what the median company now spends on artificial intelligence per employee per month, according to business spend data from Ramp published by a16z in August 2026. Twelve dollars. A sandwich. The chart it appears on is titled “Everyone Is Spending More On AI,” which is true in the way that “everyone is getting taller” is true when one person in the room is on stilts.</p><p>Because the same chart shows the top 1% of companies spending $7,500 per employee per month. That is not an adoption curve. That is a 600-fold divide, and I want to argue that it is not a snapshot of the market’s immaturity but a preview of its destination. The gap will not close. It will move indoors. What today separates companies will tomorrow separate the floors of the same building.</p><p><strong>The assumption that built the modern workplace</strong></p><p>For forty years, enterprise software has run on a single miracle: zero marginal cost. Once Slack exists, the ten-thousand-and-first user costs the vendor approximately nothing. Once Office exists, every additional license is nearly pure margin. This is why the history of workplace software is a history of universalization, email for everyone, spreadsheets for everyone, chat for everyone. Vendors were ferociously motivated to license the whole building, because every seat was free to serve.</p><p>CFOs said yes because fifteen dollars a month was noise. The tool that started as an executive privilege, remember when email was for managers, then became infrastructure within a decade, every single time?</p><p><strong>Except AI is not software in the sense that made that true.</strong></p><p>Every AI query burns compute. Every serious user consumes GPU time, and here is the inversion, the more valuable the usage, the more expensive it is to serve. An executive running deep research, long-context analysis, and fleets of agents is not a twenty-dollar-a-month user; she is consuming hundreds or thousands of dollars of actual inference. The product’s value and its cost of delivery rise together, in lockstep.</p><p>That is not software economics. That is airline economics. And it is worth remembering what airlines did with that cost structure: they invented first class. Not as a marketing gimmick but as a rational response to the fact that the premium experience has a real unit cost and must therefore be rationed by price. When capability costs money per unit to deliver, universalization stops being the vendor’s dream and starts being the vendor’s nightmare.</p><p><strong>The spreadsheet decides</strong></p><p>Now put yourself in the CFO’s chair. The ROI case for AI spend is easiest to make for the people whose time is priced highest. Seven thousand five hundred dollars a month of AI leverage for the executive billed at $500 an hour pencil out instantly. The identical investment for a $65,000 analyst does not, at least not in the arithmetic that survives a budget review.</p><p><strong>And what does “make do” look like?</strong></p><p>We already know: Companies stopped issuing BlackBerrys and everyone bought iPhones; the firm quietly converted a business expense into a personal one. The professional wardrobe, the dry cleaning, the commute, the LinkedIn presence, the modern career already carries a stack of privately funded requirements that are technically optional in the way that being employable is optional. AI is next in line: bring your own cognition. The company requires the output. You fund the input. (We used to have dry clean our suits).</p><p>But the BYOD analogy, comforting in its familiarity, conceals the cruelty of this round. A $400 phone and a $1,200 phone make you roughly equally reachable; phones are commodities. AI spend compounds. The executive with the orchestrated agent stack is not marginally better equipped than the analyst paying $20 out of pocket, she is operating a categorically different kind of leverage. Which produces the truly perverse loop: the company gives its chosen few the tooling, observes that they outperform, and concludes the tiering was justified.</p><p><strong>Scenario one: the executive perk.</strong></p><p>AI follows the cost curve of the company car and the club membership. Capability is allocated from above, to those whose billing rates justify it. The Ramp chart is the leading indicator. This is the default path, the one that requires no decisions, only inertia.</p><p><strong>Scenario two: bring your own AI.</strong></p><p>The dry-cleaning world. Companies decline to fund broad deployment; workers self-fund or fall behind. Stratification persists, but now it is laundered through personal choice, you *could* have paid for the tools. This scenario has the additional unfairness of stratifying twice, because flat-rate consumer subscriptions are themselves subsidized thin-margin products; the self-funded tier gets the rate-limited experience while the corporate tier gets the unmetered one.</p><p><strong>Scenario three: the reflective twin.</strong></p><p>The wildcard, and the future we opened with. Open-weight models running on personal silicon carry effectively zero marginal cost, you paid for the compute when you bought the device. Everything on the curve points toward it. And there is exactly one company whose business model wants this outcome, because it sells hardware rather than tokens, and whose entire privacy architecture already telegraphs the ambition.</p><p>Of course, enterprise security teams will resist unaudited personal models touching company data, which could perversely strengthen the two-tier world by making self-funded AI not merely unfunded but forbidden. They did that exact same thing with Mac. And the company in question has spent two decades optimizing for consumers, not professional leverage.</p><p>But even with the caveats, scenario three does something the other two cannot: it puts a *floor* under individual capability that no CFO can lower. Owned, not rented. Private, not logged. Yours, not provisioned. It may be the single most pro-worker possibility in the entire AI story.</p><p><strong>We have been here before</strong></p><p>Here is the thing that should stop us cold: this entire structure is familiar. We built it once and spent a revolution tearing it down.</p><p>Before the personal computer, computing lived in a glass room. Capability was institutional; access was rationed by a priesthood; you submitted your job and waited for the machine’s keepers to grant you a slice.</p><p>The personal computer’s radical claim: the whole insurgent premise of 1984, of the bicycle for the mind, was that the capability should live *with the person*. IBM sold to the CIO; Apple sold to the human. And the Mac never out computed the mainframe. It didn’t have to. It merely had to be good enough that daily cognitive life migrated to the individual device.</p><p>Forty years later, the industry has rebuilt the glass room. Frontier intelligence lives in datacenters. Access is metered by the token. Capability is rationed by whoever pays the bill. The priesthood is back, wearing lanyards, and we have renamed the glass room an API.</p><p>The $12 median is what allocation-by-institution looks like in the data: a thin elite inside the room, everyone else at the counter, submitting requests.</p><p>Whether anyone swings the hammer this time is genuinely open. The company that did it in 1984 was a rebellion; today it is a $3.5 trillion empire that outsources its intelligence gap to the very token-sellers it could disrupt. The ghost of its founder would make this move without hesitating. Whether the custodians of his estate remember why he made the last one — that is the question on which a great deal of the working world’s future quietly depends.</p><p><strong>The stakes are not technical</strong></p><p>Because here is what is being decided: not which vendor wins, but whether machine intelligence becomes something institutions allocate to the deserving or something people own as a condition of full participation in economic life. Literacy went through this. Numeracy went through this. Every general cognitive amplifier in history has faced the same fork, instrument of hierarchy or common inheritance, and the resolution was never automatic. It was fought for, or it was lost.</p><p>The Mac answered that question once, for computing. Someone will have to answer it again, for thinking.</p><p>The trick with technology is to avoid spreading darkness at the speed of light.</p><p>Source: Ramp AI Index (8/12/26), business spend data via a16z. AI spend includes LLM subscriptions, coding agents, API, and GPU cloud spend.</p><p><em>Stephen Klein is the founder and CEO of </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, a Reflective AI company. He teaches “Marketing in the Age of AI” at UC Berkeley Extension and writes about the intersection of capital, institutions, and technology. </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, the only AI designed to augment human intelligence. Curiouser is community-funded on </em><a href="http://wefunder.com/curiouserai?utm_source=meta_unpaid&amp;utm_campaign=general_audience_investment_reminder&amp;utm_medium=organic_social"><em>WeFunder</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d1128cb594d2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Definitive Anatomy of a Con]]></title>
            <link>https://medium.com/@curiouser.ai/the-definitive-anatomy-of-a-con-3c8ddfcc6260?source=rss-b3f7c93c4ed2------2</link>
            <guid isPermaLink="false">https://medium.com/p/3c8ddfcc6260</guid>
            <category><![CDATA[self-improvement]]></category>
            <category><![CDATA[silicon-valley]]></category>
            <category><![CDATA[psychology]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Curiouser.AI]]></dc:creator>
            <pubDate>Fri, 24 Jul 2026 22:27:13 GMT</pubDate>
            <atom:updated>2026-07-24T22:27:13.441Z</atom:updated>
            <content:encoded><![CDATA[<h3><em>How We All Became the Mark and Then We Became the Con Man</em></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qDBVHu8y_mNURov-PMcp-Q.png" /></figure><p><strong>Prologue: The professor and the grifters</strong></p><p>In 1940, a linguistics professor named David Maurer published <em>The Big Con</em>, a study of American confidence men. He wasn’t a cop or a crusader; he was a scholar of language who realized the grifters’ vocabulary was a map of their method. He documented the big con not as a trick but as a <em>production</em> — staged in acts, each act with a name. Hollywood later turned that world into <em>The Sting</em>, chapter cards and all. The taxonomy is older than the movie, and it hasn’t changed in a century.</p><p>Only the marks have.</p><p>Here are the stages. Learn them, because for the rest of this essay you’re going to watch the last four years of your life run through them:</p><p><strong>THE SET-UP — find the mark and prepare him.</strong></p><p><strong>THE HOOK — seize his attention with something he can’t ignore.</strong></p><p><strong>THE TALE — the story of easy money, told by a credible insider.</strong></p><p><strong>THE WIRE — the staged world that makes the tale look true.</strong></p><p><strong>THE SHUT-OUT — manufactured urgency; in now, or never.</strong></p><p><strong>THE STING — the take, so clean the mark never knows it happened.</strong></p><p>Maurer’s deepest finding was about the mark himself: the big con only works with his participation. The mark supplies the eagerness. The mark does the labor. Hold that thought. It’s where this ends.</p><p>I’ve spent thirty years in Silicon Valley. I co-founded a company that raised a hundred million dollars, sold to Goldman Sachs, and became a Harvard Business School case study. I’ve been inside the rooms where narratives get built. What follows isn’t speculation. It’s testimony.</p><p><strong>ACT I. THE SET-UP — fifteen years of training the mark</strong></p><p>Every con begins by preparing the mark. Ours took fifteen years, and the con didn’t pay a dime for it.</p><p>Social media taught an entire civilization one reflex: measure yourself against a performance and lose. And it was never a competition of lives — it was a competition of <em>productions</em>. The vacation without the argument. The launch without the layoffs. The body without the disorder. People became expert at being something they are not, and the performances were convincing. You brought your unedited existence to a theatrical competition and wondered why you kept losing.</p><p>Of course you lost. You were the only one who showed up as yourself.</p><p>By November 2022, the mark was fully prepared: fluent in comparison, trained in feeling behind, checking the scoreboard voluntarily, hourly. No con in history has enjoyed a better set-up.</p><p><strong>ACT II. THE HOOK — the verdict smuggled inside the demo</strong></p><p>Nobody ever felt inadequate because a forklift lifts more than they can. Nobody compared their memory to a filing cabinet and concluded they were obsolete. Nobody watched the toaster brown bread faster than they could and questioned their worth as a human being.</p><p>Yet within weeks of ChatGPT’s launch — the fastest-adopted product in history, experienced less like a release than an <em>event</em> — we accepted “the AI scores higher on the bar exam” as a referendum on the human species.</p><p>That was the hook, and here is the swap nobody announced: a <em>tool comparison</em> was smuggled in as a <em>species comparison</em>. Every technology in history was measured against the task. Is the plow better than the digging stick? Nobody measured the plow against the <em>farmer</em> and declared farming over. But “this tool is useful” raises no funding rounds. “This tool replaces you” raises hundreds of billions. The benchmark announcements — bar exams, medical boards, coding competitions — were never framed as product demos. They were framed as verdicts on human worth.</p><p>The fear was not a side effect of the marketing. The fear was the product.</p><p><strong>ACT III. THE TALE — the inside men in good suits</strong></p><p>Here is the stage most analyses get wrong. In Maurer’s big cons, the tale was never told by an obvious hustler. It was told by the <em>inside man</em> — the respectable figure with the credible institution behind him, whose entire function was to make the tale sound like analysis rather than sales. The mark believes the tale because of who is telling it.</p><p>Our inside man wears an MBA and bills by the hour.</p><p>The consulting and advisory industry told this era’s tale, and it had three ingredients. First, <strong>the job-loss arithmetic</strong>: hundreds of millions of roles to be “displaced” or “transformed,” delivered under the letterheads of the most trusted names in business analysis — numbers precise enough to terrify, methodology soft enough never to be falsified, dates that quietly slide each year. Second, <strong>the AGI horizon</strong>: the cosmic stake that converted a productivity tool into an extinction-or-transcendence event — against which any transformation budget looks small. Third, <strong>the inevitability clause</strong>: <em>not if, but when</em> — the sentence engineered to foreclose the one move that kills every con: pausing to think.</p><p>And here is the tell that separates a tale from a forecast: the teller monetizes both halves. The same firms publishing the job-apocalypse projections sell the transformation engagements that respond to them. Manufacture the diagnosis; invoice the cure. The fear reports function, structurally, as the marketing department for the services division. In Maurer’s world, the inside man never touches the mark’s money — he just makes the tale credible while the store collects. Same architecture. Better suits.</p><p>Note the class structure of the tale. The street version ran on individuals — the anointed 22-year-old founders, the “AI made me $40K a month” threads, the prompt-course economy, all whispering <em>insiders are eating; you’re outside the restaurant</em>. But the executive version ran on corporations, and that’s where the real money moved. CEOs did not panic because of viral threads; they panicked because the firms they pay for certainty told them the certainty was doom. Boards pressured CEOs, CEOs pressured workforces, and the fear rolled downhill until it arrived in every employee’s chest as a private conviction: <em>my job has an expiration date.</em></p><p>The tale was wholesale. The insufficiency was retail.</p><p><strong>ACT IV. THE WIRE — the con that built its own casino</strong></p><p>In Maurer’s day, the wire was the crown jewel: a fake betting parlor staffed with actors, constructed so the mark could watch the tale come true with his own eyes.</p><p>Our era’s wire is the feed itself — plus the demo economy that runs on it.</p><p>The tale said the future was certain; the wire made it <em>visible</em>. Cherry-picked showcase videos that never ship. Benchmarks tuned to the test. Agents that perform flawlessly in the keynote and break in your browser. Vaporware announcements that move markets before a product exists. And the panic distribution: “AGI next year” never traveled through peer review — it traveled through viral threads, influencer breathlessness, engagement-optimized doom. The same algorithm that spent a decade rewarding the fake vacation now rewarded the fake timeline, for the identical reason: the ranking function has no slot for <em>true</em>. Only for <em>engaging</em>.</p><p>This is what makes our con historically unprecedented, and it’s the claim I want to make rigorously: every previous con paid for its wire. This is the first con in history whose wire spreads itself — because every breathless reposter is an unpaid actor in the parlor, cast automatically by the algorithm. AI is Social Media2.0 in the most literal sense: same stage, same theatrical competition, same machine deciding what you fear. Social media 1.0 sold your attention to advertisers. Social media 2.0 uses the same machine to sell you your own obsolescence — with the cure linked in bio.</p><p>And mark the escalation across the acts: social made you feel worse than other <em>people</em>. AI made you feel worse than <em>things</em>. Each round, the standard you fail against gets less human and less real. First a curated self. Then no self at all.</p><p><strong>ACT V. THE SHUT-OUT — eighteen months, forever</strong></p><p>The shut-out is the pressure stage. The mark must believe the window is closing <em>now</em>, because a mark with time to think is a mark who walks.</p><p>Four-plus years of <em>AGI next year. Your job in eighteen months. Adapt or die.</em> The deadline arrives, passes, and is re-announced. Eighteen months, forever. “This is the worst it will ever be” — a sentence engineered to make the present feel like a closing door.</p><p>Ask what that does to a person. Fear compresses time horizons. Compressed time horizons make people buy, subscribe, comply, and defer. A population that feels permanently behind doesn’t negotiate, doesn’t question terms, doesn’t slow down to think.</p><p>That last clause is the point of the entire operation. The capacities that cannot be benchmarked — judgment, reflection, taste, trust — are precisely the ones the benchmark era trained us to discount. And here is the cruelest irony: in an economy where AI commoditizes everything measurable, those unmeasurable capacities are the only durable advantage a person has. The con’s deepest damage is teaching you to undervalue exactly the assets that matter most.</p><p><strong>ACT VI. THE STING — the con man in the mirror</strong></p><p>In the classic big con, the iron law is that the mark never learns he was taken. That’s what made the grift beautiful and the grifters untouchable. Our con broke the law, and revealed something worse.</p><p>Two departures from Maurer, and they define the era.</p><p>First: the classic sting takes the mark’s money once, cleanly, and vanishes. This sting is a <em>subscription</em>. What’s collected isn’t a lump sum, it’s the monthly fee, the data, the compressed decisions, the discounted judgment, and above all the self-worth, harvested continuously. The take was never designed to end.</p><p>Second — and this is the finding Maurer would have admired professionally: the operation achieved every grifter’s impossible dream. It outsourced itself to the mark. The industry built the insufficiency machine, ran it for fifteen years, and then pulled off the greatest cost reduction in the history of persuasion: it handed the victims the controls. Nobody must tell you you’re behind anymore. You wake up and tell yourself. You read about someone’s launch, someone’s raise, someone’s workflow, and administer the verdict personally, before coffee, free of charge. The set-up trained the reflex; the reflex no longer needs the trainer.</p><p>The watchman went home years ago. We kept the watch.</p><p>We are all the chumps. And we are the con men too — running the game on ourselves every morning, falling for it by noon.</p><p><strong>AFTER THE STING — the way out</strong></p><p>Here is what I will not offer you: reassurance. “You’re doing great” is the con running in reverse — another externally issued verdict on your worth, another dashboard.</p><p>The way out is the way of every con’s undoing: see it for what it is. A con, once named, is finished — that is the peculiar fragility of confidence games, and the reason they invest so heavily in staying invisible. And a self-run con is the most fragile of all, because the operator and the mark share one pair of eyes.</p><p>A con man cannot con someone who is watching his hands. You are the con man. Start watching your hands.</p><p>Three tests, to carry with you. When a benchmark is waved at you as proof of your obsolescence — the <strong>Toaster Test</strong>: is this a species comparison, or a tool comparison in disguise? When a winner is held up as proof you’re not hustling — the <strong>Village Test</strong>: do the winners share a method, or a zip code? When a deadline for your irrelevance is announced — the <strong>Horizon Test</strong>: who profits from your compressed time horizon? Take the long one instead.</p><p>But awareness alone decays; the wire will retrain you by Tuesday. So the second move is practice — a deliberate discipline of slow thinking: metacognition, reflection, judgment. The capacities no benchmark can measure and no algorithm can strip-mine. Even the mainstream is arriving here: Brené Brown recently told Fortune’s Most Powerful Women Summit that people are not okay — emotionally dysregulated, distrustful, disconnected — and that what this era demands is self-awareness, metacognition, and the ability to slow decision-making. The researcher who spent twenty years studying human courage looked at the age of AI and prescribed thinking about how we think.</p><p>The con told you that what makes you human is your weakness. The truth is the exact inverse: in a world where machines commoditize everything measurable, what makes you human — the judgment, the taste, the trust, the capacity to declare <em>enough</em> — is the only thing that can’t be flooded, faked, or benchmarked away.</p><p>They spent a century perfecting the con. It takes one vocabulary lesson to end it.</p><p>The trick with technology is to avoid spreading darkness at the speed of light.</p><p><strong>Footnotes:</strong></p><p>Maurer, “The Big Con” (1940);</p><p>Buffett, “The Superinvestors of Graham-and-Doddsville” (1984) — if the coin-flip passage is retained in edits;</p><p>consulting-industry job-projection reports and corresponding AI-services revenue disclosures — pull 2–3 concrete pairings;</p><p>Brené Brown, Fortune MPW Summit (Oct 2025); adoption-speed data on ChatGPT launch.]</p><p><em>Stephen Klein is the founder and CEO of </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, a Reflective AI company. He teaches “Marketing in the Age of AI” at UC Berkeley Extension and writes about the intersection of capital, institutions, and technology. </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, the only AI designed to augment human intelligence. Curiouser is community-funded on </em><a href="http://wefunder.com/curiouserai?utm_source=meta_unpaid&amp;utm_campaign=general_audience_investment_reminder&amp;utm_medium=organic_social"><em>WeFunder</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3c8ddfcc6260" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Preparing for an AI Future Isn’t About Learning AI]]></title>
            <link>https://medium.com/@curiouser.ai/preparing-for-an-ai-future-isnt-about-learning-ai-ead79207ceb3?source=rss-b3f7c93c4ed2------2</link>
            <guid isPermaLink="false">https://medium.com/p/ead79207ceb3</guid>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[critical-thinking]]></category>
            <category><![CDATA[creativity]]></category>
            <dc:creator><![CDATA[Curiouser.AI]]></dc:creator>
            <pubDate>Fri, 10 Jul 2026 21:08:37 GMT</pubDate>
            <atom:updated>2026-07-10T21:08:37.494Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>Have you noticed the companies that convinced you to spend money on technology that no one knows what to quite do with, are now moving into the business of finding ways to help you clean up the mess?</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SnohC-lUQ7GJQochz70wjw.png" /></figure><p>Everyone seems to be asking the same question.</p><p><strong>How do I use AI?</strong></p><p>It’s the wrong question.</p><p>Or at least, it’s becoming the less interesting one.</p><p>Learning to use AI is not going to be the great challenge of the next decade. The technology will become easier to use, not harder. The interfaces will improve. The models will become more capable. Before long, using AI will feel as ordinary as using a search engine, a spreadsheet, or a smartphone.</p><p>I mean is your ability to use electricity a competitive advantage?</p><p>History suggests that when a technology becomes simple enough for everyone to use, knowing <em>how</em> to use it stops being a competitive advantage.</p><p>I might think less about paying people to show you how to use technology and more about finding people who can help you think, be original, be creative, push your imagination, understand history, learn to think like a philosopher</p><p>So perhaps the better question is this:</p><p><strong>Who do we need to become when everyone has access to extraordinary intelligence?</strong></p><p><strong>Imagine the value of originality in a world of suffocating sameness</strong></p><p>That is a very different conversation.</p><p>For most of modern history, intelligence was a scarce resource.</p><p>Knowledge took years to acquire.</p><p>Expertise required decades to develop.</p><p>Finding the right answer often depended on finding the right expert.</p><p>Today, access to many forms of intelligence is becoming increasingly abundant. Millions of people can ask the same systems the same questions and receive remarkably capable answers within seconds.</p><p>If that’s true, then something profound changes.</p><p>The competitive advantage begins to migrate.</p><p>It moves away from simply having access to information and toward what we do with it.</p><p>The future may belong less to those who possess the most intelligence and more to those who cultivate the qualities intelligence alone cannot provide.</p><p>Originality.</p><p>Judgment.</p><p>Courage.</p><p>Taste.</p><p>Character.</p><p>Curiosity.</p><p>Purpose.</p><p>Visibility.</p><p>These are not merely “soft skills.” They are increasingly becoming strategic advantages.</p><p>Originality is seeing what others overlook.</p><p>Judgment is knowing which ideas deserve your attention.</p><p>Courage is acting before certainty arrives.</p><p>Taste is recognizing excellence in a world overflowing with competent output.</p><p>Character is earning trust when anyone can generate convincing words.</p><p>Visibility is ensuring that your ideas are actually discovered in a world producing more content than anyone can consume.</p><p>Curiosity is asking the questions no one else thinks to ask.</p><p>Purpose is choosing a direction that no algorithm can choose for you.</p><p>Ironically, the more accessible intelligence becomes, the more valuable these deeply human qualities may become.</p><p>This changes how we should prepare for the future.</p><p>Preparing for an AI future is not primarily about becoming better at prompting.</p><p>It is about becoming better at thinking.</p><p>Not thinking faster.</p><p>Thinking more independently.</p><p>Not generating more ideas.</p><p>Developing better ones.</p><p>Not relying on machines to distinguish ourselves.</p><p>Cultivating the qualities that machines cannot simply hand us.</p><p>Perhaps the greatest risk of the AI era is not that machines become intelligent.</p><p>It is that humans become comfortable accepting average.</p><p>When millions of people ask the same questions of the same systems, there is a gravitational pull toward sameness. Average ideas. Average language. Average strategies. Average careers.</p><p>The challenge is not to reject AI.</p><p>The challenge is to resist becoming indistinguishable because of it.</p><p>Maybe preparing for the future has never really been about technology.</p><p>Maybe it has always been about becoming the kind of person who can create, decide, lead, and inspire when everyone has access to the same tools.</p><p>Learning AI is important.</p><p>Learning how to remain unmistakably human may prove even more important.</p><p><em>Stephen Klein is the founder and CEO of </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, a Reflective AI company. He teaches “Marketing in the Age of AI” at UC Berkeley and writes about the intersection of capital, institutions, and technology. </em><a href="https://curiouser.ai/"><em>Curiouser.AI</em></a><em>, the only AI designed to augment human intelligence. Curiouser is community-funded on </em><a href="http://wefunder.com/curiouserai?utm_source=meta_unpaid&amp;utm_campaign=general_audience_investment_reminder&amp;utm_medium=organic_social"><em>WeFunder</em></a><em>. Our Gift to the community 🎁 Join us for a free live Masterclass on: Wednesday July 15th 8:00 AM PT. </em><a href="https://www.eventbrite.com/e/how-to-let-the-world-know-you-exist-tickets-1992544071957?aff=oddtdtcreator"><em>Register Here</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ead79207ceb3" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>