<?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 Paul Floren on Medium]]></title>
        <description><![CDATA[Stories by Paul Floren on Medium]]></description>
        <link>https://medium.com/@paul.floren?source=rss-4aaeda12beda------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*7vX4Td3K2E7Gej6mHyQjIw.jpeg</url>
            <title>Stories by Paul Floren on Medium</title>
            <link>https://medium.com/@paul.floren?source=rss-4aaeda12beda------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 01:15:03 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@paul.floren/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 may not destroy the world. It may simply make the human voice unnecessary.]]></title>
            <link>https://medium.com/@paul.floren/ai-may-not-destroy-the-world-it-may-simply-make-the-human-voice-unnecessary-a09a00cc8ead?source=rss-4aaeda12beda------2</link>
            <guid isPermaLink="false">https://medium.com/p/a09a00cc8ead</guid>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Paul Floren]]></dc:creator>
            <pubDate>Thu, 17 Sep 2026 06:45:23 GMT</pubDate>
            <atom:updated>2026-09-17T06:45:23.467Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jdF9WCvnk7f3y3g_ImM_Qw.jpeg" /></figure><p>Another week, another article(s) explaining how artificial intelligence is going to bring about the end of the world. Well-polished headlines with near-perfect prose with arguments that are excellently thought out, with all the predictable stages building to the climax of a perfect conclusion. Articles about the end of the world as a consequence of developing AI, written “quietly” (in quotes because, for whatever reason, AI models LOVE this word) by AI.</p><p>The updated plotline since the 1980s: Machines first seize control, then learn to play chess (wait, this one ends ok). Employment collapses. Autonomous weapons escape their operators. A super-intelligence decides that humanity is an obstacle best removed. The scenarios differ, but they share the theatrical logic of many disaster films: something terrible happens, everyone recognizes it, and a dystopian aftermath follows, with robots that look like Arnold Schwartznager.</p><p>No need to wonder: there is really no mystery at all because a growing share of what we read is written by AI or at least substantially shaped by it. Podcast scripts belabor the same points. Think “The Diary of a CEO”, with a presenter who could be fully replaced by an AI and no one would notice, or increasingly, headlines in the NYT that seem to be from a direct prompt: “write a headline for this article.” I remember when writing a headline was an art.</p><p>A break for some data to support my “feelings”: 2025 studies examining 186,000 American newspaper articles estimated that roughly 9% contained partially or fully AI-generated text; in its manual review of 100 flagged articles, only five disclosed the use of AI. Detection remains imperfect, but the pattern is real enough to make the irony unavoidable. (Take a look at the links below to see for yourself)</p><p>So it’s no surprise that machines are writing our warnings about machines. What is surprising is that the worst of all the apocalypses has already happened: we have collectively forgotten, or failed to pass on to the next generation, what it means to be an individual human being. We are, and our world is, how we act, speak, choose, remember, and become recognizable both to ourselves and to others.</p><p>This is disappearing because our laziness is giving it away wholesale, and thus something resembling the end of the world has already taken place, regardless of whether the trains still run on time (or don’t, since they never have here where I live in France).</p><p>The “quiet” apocalypse in today’s headlines isn’t that AI becomes conscious. It is that we become less so.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a09a00cc8ead" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Confidence That Wasn’t]]></title>
            <link>https://medium.com/@paul.floren/the-confidence-that-wasnt-b86934aa0f3b?source=rss-4aaeda12beda------2</link>
            <guid isPermaLink="false">https://medium.com/p/b86934aa0f3b</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Paul Floren]]></dc:creator>
            <pubDate>Sat, 20 Jun 2026 11:07:10 GMT</pubDate>
            <atom:updated>2026-06-20T11:07:10.309Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*Eq7W7RoyOo6Nfn0QCZcMXg.jpeg" /><figcaption>Photo: Paul Floren</figcaption></figure><p>Yesterday was Midsommar.</p><p>In Sweden, that means the longest day of the year (or just about, since it will actually be tomorrow, but no matter, close enough, I suppose), the point at which the sun stands as high as it will ever stand, and the days, imperceptibly, begin to shorten. The Latin root of solstice is solstitium: the sun standing still. For a brief, suspended moment, the light is at its absolute maximum. And from that moment on, without anyone even noticing it, the retreat begins.</p><p>But peak light is really an illusion, and I’ve been thinking about that image in relation to what is happening inside a lot of organizations right now.</p><p>AI can now generate confidence on demand. It can draft the statement, compose the brief, populate the deck, and fill in the silence even better than the most talkative of us (yes, I am referring to myself). In less time than it takes me to open a second browser tab, my model can produce a smart text that reads fluent, assured, and complete. It even gets the formatting right depending on my geographical location. The tone is professional. It passes AI detection with a score of 1/10 or even 0, and the uncertainty has been edited out.</p><p>But hold on, what about the midsummer metaphor? The sun never looks brighter than at the moment it begins its turn. And an organization that has learned to appear confident, through tools that manufacture that appearance, may not notice when it has stopped being confident because it has handed the steering wheel, or in this case, in Sweden, the helm, to a system that does not know what it does not know.</p><p>A recent American Psychological Association (APA) study found that heavy reliance on AI at work can erode confidence in independent thinking and reduce people’s sense of ownership over their own ideas. Passive users feel less certain in their own judgement, not more. And, perhaps not surprisingly, passive reliance on AI doesn’t just weaken judgement; it seeds a new, AI-flavored form of passive aggression within the organization.</p><p>Wharton researchers have gone further. In May 2026, they introduced what they call “Tri-System Theory” , extending Kahneman’s (yes, him again, for those that read my long pieces) classic fast/slow thinking framework with a third mode: cognitive offloading to AI. The specific risk they name is cognitive surrender: the moment when AI stops being a tool and becomes the decision-maker, without the person in the loop recognizing that the transfer has occurred. The research is clear that people who trust AI more are substantially more likely to follow incorrect AI advice and less likely to push back.</p><p>A BCG (and yes, I have used this reference before in several other posts) survey of more than 200 chief communications officers found that 68% describe their function as an AI laggard. But being behind on capability and leading on confidence are not mutually exclusive. In fact, they may be especially compatible. The less fluency a team has with AI, the more impressed they tend to be by its outputs. I know this was the case with me when I was first building Intradiegetic. But the more impressed we are, the less likely we are to question them.</p><p>Finally (actually, I stopped the first draft round about here, but leaving the finally to trick you into reading further, since it gets better), the deeper issue here is not cognitive but, like most things, an emotional maturity problem.</p><p>Emotionally mature organizations can hold uncertainty without formatting it away. They can say “we don’t know yet” without treating it as a communications failure (although, globally, this is not only an emotional problem but also a cultural one). They can distinguish between the confidence that comes from genuine clarity, earned through analysis, experience, and honest internal debate, and the confidence that comes from a well-structured paragraph that happens to end with a strong declarative sentence.</p><p>Leadership psychologists describe this capacity as emotional regulation: the ability to maintain cognitive access to judgement when identity or authority feels challenged. It is different from calm. It is the active, disciplined choice to stay in uncertainty rather than reach for a tool that will make it disappear on screen.</p><p>Daniel Goleman’s foundational work on emotional intelligence identified self-awareness and self-regulation as the core leadership capacities. Both require the same thing: a pause between stimulus and response. A willingness to sit with what is not yet resolved. AI, used passively, eliminates that pause. It gives the response before the reflection has happened.</p><p>Communications departments have always had access to tools that manufacture the appearance of confidence. Press releases. Holding statements. The classic “we are monitoring the situation closely.” These were never innovations in honesty, at least not in the way most of us used them. If we innovated at all, and I am not saying that we did, it was in the formatting.</p><p>What has changed is the speed, the volume, and the seamlessness. A statement that would have taken a team two hours to draft, and perhaps two weeks to route through legal, and wordsmith into acceptable vagueness, can now be produced in thirty seconds. The friction that used to slow the process down, the argument in the corridor, the lawyer who wanted another pass, the CCO who asked whether we actually wanted to say that, and the French exec who was convinced his English was better than the mother-tongue writers’, are disappearing.</p><p>The midsummer solstice is a celebration, and it should be. The Norse and Celtic traditions understood the turning point not as loss but as a threshold: a moment of awareness, of marking where you are before the cycle continues. The point was not to stop the sun. It was to stand in the fullness of the light, consciously, before it changed.</p><p>The equivalent move for organizations is not to stop using AI. It is to mark the threshold.</p><p>To ask: when we produce something with AI assistance, do we know what we actually believe? When we send the statement, can we defend not just the sentence structure but the claim? When the model fills in our silence, are we borrowing its confidence or have we genuinely earned our own?</p><p>Yesterday, in the long Scandinavian light, there was a moment after the dancing, the herring, the aquavit, the strawberries, when the sun barely moved, and the sky stayed bright all through what should have been night. It is one of the more disorienting experiences of living this far north. The light looks permanent. It is not.</p><p>Neither is manufactured confidence.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b86934aa0f3b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Optimism and Trust]]></title>
            <link>https://medium.com/@paul.floren/optimism-and-trust-3db88faa0633?source=rss-4aaeda12beda------2</link>
            <guid isPermaLink="false">https://medium.com/p/3db88faa0633</guid>
            <category><![CDATA[history]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[business]]></category>
            <dc:creator><![CDATA[Paul Floren]]></dc:creator>
            <pubDate>Thu, 18 Jun 2026 06:17:38 GMT</pubDate>
            <atom:updated>2026-06-18T06:17:38.659Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*02Jg9n70HKTMb4TjSRSC6g.jpeg" /><figcaption>Photo: Paul Floren</figcaption></figure><p>I can’t help but examine a pattern that recurs throughout history when I find one. I especially like an inverse one, even if they rarely end well.</p><p>So let’s take a little walk (I promise it will be a short one) along the paths of the Roman Republic’s last century, the Gilded Age’s first decades, and the early internet boom.</p><p>In Rome, the Republic expanded and prospered long after trust in the Senate had begun to erode. In the Gilded Age, industrial technology and wealth creation surged while public faith in financiers and political institutions frayed. And in the late‑1990s dot‑com boom, the world became convinced the internet would transform everything, even as questions about business models, governance, and privacy were largely waved away right up until the crash forced a reckoning.</p><p>Each time, optimism outpaced trust. People believed in the technology, the system, the institution, but they didn’t trust the people running it. And when the gap widened enough, the backlash wasn’t proportional. It was total.</p><p>I can’t help but believe we are watching the same dynamic play out with AI globally, and right now.</p><p>In Europe, a majority now believe AI’s benefits will outweigh harms, but confidence that companies will use AI responsibly sits in the low 30s.</p><p>Across Africa and South Asia, optimism is enormous: 58–75% see AI as a driver of healthcare, education, and economic growth, but trust in AI companies is at 22–30%.</p><p>In China and Indonesia, 80–83% say AI brings more benefit than harm. In the UAE, trust is higher than in almost any other country (around the mid‑60s). In the US and Germany, public trust clusters around the low 30s, alongside some of the highest ethical anxiety in the world.</p><p>One signal from the Stanford 2025 AI Index: countries with the highest AI optimism are not the ones with the highest trust in AI institutions. The two indicators are moving in opposite directions.</p><p>This is what happened in Rome when the Senate ceased to be credible, even as the idea of Rome remained sacred.</p><p>It’s what happened when Wall Street in the early 20th century outran any governance structure capable of holding it accountable, and the Depression became the corrective mechanism nobody had imagined possible.</p><p>When people believe in a system’s potential but not its stewards, they don’t stay patient forever. They look for someone to blame. Or they abandon the system entirely and reach for something cruder.</p><p>The Roman historian Livy wrote that every institution eventually faces the moment when its reputation outruns its reality, and that moment is when it becomes fragile rather than strong.</p><p>We’re not there yet with AI. But we’re building toward it faster than most people realize.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3db88faa0633" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Byblos Problem]]></title>
            <link>https://medium.com/@paul.floren/the-byblos-problem-a28a15f030c1?source=rss-4aaeda12beda------2</link>
            <guid isPermaLink="false">https://medium.com/p/a28a15f030c1</guid>
            <category><![CDATA[business-strategy]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[history]]></category>
            <dc:creator><![CDATA[Paul Floren]]></dc:creator>
            <pubDate>Mon, 15 Jun 2026 05:58:49 GMT</pubDate>
            <atom:updated>2026-06-15T05:58:49.097Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*onbN7mPeUvyYGXS4yvWsAA.jpeg" /><figcaption>The ancient port of Byblos at sunset, now in modern-day Lebanon</figcaption></figure><p>Three thousand years before anyone could fathom what a PowerPoint was beyond the sharp end of a stick, the people of Byblos solved the same problem every organization is failing to solve today.</p><p>They were a small Phoenician port city on the Lebanese coast, and they had something Egypt desperately needed: cedar. The great forests of Lebanon produced timber that the Nile delta could not. Egypt had gold, linen, and grain. Byblos had the wood to build ships, temples, and coffins for pharaohs.</p><p>The trade route was obvious. What was not obvious was the organizational design required to sustain it.</p><p>The merchants of Byblos did not simply load timber onto boats. They redesigned how decisions moved through their city. They created specialized roles for negotiation, logistics, and relationship management with the Egyptian court. They built information networks along the coast so that price signals, political shifts, and fleet movements reached the right people before they reached anyone else. They separated the people who detected opportunity from those who acted on it, and they connected the two groups with clear lines of accountability and speed.</p><p>The result was not just a successful trade. It was dominance. For centuries, Byblos was the most powerful commercial force in the Mediterranean. The Egyptians named the city in their records. The word for book in Greek, biblos, comes from this place, because so much papyrus passed through its harbor that the city became synonymous with the written word itself.</p><p>Lebanon’s forests were available to anyone willing to sail there, and having the best cedar was not their competitive advantage, allowing them to win. They won because they built an organization that could move insight into action more effectively than anyone else in the region.</p><p>The technology, in their case, was maritime trade. In ours, it is artificial intelligence. The design problem is identical. MIT has been circling this for months now, and they finally published on it properly last week.</p><p>In their coverage of agentic AI and organizational design, MIT Technology Review put into words what many in technology have been feeling but not quite articulating: most companies are failing because their organizations were not designed to make decisions with AI in the loop.</p><p>I have spent the last two years building something I call Intradiegetic. It is a distributed AI infrastructure, yes, but more than that, it is a thinking system. A way of asking: what does it look like when a human and a machine build a decision architecture together, from first principles, rather than layering intelligence onto a structure that was designed for a different era?</p><p>The answer, it turns out, is that you cannot separate the technology from the philosophy of who owns what.</p><p>Let me start with the MIT observation. Their research, and the parallel work from Davenport and Bean at MIT Sloan, converges on a single finding: only 23% of AI deployments in organizations today generate a clearly measurable return on investment. The failure rate of corporate AI projects sits above 45%. And when you ask why, the answer is almost never the model. It is the organization.</p><p>Specifically, organizations are layering AI agents onto decision structures built for human-speed, hierarchical, politically managed certainty. And AI produces something different. It produces probabilistic, fast, sometimes counterintuitive insight. The structure cannot absorb it. The insight disappears. The investment disappears with it.</p><p>Jim Wilson of Accenture, speaking at MIT’s BIG.AI conference in April, framed it this way: the ROI question for AI should not be which tool to buy. It should be determined by whether the organization is structured to adopt it well.</p><p>I think that framing is close, but it still understates the problem.</p><p>The deeper issue is not just structure. It is accountability.</p><p>Here is the pattern I have seen repeatedly in corporate communications, infrastructure operations, and investor relations. An AI system surfaces a signal. A risk pattern, a sentiment shift, a crisis emerging at the edges before it reaches the center. The system is correct. The insight is actionable. And nothing happens.</p><p>People know the information. They have it in front of them. But nobody is accountable for the cost of ignoring it.</p><p>Aristotle made a distinction between two kinds of knowledge: episteme, the knowledge of facts, and phronesis, practical wisdom, the capacity to act well in the face of uncertainty. Most organizations are investing heavily in episteme. They are buying the best models, data pipelines, and dashboards. And they are treating phronesis as somebody else’s problem.</p><p>But phronesis is not a personality trait you can hire for. It is a structural condition. It emerges when the person who receives the insight is also the person who bears the consequence of ignoring it. When those two things are separated, you do not get wisdom. You get information theatre.</p><p>The crisis communications world is a useful lens here because the stakes are visceral.</p><p>Mid-market companies, the 500- to 5,000-employee organizations that form the backbone of most economies, face exactly the same crisis exposure as large enterprises. Cyber incidents are now the second-most common trigger for formal crisis plans. Reputational events move faster than ever. And yet no integrated, agentic solution exists for this tier of company.</p><p>The enterprise platforms are designed for governments and multinationals. The cost and complexity of implementation place them outside the reach of most mid-market buyers. The gap is real and widening as the pace of crises accelerates.</p><p>What does an agentic crisis communications architecture actually need to do? It needs to detect a signal before it becomes a headline. It needs to draft a holding statement in the time it takes to convene a call. It needs to route the right information to the right people without a human having to manually triage every alert at 3am.</p><p>All of that is technically achievable today with off-the-shelf open source tools. The architecture exists.</p><p>What most systems get wrong is the human gate.</p><p>In Intradiegetic, I have a rule I have come to call the distribution principle.</p><p>Automation handles speed at detection and drafting. Human judgment owns distribution consequences. Always.</p><p>It is a philosophical position much more than a liability hedge.</p><p>The philosopher Paul Ricoeur wrote about what he called the attestation of self, the idea that identity is not something you possess statically but something you demonstrate through action under uncertainty. You do not know who you are in a crisis by reading a policy document. You know it in the moment of decision, when the draft is ready, and the send button is visible, and something in you pauses.</p><p>That pause is the system working.</p><p>Every agent pipeline I build has what I call a decision authority policy: a documented answer to what the agent can do autonomously, what requires human approval before execution, and what is entirely off-limits to automated systems. I don’t particularly distrust the models, but I believe that the moment you remove human accountability from a reputational decision, you haven’t made the process faster; you have made it hollow.</p><p>MIT Sloan’s Davenport and Bean predicted, at the start of this year, that 2026 would be a year of reckoning for agentic AI. The bubble, they argued, would begin to deflate as organizations confronted the gap between what agents promise and what existing structures can absorb.</p><p>I think they are right about the reckoning. I think they are slightly wrong about the cause.</p><p>The bubble is not inflated by bad technology. It is inflated by a category error: the belief that deploying an agent is a technology decision rather than an organizational one. Companies are asking their CTO to answer a question that is actually meant for their Chief of Staff, General Counsel, or Chief Communications Officer.</p><p>Who owns the decision that follows the model?</p><p>If you cannot answer that question for every output class your AI system produces, you do not have an AI strategy. You have an AI pilot that is about to teach you something expensive.</p><p>What I find philosophically interesting about this moment, and I think it is genuinely interesting rather than just professionally convenient, is that the question AI is forcing onto organizations is one that philosophy has been asking for millennia.</p><p>What does it mean to act well under uncertainty?</p><p>Epictetus drew the line at what is within our power and what is outside it. The Stoics built an entire ethics around the discipline of distinguishing between the two. And what AI is doing, structurally, is redistributing that line. Some decisions that previously required a human now sit firmly outside human capacity, in the sense that a human cannot process the signal fast enough to act on it. The agent can. Other decisions, the ones with reputational weight, with ambiguity, with consequences that ripple in ways the model cannot fully anticipate, remain inside the human domain in a way that no level of model sophistication will change.</p><p>The organizations that figure out where that line sits and build their architecture around it deliberately are the ones that will generate that elusive 23% return. The others will generate beautiful dashboards and expensive lessons.</p><p>Intradiegetic is my attempt to work this out in public, one build at a time.</p><p>It is named for a concept from narratology: intradiegetic, existing within the story world, as a participant rather than a narrator. I chose it because I think that is the only honest posture for someone building AI systems right now. You are inside the story. The design choices you make are not neutral observations about technology. There are arguments about how decisions should be made, who should be accountable for them, and what it means to act well when the information arrives faster than the wisdom to use it.</p><p>The question I keep coming back to is whether organizations are sufficiently accountable to act on what their agents know.</p><p>That is the design problem. And until it is treated as a design problem, the investment will keep flowing, and the value will keep disappearing into the gap between insight and decision.</p><p>What does accountability look like in your organization when the model already knows what needs to happen?</p><p>The merchants of Byblos did not ask whether cedar was valuable. They asked who was accountable for getting it to Egypt before the next ship from Tyre did. That question built a civilization. It is still the only question that matters.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a28a15f030c1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Alphabet Outlasts the Army]]></title>
            <link>https://medium.com/@paul.floren/the-alphabet-outlasts-the-army-1652f99da71f?source=rss-4aaeda12beda------2</link>
            <guid isPermaLink="false">https://medium.com/p/1652f99da71f</guid>
            <category><![CDATA[business]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[history]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Paul Floren]]></dc:creator>
            <pubDate>Mon, 08 Jun 2026 07:15:45 GMT</pubDate>
            <atom:updated>2026-06-08T07:15:45.796Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/825/1*CSB2931aYS9XWlSs5FTbgA.jpeg" /></figure><p>There are moments in our lives, as there are in stories, that deserve more time than we give to simply reading the words themselves. The story of Elissa is one of these.</p><p>Elissa had just fled Tyre. Her brother Pygmalion had murdered her husband and seized the throne. She took what she could: a loyal circle, a cache of gold, the temple priest of Astarte, and sailed west into a sea that most people of her era treated as a wall rather than a road. She was not the first Phoenician to do that. Her people had been threading these waters for three centuries already. But she was the first to plant a universe of new ideas permanently on the North African shore.</p><p>When she arrived and negotiated for land with the Berber king Hiarbas, she asked only for as much ground as could be covered by an ox hide. He agreed, assuming he was being generous to a refugee. She cut the hide into strips so thin they could encircle an entire hilltop. The city she traced on that hill became Carthage. Qart-Hadasht in Phoenician. The New City.</p><p>The Phoenicians are the most underrated civilization in the Western tradition, but like the Assyrians, they are often overlooked because they did not build enduring monuments. They also did not write epics that became school curricula. What they did build was at least just as significant, though. They built a network of over 300 coastal colonies from the Levant to the Iberian Peninsula, connected by ships that could go farther and carry more than anything their rivals had. They invented a functional alphabet, which is so much more than a mere literary project; it is the architecture upon which those projects can be built. It became the first CRM. Merchants needed to record transactions faster than cuneiform allowed, and so they simplified. Twenty-two characters, multi-platform OS. Adaptable to any language. Even the Greeks borrowed it. Then the Romans borrowed it from the Greeks. Regardless of whether you are shaking your head or about to put down this article for something else, you are reading its direct descendant right now.</p><p>This is the thing about the Phoenicians: they were not empire builders like so many before and after them; they were network builders. Contrary to the empires of the time, their power was relational and informational. Whether it was knowing the routes or holding the chokepoints, they understood that the real currency of the ancient Mediterranean was not tin or grain or purple dye, though they traded all of those; it was knowing what others did not know and being capable of what others were not.</p><p>Carthage grew from that logic. By the third century BC, it was the second-largest city in the Mediterranean, behind only Alexandria. Its fleet dominated western trade. It had embedded colonies in Sardinia, Corsica, Sicily, and Iberia. It was far more than a single monolithic power directed from a throne. It was a distributed system, with semi-autonomous nodes operating under shared commercial incentives. It had managed to survive centuries of pressure from Etruscans, Greeks, and the slow consolidating rise of Rome to the north.</p><p>I hope you are enjoying the story thus far, and don’t worry, we are nearing the last chapter.</p><p>So now imagine that we have moved up in time to when Rome decided it could no longer tolerate a distributed, adaptive, commercially dominant rival operating across the sea.</p><p>The Punic Wars, three of them, spanning 118 years, are usually taught as military history. Hannibal crossed the Alps with war elephants. The double envelopment at Cannae. Scipio’s eventual victory at Zama. But that framing misses what was actually being decided. These were not territorial wars in the traditional sense, even if they looked horrifically familiar. They were a contest between two fundamentally different theories of power.</p><p>Rome’s theory: centralized command, standardized structure, scalable legions, unified law, and the conviction that the only stable world is one where everyone operates within “your” system.</p><p>Carthage’s theory: adaptability, commercial sovereignty, distributed nodes, and the belief that a network of capable autonomous actors is more resilient than any monolithic structure.</p><p>I think you are starting to suspect where I am going with this. But bear with me for a couple more paragraphs.</p><p>For most of the Second Punic War, Carthage’s theory was winning. Hannibal crossed the Alps because it was unexpected, even so far from the most direct or practical route. He fought in Italy for fifteen years without a fixed supply line, because he had built an army capable of sustaining itself from the environment. At Cannae, he faced a Roman force nearly double the size of his own and destroyed it with a double envelopment, one of the most studied tactical moves in military history, precisely because he refused to fight Rome on Rome’s terms. He adapted. Constantly.</p><p>Students of history will now shake their heads at me because even if they won, Carthage still lost.</p><p>Not at Cannae. Not even at Zama. It lost across decades, because the Roman state was willing to absorb catastrophic short-term losses in the service of a long-term structural objective. They had the winning architecture. Rome rebuilt its navy from scratch after the First Punic War, it had no seafaring tradition, so it captured a Carthaginian ship and used it as a template. It adapted too, but it adapted to eliminate rather than to compete. In 146 BC, after the Third Punic War, Rome did not occupy Carthage. It burned it for seventeen days. It leveled the foundations. It salted, according to some accounts, the earth. It wanted to ensure that no one could ever rebuild on that site and become that kind of rival again.</p><p>The phrase “Carthago delenda est”, Carthage must be destroyed, had been a political mantra for years before the siege. Cato the Elder reportedly ended every senate speech with it, regardless of the topic. It was not really about Carthage anymore. It was about the refusal to allow an alternative model of civilizational organization to persist.</p><p>Ok, so finally here we are, the point I am trying to make, and you are forgiven if you skipped over the “super fun” trip through history to get to it.</p><p>If you have spent any time thinking about what is happening with AI right now, the echo should be loud enough to ring between Tunisia and Sicily.</p><p>We are living through a moment that maps with precision onto the late third century BC.</p><p>On one side, you have two or three large centralized systems built by organizations with near-unlimited compute, trained on effectively all of human writing, distributed through controlled APIs, operating under terms that route your data, your queries, your dependencies, back through a single point of control. They are genuinely impressive. They are also structurally Rome. Scalable. Standardized. Oriented toward total absorption.</p><p>On the other side, you have something much younger and considerably messier: local AI. Models that run on consumer hardware. Inference stacks that stay within your own infrastructure. Systems that you own, that see only what you show them, that can be fine-tuned on your actual context rather than the average of all human production. Ollama. llama.cpp. Small, capable models from Mistral and the growing open-source ecosystem. The Phoenician alphabet of the present is practical, adaptable, and not beautiful in the way cuneiform was, but powerful precisely because anyone can use it.</p><p>The instinct in most organizations is to centralize. Buy the enterprise license. Route everything through the API. To consolidate in the hopes of being efficient, and managers come across as being decisive. AND it mirrors what Rome always told itself: that unity and standardization are the conditions of strength.</p><p>BUT Hannibal already disproved that. His army, operating without a fixed base in hostile territory for fifteen years, was not a symbol of chaos that so many see in emerging AI today, but rather a demonstration that a sufficiently adaptable distributed system can outperform a monolithic one in almost every tactical engagement. He just could not outlast Rome’s institutional patience and its willingness to absorb the cost of eliminating him as a model.</p><p>What does this mean for us as we consider how to integrate AI right about now? For anyone building something they still want to own in a decade, it’s not about “which model is best.” It is: are you building a network or becoming a dependency?</p><p>Elissa did not build a replica of Tyre. She did not try to reproduce the city she came from. She read the geography of the North African coast; she understood what was available; she negotiated within the constraints she had; and she built something new, informed by Phoenician principles without being limited to Phoenician form. The ox hide trick was a problem-solving under constraint.</p><p>Some tend to see Local AI as a consolation prize for people who cannot afford the big models. It is a different design philosophy. Data stays inside your perimeter. Latency disappears. Inference costs approach zero after the initial hardware investment. Your institutional knowledge, the documents, the conversations, the decisions that make your organization what it actually is, can become part of the model without ever leaving your control. You are no longer a customer; you are much more: an operator.</p><p>The companies and individuals building this way right now are doing what the Phoenicians did when the Assyrian and Persian empires were pressing against their eastern borders: they played the dominant powers where they had to, and they built independently where they could. They did not resist the empires frontally. They maintained the parts of their network that the empires could not easily reach.</p><p>Adaptability is not the same as compromise. The Phoenicians were not weak because they moved laterally around power. They were strong precisely because they understood that network resilience depends on not putting everything under a single point of failure.</p><p>Carthage fell because it eventually had to fight Rome frontally. It had no choice. The wars came to it. And when you fight a centralized monolith on its own terms, the monolith usually wins.</p><p>The lesson is not that distributed systems always lose. It is the moment a distributed system stops adapting and starts imitating the structure of the power it is competing with that it loses the only advantage it ever had.</p><p>The AI landscape in 2026 is not yet 146 BC. We are, if anything, somewhere in the middle of the Second Punic War. The outcome is not written. There are organizations and individuals right now building real, sovereign AI infrastructure, and they are doing so because they understand that owning your inference, your context, your knowledge layer is a structural advantage that compounds over time.</p><p>Elissa cut the ox hide into strips because she understood the problem. She was trying to get enough ground to build something that would outlast the conversation.</p><p>What are you actually building, and how thin are you willing to cut the strips?</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1652f99da71f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Can You Read My Mind, or Are You Just an AI Agent?]]></title>
            <link>https://medium.com/@paul.floren/can-you-read-my-mind-or-are-you-just-an-ai-agent-0fad51890c12?source=rss-4aaeda12beda------2</link>
            <guid isPermaLink="false">https://medium.com/p/0fad51890c12</guid>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[business]]></category>
            <dc:creator><![CDATA[Paul Floren]]></dc:creator>
            <pubDate>Thu, 04 Jun 2026 07:20:04 GMT</pubDate>
            <atom:updated>2026-06-04T07:20:04.596Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*o0dBP6C5J331u3raGeCfQQ.jpeg" /></figure><p>There is an old philosophical problem. Hume argued that memory does not just record the past; it constitutes the self. What you remember shapes what you perceive, what you decide, and what you ignore.</p><p>We tend to believe that the era we live in is unique. For the most part, it is not because people are people, and their hopes and fears remain the same over the course of evolution. But, in some ways, they are unique, and this is where we find the small gap of what matters.</p><p>Recently, the news was full of stories about Anthropic running pilots inside Fortune 100 companies, where agents monitor internal chats, documents, and meetings and begin surfacing patterns that influence who gets hired, what gets funded, and how risk gets assessed. We do the same thing locally at Intradiegetic, although not at the same scale, but with A LOT more impact.</p><p>The agent reads the stream of organizational life and acts on it; indeed, it is reading what I am writing right now as it is saved in the general writing corpus and ingested. The impact: constant comparison to other activities going on in the company; what has been written in the past and is being worked on right now; and what belongs only to us at Intradiegetic. It is local AI, as a partner, not a service.</p><p>Institutional memory has always encoded the priorities, biases, and blind spots of the people who built it. The difference now is that this memory is being processed at machine speed, with recommendations attached.</p><p>That creates a structural problem that most firms are not ready to answer. When an agent surfaces a pattern in real time, who decides what a pattern worth surfacing looks like? Where does the accountability architecture live? These are governance questions dressed in technical clothes.</p><p>The data these agents learn from is not neutral. It reflects the organization as it was, not necessarily as it should be. A system trained on your internal communications will learn your politics, your silences, and your defaults. The question worth asking is whether you want those defaults encoded into the decision layer.</p><p>The real shift is in how we think about knowledge: do we expect to file it, or is it to be read back in real time?</p><p>That is a fundamentally different relationship with organizational memory. And the design challenge is not how to make the agents faster or cheaper. It is about building systems that can surface what the organization does not already know it believes.</p><p>Hume would be awed. He might also remind us that what we choose to remember is already a decision</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0fad51890c12" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The (Almost) End of a Love Story on the Øresund Bridge. My AI fixed it.]]></title>
            <link>https://medium.com/@paul.floren/the-almost-end-of-a-love-story-on-the-%C3%B8resund-bridge-my-ai-fixed-it-c89eb2620a3f?source=rss-4aaeda12beda------2</link>
            <guid isPermaLink="false">https://medium.com/p/c89eb2620a3f</guid>
            <category><![CDATA[saab]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Paul Floren]]></dc:creator>
            <pubDate>Wed, 03 Jun 2026 05:19:21 GMT</pubDate>
            <atom:updated>2026-06-03T05:24:29.075Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aBUY07YdFJ18iL9qTOvodw.jpeg" /></figure><p>Love stories always start differently, or at least mostly, and no, I am not going to go on and quote Tolstoy here, even if I am dying to make the connection. They mostly end differently, though, and mine with my SAAB was on the verge of a dramatic breakup yesterday. It was early in the morning when I set off from my home outside Paris for a leisurely drive to Sweden. Lots going on in Sweden, especially in the AI space.</p><p>I joked to myself that my old SAAB was on its way back to its roots, and caught myself laughing at how silly I can be for anthropomorphizing a car. In the tunnel, and then halfway across the bridge from Denmark to Sweden, the dashboard lights started to flash and the dreaded “limp mode” activated. Could it have been that the emotions of coming home were too much for it? Could this be the end of a beautiful relationship?</p><p>But wait, that is not what this post is about; it’s about Agentic AI and especially that success is no longer measured by problems solved. It’s measured by the speed with which failures are contained.</p><p>My first call was not to a tow truck; it was to my two Agentic AIs, which have access to my personal corpus and thus the mechanical history of my cars. Arthur, the GPU powerhouse, did the detailed root analysis, while Pippa, the Open Claw agent, did the “next” steps problem-solving. Long story, short, they solved the problem and gave me a series of snaps and clicks with the pedals and lights (like something out of a Monty Python), and in no time, I was back on my way like nothing had happened at all.</p><p>They didn’t fix the car. But they helped me decide what to do next — and that distinction matters more than it sounds. How does this connect to Enterprise AI? It isn’t about solving problems anymore. It’s about managing the consequences of solutions.</p><p>AI is increasingly engineered less to optimize outcomes and more to absorb imperfect execution, to help fix the SAAB in the middle of the bridge. The value proposition is changing from “here’s what AI can do for you” to “here’s how AI can contain the damage when your decisions fail.” The underlying assumption is that AI’s strategic value lies in its capacity to obscure the downstream consequences of human decisions. In this case, driving 18 hours to Sweden instead of taking a 90-minute flight (ok, here you are justified to raise your hands in the air and say “what were you thinking?”)</p><p>A long-winded story to say that the next frontier won’t be defined by better models. It will be defined by systems designed to fail gracefully, engineered for reliable collapse rather than reliable performance.</p><p>A 20-year-old SAAB is on the bridge back to its country of origin, which is building the corporate roots of the future. Not fixing them, constructing them.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c89eb2620a3f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to Fall in Love, Not Quietly]]></title>
            <link>https://medium.com/@paul.floren/how-to-fall-in-love-not-quietly-4c007c70c57d?source=rss-4aaeda12beda------2</link>
            <guid isPermaLink="false">https://medium.com/p/4c007c70c57d</guid>
            <category><![CDATA[language]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[love]]></category>
            <category><![CDATA[business]]></category>
            <dc:creator><![CDATA[Paul Floren]]></dc:creator>
            <pubDate>Mon, 01 Jun 2026 04:20:19 GMT</pubDate>
            <atom:updated>2026-06-01T04:20:19.023Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/737/1*YOGzlYB0ceEhm6dvvRuhUw.jpeg" /><figcaption>A crumpled love letter. Illegible. Unmistakably human.</figcaption></figure><p>Words are important. Arranged properly, they can let us set sail for some of the most exciting journeys the human experience has to offer. Tales of love, tales of loss, tales of woe, but especially of love. The words we choose allow the reader to fall in love with our characters, or not, but also with the writer. Their emotion, passion, and a glimpse into the subconscious are why writing is among the most vulnerable forms of art.</p><p>Quietly.</p><p>“Quietly” is the perfect AI word. It feels literary and, as such, intentional, as if this is where the writer paused, chose carefully, and wanted you to sense something unfolding beneath the surface. Quietly.</p><p>But it does nothing. It adds no information. It changes no meaning. Remove it from any sentence it lives in, and the sentence gets better. Much better, and ironically, it’s everywhere now in virtually every LinkedIn article, in newsletters, in job descriptions, and even in a headline in “the paper of record” on the front page of today’s NYT.</p><p>Here’s why AI loves it: it collapses two things that are hard to write and define us as humans, drama and restraint, into a single adverb. It is basically AI trying to be clever and hide that it is performing intelligence while making it sound like depth. It doesn’t know how to signal that what is happening is significant, so it is going to whisper it to sound more sophisticated because screaming is somehow undignified. Screaming is, but it is also invariably human.</p><p>AI doesn’t know it’s doing this. It’s pattern-matching on millions of texts where humans used “quietly” to signal gravitas. So it reaches for it. If AI could think for themselves, it would, at this point, be congratulating itself for being clever.</p><p>The problem is not the word. The problem is what it reveals: writing that optimizes for the feeling of meaning rather than for meaning itself.</p><p>That’s the real AI tell. Not “delve.” Not “furthermore.” Not “it’s worth noting that.”</p><p>It’s adverbs that create depth rather than deliver it.</p><p>Quietly is the most honest dishonest word AI has ever learned, but more importantly for us humans, you cannot fall in love with a writer who whispers on cue.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4c007c70c57d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Who Owns the Idea?]]></title>
            <link>https://medium.com/@paul.floren/who-owns-the-idea-a5f6fe4e172a?source=rss-4aaeda12beda------2</link>
            <guid isPermaLink="false">https://medium.com/p/a5f6fe4e172a</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ideas]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[history]]></category>
            <category><![CDATA[philosophy]]></category>
            <dc:creator><![CDATA[Paul Floren]]></dc:creator>
            <pubDate>Fri, 29 May 2026 04:12:40 GMT</pubDate>
            <atom:updated>2026-05-29T06:02:40.247Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rzLm1yZ5wqg4CFMsr2OfnQ.png" /></figure><p>In Sumer, the gods held the Me.<br>In Egypt, Thoth wrote the words before you did.<br>In Athens, you asked a Muse.<br>In 2026, you open a tab and ask a model.</p><p>Working all day with advanced AI systems is fascinating. There is a lot of coding, but for a communicator like me, it’s interesting and oh-so-satisfying to see your ideas take shape. The fascinating bit comes from the combination of structure, psychology, philosophy, and pure creativity. And then sometimes a particular idea or a solution comes from across the ether onto your screen, and the idea is original and perfect and purely from an AI, which pushes larger inquiries. If I didn’t generate it, do I own it? Am I accountable for it?</p><p>For most of recorded history, the question “where do ideas come from?” was a spiritual one, with an answer rooted in culture and tradition that said something important about who you were and what you were responsible for.</p><p>Skip to today, and it seems we’ve swapped that ancient question for efficiency. Most organizations haven’t noticed what they gave away in the trade.</p><p>The oldest written framework for where ideas come from doesn’t belong to the Greeks (yes, if you read some of my other posts, you would be justified in thinking that I was going there, and I will, but not quite yet). It belongs to the Sumerians, a civilization in southern Iraq that was already ancient when Athens was a village.</p><p>Their concept was the Me (pronounced “may”): a set of divine powers that underpinned all of civilization — writing, judgment, music, wisdom, craftsmanship, law, and even descent and lamentation. The Me existed before any human held them. In the myth of The Gifts of Inanna, the god Enki, patron of wisdom and the sciences, holds the Me and transmits them to Inanna, who carries them to humanity. Ideas, in this framework, were not generated. They were received, transmitted, and stewarded. The human was the channel. The Me was the source.</p><p>Make sure to read the sobering fine print. The Me also included darker forces, deceit, envy, and warfare. The Sumerians understood that the same divine transmission that gives you wisdom also gives you the capacity for catastrophic misuse. The gift came with a shadow attached. Ancient Mesopotamian scribes performed a ritual acknowledgment before beginning their work: the knowledge they used had a source, and they were accountable to it.</p><p>Maybe you get where I am going with this, but not just yet, since the journey in this case is so much more interesting than the destination. So now, Egypt-bound, where and when knowledge was morally weighted.</p><p>In Egypt, hieroglyphics that we have all seen in museums had a name: medu netjer, “the words of the gods.” Writing was not invented by humans. It was given by Thoth, god of wisdom, knowledge, and cosmic balance, a deity venerated continuously for thousands of years, from the earliest dynasties through the Ptolemaic period. Before starting work each day, scribes would pour out a drop of ink in Thoth’s honor. The tradition held that Thoth had encoded all knowledge humanity would need into 42 books, some reserved as sacred, revealed only to initiates.</p><p>But the more structurally interesting element is Maat, not just truth or justice, but the order of the cosmos itself. Ideas aligned with Maat carried cosmic weight. Ideas that violated it were existentially wrong, not just incorrect. At death, your heart was weighed against Maat’s feather. This is a system where knowledge is always already morally loaded, not neutral, not a resource. Every idea carried accountability built in.</p><p>Nope, we haven’t arrived at our destination yet. Board the next ship further with the light of the Pharos of Alexandria in our eyes, and off to West Africa and then to the Australian desert, where the self you choose, the land that remembers becomes beauty in truth.</p><p>In Yoruba cosmology, Orunmila is the deity of wisdom and omniscience, the Witness of Fate, present at the creation of all souls. But the more striking concept is Ori: the personal divine consciousness that each soul chooses before birth, the inner source that shapes all decisions, creativity, and thought throughout a life.</p><p>This is close to the Roman genius, but with a crucial difference. You don’t receive Ori from outside. You select it before incarnating. Authorship and responsibility are entangled from the start, ideas don’t come from a god external to you, they come from the Ori you chose, which is inseparable from who you are. The Ifa divination system that Orunmila governs functions as a vast knowledge framework: wisdom is already encoded in the corpus; the human task is to access it correctly and interpret it for the specific situation. A system that holds the knowledge. A person who is responsible for the application. Sound familiar? I know that it was at this point in my research that I was nodding at the screen, thinking, “Wow, AI is not that unique in the way it builds.”</p><p>A bit more patience, and we are getting there.</p><p>The most radical version of this framework belongs to Aboriginal Australians, the oldest continuous culture on earth, spanning more than 50,000 years. In their tradition, the Dreaming (Jukurrpa) is not a creation story confined to the past. It is a permanent reality that exists across the past, present, and future simultaneously. Ancestor Beings encoded knowledge, law, and meaning into the land itself. Songs, stories, ceremony, and art are not representations of ideas. They are the ideas, maintained in the country.</p><p>In this framework, ideas don’t come from individuals at all. They come from the land, the Ancestors, and the relational practice that keeps the transmission alive. The human is custodian and transmitter — not creator. To transmit poorly, or to transmit sacred knowledge to the wrong context, is a breach with consequences that extend across community and country. Accountability more total than anything most organizations have ever designed.</p><p>Ok, so finally we are here, you are thinking, and you would be right. It’s a bit long-winded, but often the best philosophical points are, and who am I to muddy tradition?</p><p>So now we fast-forward across fifty centuries. When you type into an LLM, you are not invoking Thoth or asking the Muses or consulting Orunmila. You are tapping something closer to an industrial commons, a system trained on the aggregate of written human thought, optimized to return the most statistically probable next word given what you asked. It is, by design, an averaging engine.</p><p>The research is becoming hard to ignore. A 2024 study of 300 writers found that AI assistance improved individual outputs but significantly reduced collective diversity, pulling everyone’s work toward the same center. A 2025 paper in Nature confirmed that AI models trained on AI-generated content undergo “model collapse”: successive generations lose the edge cases, the rare inputs, the things that make outputs genuinely distinct, eventually degrading into what researchers described as “unrelated nonsense.” The World Economic Forum noted plainly that models trained on the same data will, over time, generate the same outputs. And I and others have published numerous posts and articles on this right here in this forum.</p><p>Yet, research published in Nature Human Behavior in early 2026 went further, showing that AI adoption is producing a scientific monoculture converging not just in what is studied but also in how questions are framed, which answers appear credible, and what language is treated as legitimate. The feedback loop: AI-assisted work becomes more visible, more cited, more rewarded, which accelerates the convergence of the very thought that gets fed back into the next generation of training data.</p><p>Yes, it’s about architecture, but moreover, homogenization in this case is not a bug in a few writing tools. It is structurally built into the system.</p><p>Jay Barney (author of the most-cited paper in strategic management) published a blunt conclusion in MIT Sloan Management Review in 2025: AI will not provide sustainable competitive advantage. His logic is clean. For a resource to generate a durable competitive advantage, it must be Valuable, Rare, and Inimitable. AI is valuable. It is not rare. It is not inimitable. When every function in every organization uses the same models and templates, convergence is not a risk. It is the default.</p><p>So is everyone consulting the same oracle, then? Is this just a case of Delphi being overrun and the priests there getting fat on goats and ducks?</p><p>Most of the conversation about AI homogenization has stayed in one room: content, brand voice, and communications. Does everything start to sound the same? Yes, measurably. But stay in that one room, and you miss the deeper problem.</p><p>Strategy decks drafted with the same prompts on the same three models. HR policies written by systems primarily trained on other companies’ labor laws and risk tolerances. Product decisions filtered through “best-performing” templates harvested from industry averages. Financial forecasts shaped by models that have learned what forecasts are supposed to look like. Sales approaches optimized toward the same conversion patterns everyone else is optimizing toward.</p><p>Individually, none of these is catastrophic. Collectively, they push the entire organization toward a single, statistically optimized version of “good,” safe, legible, average. The Deloitte 2026 Global Human Capital Trends report is direct: competitive advantage now depends less on technology differentiation and more on “cultivating the human edge,” adaptivity, creativity, and judgment under uncertainty. Technology is replicable. People aren’t. And the dominant workflow pattern of 2026 is to systematically reduce the influence of what is not replicable.</p><p>What we are building, function by function, is an organization that thinks like the average of the internet.</p><p>So maybe the question the ancient traditions (and my long-winded way of getting you here) was actually worth asking.</p><p>Here is the thing that most frameworks miss: ancient cultures didn’t locate ideas outside humans because they lacked confidence. They did it because it clarified something essential about responsibility.</p><p>Plato’s Ion makes this argument inside out. Socrates tells Ion that poets are “interpreters of the gods,” composing not through conscious mastery but through divine possession. This sounds like a compliment. It isn’t. Plato’s point is that if the poem came from the god, not from the poet’s own understanding, then the poet cannot be responsible for its meaning. They are a passive conduit. They cannot explain what they made. They cannot be held accountable for where it leads.</p><p>We should all be as suspicious as Socrates was of inspiration precisely because it dissolved authorship.</p><p>This is the exact condition we are creating at scale, but with no gods to absorb the blame. The model generates. The employee pastes. The leader approves. When a strategy fails to account for local conditions, when a policy misses its own organization’s culture, when a product answers the wrong question, there is no accountability trail. The AI suggested it. It was statistically probable. It sounded fine.</p><p>Nassim Taleb calls this the core ethical failure of modern institutions: separating decisions from consequences. Having skin in the game is not just an incentive mechanism. It is the epistemological condition for understanding the world at all. “If you give an opinion, and someone follows it, you are morally obligated to be exposed to its consequences.”</p><p>A model has no skin in the game. It does not attend the town hall after a restructuring. It does not look the client in the eye when a strategy fails. It is not there the morning after.</p><p>You are.</p><p>The Sumerians didn’t fear the Me. They respected the obligation that came with receiving them. The Egyptian scribe didn’t resent Thoth. They honored the source by doing their craft carefully. The Aboriginal custodian doesn’t own the Dreaming. They are accountable for keeping it intact for what comes next.</p><p>In each case, the source of knowledge is larger than the individual. The individual is the point of accountability. Those two things do not cancel each other out. They define each other.</p><p>McKinsey’s 2025 global AI survey found that AI high performers are defined not by tool selection but by three things: bold transformation ambitions, intentional workflow redesign, and defined processes for human validation of model outputs. The differentiator is organizational coordination and judgment, not computation.</p><p>Barney’s practical conclusion: the advantage is not whether you use AI but what you use it on, your proprietary knowledge, your distinctive context, the judgment at the edge of the situation where the model runs out.</p><p>So the question for leaders is not “how do we implement AI across functions?” It is:</p><p>In which parts of your organization does human friction — the disagreement, the local variation, the unexpected angle — still have permission to exist?</p><p>Because that is where your ideas actually come from. That is where distinctiveness lives. That is where what cannot be replicated gets made. The Aboriginal custodian doesn’t claim authorship of the Dreaming, but is completely accountable for transmitting it faithfully</p><p>The ancient traditions were right about one thing across every culture and every century: ideas don’t belong entirely to you. They arrive through conditions, context, and connection, through the specific weirdness of your situation, your people, your moment.</p><p>The mistake is not using AI. The mistake is using AI to sand away the very friction that makes your organization worth paying attention to.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a5f6fe4e172a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[C-suite: Your AI strategy is going to cost you your job]]></title>
            <link>https://medium.com/@paul.floren/c-suite-your-ai-strategy-is-going-to-cost-you-your-job-c4e8245c74f4?source=rss-4aaeda12beda------2</link>
            <guid isPermaLink="false">https://medium.com/p/c4e8245c74f4</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[business-strategy]]></category>
            <dc:creator><![CDATA[Paul Floren]]></dc:creator>
            <pubDate>Mon, 25 May 2026 06:28:10 GMT</pubDate>
            <atom:updated>2026-05-25T06:28:10.447Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*hCDNt9oKUTp0P4jeb8EWlg.jpeg" /><figcaption>Photo: Paul Floren</figcaption></figure><p>One of the hardest things to understand about AI is not how to use it effectively to be more productive and, in the process, build pretty ROI graphs that will make most everyone happy.</p><p>To build these graphs, most companies now have AI strategies. With all the hype, you would have to be crazy not to, or at least not to have a pretty deck somewhere with big words on the headline pages stating “roadmap”, “executive mandate”, and of course, the world-shattering press release announcing a transformation that no one will read, not even your employees.</p><p>No one will read it because they are getting it fundamentally wrong. While every company has an AI strategy. Almost none have an AI operating model.</p><p>There’s a difference. And it’s costing them.</p><p>The operating model is what actually changes. Who decides what. Which inputs get trusted. Where a human still needs to be in the loop. What happens when the AI recommendation contradicts the established process.</p><p>MIT’s NANDA initiative put a number on it last year: 95% of enterprise AI pilots fail to reach production with measurable impact on the P&amp;L. It is not because the models are bad, and to be honest, this is an easy fix; it is because the organization around the model was never redesigned.</p><p>You can’t bolt intelligence onto a structure built for a different century and call it transformation. I mean, you can, but it is a transformation in the wrong direction and will, in time, cost the CEO and the C-suite their jobs.</p><p>What I keep seeing, across sectors and geographies, is the same pattern: leadership buys the strategy. Middle management absorbs the tools. And the operating model — the real architecture of how decisions get made — stays exactly as it was. It is a design problem.</p><p>What it’s not: An AI operating model isn’t a governance document. It’s not a Center of Excellence with a quarterly meeting.</p><p>What it is: It’s a clear answer to four questions: What decisions are AI-assisted? What decisions are AI-led? Where does accountability sit when the system is wrong? And what does the organizational chart look like when those answers are honest?</p><p>Most companies haven’t answered any of them.</p><p>The ones that have? They’re not running AI pilots anymore. They’re running at a different speed entirely while daily demonstrating that the gap isn’t between companies with AI and those without. It’s between the ones who redesigned to think with it and the ones who still use it as a faster version of what they already had.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c4e8245c74f4" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>