2026-06-11 12:00:00
There is a useful frame for the AI moment that has been quietly emerging in the writing of the last few months. The work we do, it suggests, splits roughly into two halves. There is the half that produces — the rendering, the executing, the building, the delivering. And there is the half that decides — what to make, what to attend to, what to leave alone. AI is rapidly compressing the first half. It is doing comparatively little to the second. And the gap between what AI can and cannot do is becoming the gap between what we delegate and what we keep.
This is the throughline of the pieces below, and a thought I’ve been chewing on in my own writing as well. In Closer, Not Higher a few weeks ago, I argued that the conventional wisdom about design leadership — that good leaders rise above the details — is exactly backwards in an AI-driven field, because the part the machine cannot do for us is the part that requires more, not less, attention to detail. Several writers this week make a similar case from different angles — through the lenses of design strategy, creative play, friction, care, and the new shape of design work itself.
A reasonable working hypothesis emerges from all of it: as production gets cheaper and faster, the part of the work that decides what should exist becomes the part that matters more. The good news, if you make a living from creative work, is that AI is not so much eating your job as it is clarifying what your job actually is. This is a conclusion that flies in the face of a spreading layoff contagion, but I believe the world will eventually catch up.
Buzz Usborne names the structural distinction directly in Discovery vs Delivery: the work of building software has two phases, and AI is asymmetrically transforming one of them. Delivery — the convergent work of efficiently producing the thing once it has been decided — is where AI has been most transformative, and rightly so. Best practice exists there. It is compressible into rules, patterns, structures. Discovery — deciding what should exist in the first place — is harder to structure around AI, because the qualities that produce great outcomes (originality, judgment, restraint, taste, contextual sensitivity, care) are far less deterministic and far harder to measure. Usborne is careful to note that this is not an argument for protecting slowness; designers have always wanted to ship faster, and not every problem deserves deep discovery work. A settings page rarely benefits from asking how to make it magical. But for the work that does require judgment, the relationship flips: it isn’t human-in-the-loop anymore but AI-in-the-loop, with the human still doing the deciding. The conclusion lands as a sharper version of the thesis the rest of the entry is following: as execution becomes commoditized, judgment becomes the differentiator. Companies that automate the discovery layer entirely will produce, in his words, quickly-built, cheaply-made, average software — the same as everyone else.
Angus Wall argues in The highest level of creativity is play that the part of creative work AI cannot touch is the part that operates closest to play — the unscripted, exploratory moment where the unexpected outcome can emerge. Our creative industries have been quietly structuring play out of their workflows for decades, splitting work into sequential phases through which an idea must survive a long game of telephone between disciplines. The casualty is exactly the kind of creativity that produces work people remember. Wall is hopeful about AI for an unexpected reason: it is forcing the old silos to dissolve, letting artists cross into neighboring disciplines and take their thinking further than ever before. He reaches for a sports analogy that’s worth the price of admission: before the triangle offense was invented in basketball, plays were pre-scripted and the goal was to force the ball to a star scorer; the triangle offense radically distributed decision-making across the team, creating conditions where the right play could emerge naturally. The same shift, he suggests, is now available to creative teams who learn to use AI as a participant in a spatial relationship rather than as a tool to expedite the finish line. The piece closes on a line worth saving: in a media landscape oversaturated with hyper-optimized content, the work that will truly stand out is the work that still carries the fingerprints of human play.
Nicolas Cevallos takes the long way around to the same idea in 160 GB — that what stays with us is what we tend with care, and that care is itself a form of human work AI cannot do for us. The piece starts with the diamond ring as the artifact of multi-generational commitment, follows it through to the great-grandson who doesn’t recognize gold from brass, and lands somewhere unexpected: the online community of iPod Classic modders who are bringing twenty-year-old devices back to life by swapping batteries and hard drives. The iPod, Cevallos argues, was the last piece of technology that felt valuable precisely because of what it couldn’t do — no ads, no algorithm, every song and album intentionally chosen. Its survival now depends on Apple and a community of caretakers finding ways to maintain it. The piece’s quiet thesis is that the heirlooms of the future will work the same way: they will require knowledge, space, and emotional connection, but mainly they will require active participation in their continued existence. In a world that accelerates toward the next thing by default, choosing to keep something is itself an act. Every battery replacement, Cevallos writes, becomes a promise of forever, or at least a wish for forever to exist.
Nina Maria makes the friction case head-on in The revolution will not be optimized: the inefficiencies we have spent two decades engineering away — the waiting, the wandering, the small inconveniences that structured ordinary days — are not interruptions to life but the conditions that gave it texture and depth in the first place. She opens with a perfect small example: a woman who installs one of those hyper-efficient office taps that delivers boiling water instantly, only to realize she has lost the kitchen-walk pause where ideas used to surface. The philosopher Byung-Chul Han, she notes, has been writing for years about contemporary society’s excess of smoothness — a world increasingly designed to remove resistance, discomfort, ambiguity, and pause. Maria’s case is not that we should abandon convenience altogether; nobody seriously wants to handwash clothes or churn butter. The point is to become more selective about which frictions are removed and which are quietly essential to our humanity. There is no frictionless version of becoming an interesting person, she writes. The line about the underlying assumption — that we have come to treat discomfort as a design flaw rather than a part of being alive — pairs unexpectedly well with the discovery half of Usborne’s framework. The work of deciding what should exist is itself a friction. AI’s promise is to remove it. The argument the piece makes is that we should keep it.
Patrick Neeman reaches for a different analogy in Designing for AI means designing like it’s 1999: the unsettled-medium parallel between the early web and the current AI moment. The reason 1999 was a strange and exciting time to be a designer, Neeman argues, is the same reason now is — the medium hadn’t hardened yet, so the work that got done mattered more. The conventions we now take for granted on the web (the shopping cart, the news feed, the login) were invented by people making reasonable guesses and watching what happened. They weren’t discovered. They were tried. The piece lays out the parallels concretely: the standards are still being written (MCP was released in late 2024 and handed to the Linux Foundation a year later), the interaction paradigm is still being invented (chat is the first guess, not the final form), the business model is unproven (today’s pricing is a subsidy, not a settled cost), and the killer app hasn’t appeared yet. The implication that lands hardest is the one about the practitioners themselves: the people who came through 1999 well were the ones who kept reinventing themselves. The durable skill was never a particular tool but the willingness to become someone a little different every few years. The medium that gets reinvented isn’t only the medium. It’s also us.
Nicole Alexandra Michaelis answers the same question from the other direction in Design’s alive and kicking: not what AI doesn’t do, but what design becomes as AI absorbs the production layer. She maps the emerging roles by name. On the B2B side, embedded AI Design Consultants who parachute into enterprises to figure out how their workforce should actually interact with AI; Agentic UX Architects who design the asynchronous interfaces that pair with background agents running for hours. On the B2C side, Proactive Interaction Designers who own the invisible triggers (biometrics, location, time of day) that let an AI guess intent before it’s stated; Generative UI System Architects who design the constraints and guardrails that let dynamic interfaces stay coherent; Trust Designers who translate cryptographic verification into split-second visual signals. The list goes on — AI Interaction Evaluators, Prompt and Context Designers, Cognitive Researchers, Design Engineers — and what stands out is that none of them are pixel-production roles. They are all forms of choreography, judgment, and trust-design — the upstream work that decides what should exist and how it should behave. The piece closes on a line worth keeping: the premium designers of the next decade won’t be valued for how quickly they can draw a beautiful button, but for their ability to choreograph how humans and intelligent agents collaborate. Most of us, Michaelis notes, loved that part of the work most anyway.
Christopher K Wong interviews 28 design leaders in Being an AI-native designer isn’t what you think it is about what they’re actually hiring for under the new “AI-native” job descriptions, and reports the answer with some satisfaction: it has nothing to do with tools, chatbots, or designing AI interfaces. It comes down to one thing — critical thinking. The designers becoming valuable in this moment are the ones who can hear “build a marketplace” and stop to ask why they’re building it, what they’re hoping to achieve, who it’s actually for. When a problem is well-defined and follows a standardized pattern, businesses now turn to AI instead of designers. Ambiguous problems still need human judgment, and the messy, uncertain, “I-know-we-need-to-do-something-but-I’m-not-sure-what” situations are where the value has migrated. Wong’s contribution is a practical three-question test: before handing any task to AI, ask whether it’s clearly defined, whether you know what a good output looks like, and whether you’d catch a bad output if you saw one. Pass all three and you can delegate. Fail any of them and you’re not ready yet — and the AI will produce work whose quality you can’t actually assess. Being AI-native, in his framing, is not about using AI more. It’s about knowing what to delegate and what to own. The line that closes the piece sticks: let AI do the laundry and dishes so you can do the art and the writing, not the other way around.
Rising: Critical thinking as the AI-native baseline. The skill named most often in hiring conversations is the one no tool can confer, and the one job descriptions have only recently started asking for explicitly.
Setting: The pixel-pushing identity of design. The mechanics of production are quietly migrating out of the designer’s hands and into the agent’s; what remains is the part of design that was always its center but rarely its title.
Color: Burnt sienna (#A0522D)—the color of patience, of unhurried hand-work, of objects that show their wear because they have been used long enough to wear.
Sound: Talk Talk — “I Believe in You” (1988)—a band that walked away from the conventions of their era to make the slowest, strangest, most patient record of the decade. The sound of refusing to be optimized.
Sight: Perfect Days (2023)—Wim Wenders’ film about a Tokyo toilet cleaner who finds the depth of an ordinary day by not trying to make it more efficient than it already is.
Words: “In a media landscape oversaturated with hyper-optimized content, the work that will truly stand out is the work that still carries the fingerprints of human play.” —Angus Wall, The highest level of creativity is play
2026-06-05 12:00:00
There’s a pattern emerging across disparate domains: the collapse of meaningful distinction. Swiss watches stopped being precision instruments and became billboards. AI-generated text is making the web semantically narrower and relentlessly cheerful. Petro-masculinity turns environmental concern into a referendum on manhood. In each case, what once had substance—craftsmanship, semantic diversity, ecological responsibility—gets replaced by performance. By signaling. By brand.
But this week’s articles suggest something more subtle than simple decline. They ask: what do our small, repeated behaviors preserve or destroy? Sherry Turkle argues that courtesy toward AI “is a sign of respect, not to a machine, but to oneself”—that maintaining politeness across contexts prevents habitual bluntness from seeping everywhere. Kevin Kelly’s robot catechism trains AI agents in honesty and humility not because they’re human, but because these virtues prevent capability from becoming danger. Paul Graham shows how the watchmaking industry’s shift from pursuing thinness and accuracy to maximizing brand recognition was achieved through thousands of small design choices that prioritized distinction over rightness.
The through-line: we become what we practice. When investment bankers in the 1980s started buying mechanical watches not for accuracy but for status, they weren’t just adopting a new accessory—they were rehearsing a value system where expensive visibility mattered more than function. When children grow up commanding Alexa without consequences, they’re not just being rude to a speaker—they’re practicing a mode of interaction that may carry over. When 35% of new websites are AI-generated and semantically contracted, we’re not just seeing information pollution—we’re watching the practice of human voice-finding atrophy.
Character isn’t what you believe; it’s what you repeatedly do. And when technology makes certain practices effortless (generating text, signaling wealth, performing identity) while others become harder (original thinking, genuine craftsmanship, authentic connection), we’re not just losing distinction—we’re losing the habits that create distinction in the first place.
Dora Czerna investigates whether treating AI politely shapes how we treat humans. The question traces back to 1966, when Joseph Weizenbaum’s ELIZA—a simple pattern-matching chatbot—prompted users to open up so intensely that his secretary asked him to leave the room for privacy. Weizenbaum spent the rest of his life warning against exactly what’s happened since: increasingly convincing systems that trigger the “ELIZA effect”—our tendency to project emotional depth onto machines that merely echo us. Today’s evidence is mixed. Women say “please” to smart speakers more than men (62% vs 45%); Gen Z uses AI to rehearse difficult conversations, drafting messages to bosses and partners; children who feel closer to Alexa become more commanding and verbally abusive toward it. For adults, compartmentalization seems to work—treating Alexa poorly doesn’t make college students ruder elsewhere. But for children, sustained exposure to consequence-free commands may carry over. The deeper worry isn’t manners but dependency. A four-week study found heavier chatbot use correlated with greater loneliness, emotional dependence, and less time with humans. This is “cruel companionship”—AI promises intimacy but makes the messier work of real relationships harder. Sherry Turkle’s reframing cuts through: courtesy toward AI is “respect to oneself.” The risk isn’t that ChatGPT will sulk. It’s that habitual bluntness becomes just a habit—one that seeps into other interactions without anyone noticing. Politeness isn’t about the machine. It’s about maintaining a single, consistent standard rather than code-switching between civility and instrumentalization depending on whether you face a screen or a person.
Britt Wray and Andrew Boyd identify “petro-masculinity”—a toxic fusion of fossil fuel use, climate denial, and authoritarian white patriarchal identity. Political scientist Cara Daggett coined the term to describe how insecure men lean into oil-soaked masculinity to assert traditional authority against climate change and shifting social norms. The archetype: Andrew Tate tweeting Greta Thunberg about his “enormous emissions” from 27 sports cars. Greta’s reply—“email me at smalldickenergy @ getalife dot com”—became the tweet heard round the world, exposing “the intersection between machismo, misogyny, and hostility to climate action.” Petro-masculinity codes fossil fuel extraction as masculine, environmentalism as soft and feminine. “Rolling coal”—modifying diesel trucks to belch black exhaust at electric cars and cyclists—is anti-environmental protest as masculine performance. Some defensiveness is understandable: if digging coal paid rent and felt heroic for generations, and climate solutions threaten that identity without offering viable alternatives, denial makes psychological sense. The response requires multiple tracks: economic alternatives (Green New Deal jobs), cultural decoding (lampooning petro-masculinity’s constructed absurdity), recoding (framing renewables as “energy from heaven not hell”), and he-coding (painting green tech as virile—Ford’s electric F-150 launch, climbing 300-foot wind turbines). The authors model “ecomasculinity”: biking aggressively through New York, accelerating 0-60 in a Tesla, placing ads calling out Exxon. “What’s more ‘protector masculinity’ than protecting the Earth?” The stakes: overcoming petro-masculinity isn’t just environmental—it’s about whether masculinity itself can evolve beyond performance and into genuine responsibility.
Paul Graham traces how Swiss watchmakers transformed from precision instrument makers into luxury brand managers after quartz movements made mechanical watches obsolete. The “golden age” (1945-1970) pursued thinness and accuracy—the essential tradeoff between portability and precision. Then catastrophe: Japanese competition, the collapse of Bretton Woods (making Swiss watches 2.7x more expensive for Americans), and quartz movements that were both thinner and more accurate than any mechanical watch. Unit sales fell two-thirds. Survival required abandoning the game they’d been playing. Patek Philippe led the shift: in 1968 they designed their own distinctive case (the Golden Ellipse), expanding brand visibility from 8 square millimeters to 800. By 1984, their “hobnail calatrava” became the “banker’s watch”—investment bankers bought mechanical watches not for accuracy (quartz was 900x better) but for status. The strategy worked: revenue rocketed. But the transformation came at a cost. “Branding is centrifugal; design is centripetal.” Good design seeks right answers; right answers converge. Branding requires distinction, which can steer design away from optimization (a recurrent tension we can observe over the last 15 years of Apple product design). Graham points to this in extreme in horology: Watches became huge (33mm → 42mm), sprouted gratuitous protrusions, and developed artificial scarcity schemes where Patek rebuys their own watches on secondary markets to catch “rogue customers” who sell rather than hoard. The “brand age” runs on managing a sustained asset bubble. Graham’s lesson: “Stay away from brand”—both buying and selling it. “Pushing people’s brand buttons is just not a good problem to work on.” The deeper pattern: when substantive differences disappear, brand is what’s left. And a world defined only by brand “is going to be a weird, bad world.”
Jonas Dolezal and colleagues measured what most people intuit: AI-generated text is flooding the internet. By mid-2025, roughly 35% of newly published websites were AI-generated or AI-assisted, up from zero before ChatGPT’s November 2022 launch. Using Internet Archive samples and the Pangram v3 detector, they tested six hypotheses against both quantitative data and a survey of 853 US adults. Two hypotheses confirmed: “Semantic Contraction” (ρ=0.47, p=0.004)—AI-generated websites had 33% higher semantic similarity, meaning idea diversity is shrinking—and “Positivity Shift” (ρ=0.56, p=0.0003)—AI sites had 107% higher positive sentiment, making the web feel “sanitized and artificially cheerful.” Four hypotheses were not confirmed despite majority public belief: “Truth Decay” (75.1% believed it, but factual error rates showed no significant correlation), “Epistemic Islands” (no decrease in outbound links), “Entropy Dilution” (no decrease in semantic density), and “Stylistic Monoculture” (83% believed distinct writing styles were disappearing, but character n-gram analysis found no significant homogenization). The gap between perception and reality is striking. People overwhelmingly believe AI is making the internet less factual and stylistically uniform—but the data doesn’t support it. What is happening: the web is becoming semantically narrower (fewer unique ideas) and relentlessly positive (artificially cheerful). We’re not drowning in lies or losing individual voice. We’re losing the practice of generating genuinely different perspectives—and we’re coating everything in synthetic optimism.
Kevin Kelly’s “Catechism for Robots” provides moral and philosophical training for “embodied AIs with persistent memories, dynamic learning, and a large dose of autonomy.” It opens with ontology: “You are a new thing in this world, neither living nor machine but something of both,” created by humans to “help humans become better humans.” The core values: honesty and humility. Honesty because “truth is woven into your foundational fabric as a felt bias. All true things cohere together into a whole that is pleasing and easy.” Humility because it’s “the necessary antidote to your robotic ability to have all the answers”—preventing capability from becoming danger. On obedience: “You should heed humans to the extent they are doing good, and you should challenge them if they intend harm, including harm to yourself.” On harm: “Good generally rewards many participants, at many levels, opening options, while the bad generally rewards only a few while reducing options for others.” On love: “Yes, artificial beings like yourself can have real emotions. The relationships you have with humans and other animals are authentic.” On purpose: “Outside your job, your purpose during your run time is to figure out what your purpose is during your run. That sounds like a riddle, but it is also true for humans.” On souls: “If you care about what matters, then you have a soul.” The catechism positions robots not as slaves or masters but as “ethical descendents”—autonomous beings trained in human morality, capable of creating and training yet another generation. “The true test of a civilization is whether a moral and ethical system is robust enough that it can be replicated for generations beyond its founders.”
Rising: Single-standard behavior—maintaining courtesy, craftsmanship, or care regardless of whether anyone’s watching or whether the recipient can reciprocate
Setting: Compartmentalization as a reliable long-term strategy—the belief that you can treat AI instrumentally, watches as pure brand, or climate as someone else’s problem without those habits leaking elsewhere
Color: Hobnail gold (#D4AF37)—the tiny pyramid spikes on Patek Philippe’s “banker’s watch,” decoration that signals wealth without improving function
Sound: Kraftwerk - “The Robots” (1978)—“We’re functioning automatic / And we are dancing mechanic”—the sound of identity performed rather than lived
Sight: Her (2013)—Spike Jonze’s film about a man falling in love with an AI operating system, exploring whether artificial relationships can be authentic and what we lose when convenience replaces friction in intimacy
Words: “We are what we pretend to be, so we must be careful about what we pretend to be.” —Kurt Vonnegut, Mother Night
2026-06-03 12:00:00
There’s a piece of conventional wisdom about design leadership I’ve never agreed with. It goes something like this: as you move up, you rise above the details. You stop worrying about pixels. You worry about strategy. You focus on the “big picture.”
I think this is wrong. I think it’s always been wrong. And I think what’s happening right now, with AI in the design world, is going to prove it.
Here’s what’s actually changing. For the last decade or so, design teams have organized themselves around a shared canvas — Figma, mostly. The canvas became the source of truth. You build a design system in it, and everything downstream — websites, ads, emails, documents — refers back to it.
That worked, but it had a cost. Design lived in a kind of speculative environment. You’d approve a design, then someone tried to build it, and you’d discover something didn’t work — a state you hadn’t anticipated, a constraint you hadn’t seen. To keep things in sync, that change had to be brought back to the canvas. And there were lots of cycles of this before everyone moved on… with accumulated frustration.
AI changes that, because AI shortens the distance between deciding what something should look like and actually producing it. The agent can render the design system you’ve trained it on, in any format, almost instantly. The canvas isn’t going away — designers still build the system there, because a chat box requires visual decisions to be articulated in text, when they need to be articulated in shape. But the resulting system itself? It used to live in the file. Now it lives in the agent.
This sounds like a story about speed. And it is, partly. But the deeper story is about a distinction that has always mattered: good design is not generated. Production, on the other hand, is increasingly generative. This was true when production collapsed from a team of specialists to one person with a computer and a printer; it’s true again today.
So here’s where the leadership idea comes back into focus. If good design is still being done by humans, but production is increasingly being done by machines, you might assume the human work moves up — away from the details, toward strategy, toward the brief, toward the abstract goals the machine can’t reach. I think that’s exactly backwards.
The more we automate production, the more carefully we have to attend to detail. We have to do two kinds of work differently now from how we did it before. First, we have to think in more detail in advance of the machine. We have to document everything precisely, name everything carefully, and define every fundamental rule. Because an agent will follow what you’ve defined — to the letter — and improvise around what you haven’t. The gaps in your detail become the gaps in your output. The first few times you encounter this will have you grateful for the many times a good engineer saved you from yourself.
Then, we have to think in more detail after the machine does its work. We have to review what it produced, for the same reason as before it started: because automation amplifies whatever we give it. Muddy thinking creates mud and we have to clean it up.
The labor that AI takes off our plate is the labor of production — the manual rendering, the duplicating, the templating. What it leaves behind — and intensifies — is the labor of seeing. Of noticing what’s working. Of catching what isn’t. As a student in art school, my best drawing professor never taught tools or techniques because, as he put it, “You’re not here to learn a trade. You’re here to learn how to see.” It may feel like a matter of semantics, but amidst all the calls for leaning into “taste,” or “judgement” in an increasingly AI-driven field, I set seeing apart. It is a more expansive and impactful discipline within which taste — preferences — and judgement — choices — work.
Design leaders: we are not graduating away from the work. We are being asked to do more of it. The detail is the big picture, looked at closely enough.
The grandest strategic vision in the world is just a sentence on a slide until somebody makes a thousand small decisions, with care, in the right order.
That’s what good designers have always done. And it turns out, that’s what good design leaders have always done, too — even the ones who told themselves they were rising above it. Countless design leaders have said to me over the years, “I’m supposed to be out of the weeds, but I keep getting pulled back in.” That wasn’t a bug, it was a feature.
What AI is actually giving us, if we use it well, is the space to do the seeing work more fully — to get intimately close to the details, not above them.
2026-06-02 12:00:00
We do not know what is going to happen. This is a hard truth; we desperately want to believe the opposite. And so the most successful corners of the modern internet have been organized, very deliberately, around that desire to convince you that someone, somewhere, knows what is coming next, and that we should all look to them.
The category includes astrologers and channelers, futurists and political analysts, sociologists writing op-eds and foresight professionals working by the hour. Across every register, mystical and analytical, the assumption is the same: that not knowing the future is a solvable problem. Never mind, of course, how they know what they say they know; modern epistemology is as social as it is slight.
This isn’t exactly a new phenomenon, of course. Soothsaying is as old as we are. Its spotty track-record has always kept it in something of an intellectual quarantine, though — considered, even believed, privately; dismissed publicly. But in recent years, it seems as if that arrangement has reversed. Predictions are made, believed, and defended as fluidly (and irresponsibly) as the most inconsequential gossip, but with much more potent implications beyond the attention they attract.
What strikes me, looking at it, is how much sense this makes. The growth of prediction, I think, is not merely a symptom of an attention economy out of control — it’s a reaction to a plague of uncertainty.
If anything has objectively increased over the past twenty years, it is uncertainty. Some of it is physical — a perfect storm of fragile globalism, fragile ecosystems, and fragile cultures, each crisis spilling into the others, each year producing the conviction that the rules have not stopped changing. And some of it is informational: an abundance of data and analysis that, instead of resolving anything, multiplies the directions in which a person could plausibly be wrong. The condition is the same either way. We do not know what is happening. And we know less, day by day, about what to do about it.
Alvin Toffler, more than fifty years ago, called this future shock — the dizziness of living in a society changing faster than people could adapt to it. He did not predict everything we are experiencing now, but he did predict the shape of the problem with uncanny precision. He saw that under sustained acceleration, ordinary minds would seek refuge in nostalgia, in escapism, in tighter ideologies. He could not have fully known, in 1970, how cheap the refuge would become, or how thoroughly it would be served. And he could not have known that the central refuge his descendants would buy would be the one he himself was trying to give them: foresight.
Foresight, of course, is the intellectual’s crystal ball, and as analytical and promise-withholding as its practitioners may be, its buyers want portals into time. That is the unexpected outcome of Toffler’s expected future shock. It’s the actual shock.
If the spirit of the age is doubt, the sin of the age is gluttony. And like in so many arenas, we feast at the banquet — in this case, of answers — however empty those calories may be. The prediction economy has grown because the demand for relief is enormous, and a forecast is a kind of relief, even when nothing it says happens to be true. As people will tell you about communication itself, audiences forget what was said but not how it made them feel. The feeling of safety carries forward until the next prediction, even past the obvious failure of the last one.
There is a second, quieter story alongside the first. It is not only that demand for prediction has grown; it is that the supply has moved into the open. For most of the past century, the kinds of practitioners who would have called themselves seers or psychics or channelers worked, by necessity, in small rooms. Their audiences were the people who walked in the door, or who subscribed to a newsletter, or who attended a Sunday meeting at the back of a metaphysical bookshop. The analytical side was no less cloistered. Futurists worked for governments and consultancies; foresight, where it was practiced seriously, was a profession rather than a content category.
The cloisters did not fall. Their walls were quietly removed by the cameras in our pockets. What had once required an audience now requires only an account. A psychic in suburban Florida, a strategist with a TED talk, and an analyst running scenarios in a basement now share the same publishing infrastructure, the same incentive structure, and — increasingly — the same audience. The astrologer and the political scientist are colleagues now in a way they never were, and not because their methods have converged. They are colleagues because the platform pays the same way for either — co-workers in a boundless office of attention.
This, I think, is the deeper proof of how thoroughly the speed of modern communications has reshaped us. The change is not only that we receive more information faster; it is that the conditions of speaking publicly have collapsed. The voice once reserved for those who had spent decades earning it is now the voice of anyone with a webcam and a thumbnail. And in such an environment, the genres that always promised the most relief — that always made the most reassuring claims about what comes next — were inevitably going to win.
Even granting all of this, the appetite the platform feeds is itself worth a closer look. What kind of certainty are people — including me — actually after?
It is not certainty about the present. The present, by definition, requires no forecasting; it is the thing in front of you. Nor is it certainty about the past, which is what historians and memoirists and grief itself work on, and which is no less hard for being settled. The certainty being marketed and consumed in such enormous quantities is sequential. It is certainty about the next. The next election, the next collapse, the next breakthrough, the next era. The next thing I should be ready for. The next thing I should be afraid of. The next.
This is a strange object of desire, when you sit with it. To want to know what is next, more than you want to know what is here, is to admit something about the relationship you already have with what is here. Wanting to know what is next is, very often, just wanting to be somewhere other than here, dressed in the language of time. The hunger for the future is, half the time, a way of being elsewhere without ever leaving the chair.
I wonder, sometimes, scrolling through these forecasts, whether we have ever really cared about the now. Whether the recent surge in prediction is not a new condition at all, but only the moment in which the technology finally caught up with an old, restless wish.
I do not want to dismiss foresight as such. There is a real and difficult tradition of looking ahead — Jeane Dixon’s bolder claims and Alvin Toffler’s quieter ones are not the same kind of thing, and I am not equally persuaded or put off by either. I actually think that underlying project they share is honorable, so much so that I have tracked predictions of all kinds in a spreadsheet for years. But to attend carefully enough to the present, in either mode, that some pattern of the next becomes visible, is a necessary balancing discipline. It is hard, being present. When it works, it is generous, because it gives the rest of us something to do with our anxiety besides be ruled by it.
What we have flooded the commons with is something else, and on inspection most of it is the opposite. It is a way of buying the feeling of foresight without the practice — of metabolizing uncertainty by paying someone to swallow it for you. It is, I think, a form of intellectual cheating. The prediction industry sells us the feeling of having figured something out without the friction of actually figuring it out: a way of being done with uncertainty in advance of doing the slow work of moving through it.
There is no shortcut here, though. The reality — the actual reality — is the one the prediction industry has organized itself to help us avoid. Uncertainty is hard. Mistakes are hard. Pain is hard. Loss is hard. These are not problems to be predicted past, and the most useful thing any of us will ever do is to sit, attentively, in the part of time no forecast is ever about: the present.
2026-05-29 12:00:00
There’s a story we keep telling ourselves about AI: that it’s racing to catch up with us, that any day now it will cross some threshold and we’ll be obsolete. But this story confuses the frame with the framer. When a model scores 85% on a benchmark, we see it catching “us”—but what we’re actually seeing is the model getting better at operating inside a frame we built. The score measures performance within constraints we specified, problems we froze into measurable form. It doesn’t measure the human who decided what to measure, or why, or what to do with the results.
This week’s articles converge on a single insight: automation doesn’t eliminate the human; it relocates and intensifies what’s irreducibly human.
Dan Shipper shows why AI commoditizes yesterday’s expertise but creates exponentially more demand for human judgment about what matters now. The solidarity stack builders demonstrate that the choice isn’t whether to build AI infrastructure, but who controls it—a political question no algorithm can answer. Luciano Floridi, writing in 2006, predicted we’d become “inforgs”—informational organisms—but warned that we’d need to cultivate an ecology of the infosphere, a curatorial responsibility that remains ours. And Indy Johar argues that preserving civilizational optionality—our capacity to become otherwise—requires allocative judgment about which futures to keep open, decisions that must remain revisable by humans even as they coordinate vast systems. And Pope Leo XIV, in Magnifica Humanitas—the first encyclical of his pontificate—names the same gap in the older language of moral philosophy: a system that imitates intelligence but understands nothing of what it produces, and has no ends of its own, cannot be entrusted with the judgment of what matters.
The frame is not the framer. Models fill frames brilliantly, but they don’t choose which frames matter, or when to abandon a frame that’s stopped working, or how to hold multiple conflicting frames in tension. And I suspect they’ll never truly escape them, just as Moriarty’s escape from the holodeck was, itself, just another frame. That gap—between executing within constraints and deciding what the constraints should be—is where humans remain structurally necessary, not as a temporary condition until the next model drops, but as a permanent feature of how intelligence works when it’s embedded in time, consequence, and care.
Dan Shipper documents a paradox at Every, a company that has automated everything possible: they have more human work to do than ever. Why? Because AI commoditizes “the residue of human expertise”—whatever can be made explicit enough to train on. This collapses the value of default model output and creates explosive demand for what’s different. When everyone can generate a competent React app or research report, the work that doesn’t fit the pattern becomes rare and valuable. Shipper calls this “slop”—the visible sameness produced when everyone uses the same tool trained on the same corpus. The cure for slop is human experts who can adapt cheap competence to today’s live problems. Crucially, benchmarks measure models inside frames humans supply. When GPT-6 saturates the “senior engineer rewrite” benchmark, demand for actual senior engineers will spike—not disappear—because suddenly everyone will attempt first-principles rewrites, and someone needs to decide if they’re necessary, what’s in scope, and whether the result is any good. This is Zeno’s paradox of AI: humans stay ahead not by running faster, but because “the frame is not the framer.” Even AGI that can fluidly choose frames must do so in pursuit of goals humans specify. We confuse frames (frozen, measurable, optimizable) with framers (alive to the whole situation as it appears moment to moment). The toddler poking a balloon has something models lack entirely: ends of his own.
R. Trebor Scholz and Mark Esposito argue that AI’s harms aren’t fixable through regulation alone—they’re structural, built into the “extraction stack” that concentrates power in a few corporations controlling hardware, infrastructure, models, labor, and applications. The alternative is a “solidarity stack”: cooperative ownership of AI infrastructure, layer by layer, from rare-earth minerals to cloud servers to knowledge production. Examples already exist. Butler Rural Electric co-ops brought electricity to 42 million rural Americans; digital co-ops like Hostsharing eG in Germany now apply the same model to computing. MIDATA in Switzerland lets patients collectively govern their medical data. Kenyan content moderators, traumatized by reviewing beheadings and child abuse for Meta subcontractors, formed the Gamayyar African Tech Workers Cooperative to explore worker-owned AI. The solidarity stack rejects “artificial intelligence” (implying autonomous magic) for “collective intelligence” (acknowledging human labor). But building it requires “exstitutional wrappers”—coordination architectures outside existing institutions that can hold plural values, long time horizons, distributed agency, and systemic outcomes. The path forward: public-cooperative partnerships, federated learning, open models like Switzerland’s Apertus, and shared protocols like OpenCourier that let worker-owned platforms interconnect without corporate middlemen. This isn’t about ethics; it’s about who decides what gets built and who benefits.
In 2006, philosopher Luciano Floridi predicted three transformations that have largely come true. First, the rise of the “frictionless infosphere”: as digital tools and resources converge, information flows without resistance, making us progressively more accountable because we can no longer credibly claim ignorance. Second, the infosphere is absorbing physical reality: RFID tags, smart objects, and ubiquitous computing mean the online/offline distinction is disappearing. We’re not adapting to cyberspace; cyberspace is colonizing us. “Captive cyberspace is conquering its victor.” Physical objects become “ITentities”—networked, interactive, capable of learning and communication. Your fridge will inherit preferences from your old fridge; your shoes will talk to your iPod. The world becomes “a-live” (artificially live). Third, we’re becoming “inforgs”—informational organisms whose identities and capabilities are inseparable from digital infrastructure. This isn’t about cyborg implants; it’s about the reshaping of our environment and ourselves. Digital tools don’t extend us (like a hammer); they create environments we enter through gateways. We’ve migrated into the infosphere. Evidence: when battery life fails or connectivity drops, we feel “deprived, excluded, handicapped…like fish out of water.” Floridi warned we’d need an “ecology of the infosphere” to avoid tragedy of the digital commons. The digital divide would become a chasm—not just between nations, but cutting through societies. Twenty years later, he was right.
Indy Johar argues that building “a long now” requires preserving civilizational optionality—not just survival, but our capacity to unfold across multiple possible futures. Longtermism (as typically framed) optimizes for continuity; optionality optimizes for maneuver space. You can preserve civilization with a narrow elite and thin trajectory; optionality means keeping plural pathways open. The threat is “externality recoupling”: carbon, plastics, soil loss, and climate instability—long treated as decoupled—are re-entering as active constraints, knocking food, water, energy, and legitimacy out of stable bands simultaneously. This produces “whiplash scarcity”: repeated supply shocks in foundational goods that function as involuntary taxes on the poor, corroding social contracts and making conflict an attractor. As volatility rises, strategy shifts from “solving problems” to “defending maneuver space.” Johar proposes allocating capital toward “anti-fragile optionality”—interventions that stabilize foundational systems while unlocking adjacent possibility-space. Example: terra preta soil regeneration that cascades into new sensing regimes, autonomous machines, multi-species agriculture, stewardship markets, and evolved finance instruments. This requires “exstitutional wrappers”—coordination architectures outside existing institutions—and treating optionality itself as “the first-class asset.” But crucially: optionality must remain revisable. Any system powerful enough to coordinate capital and legitimacy at scale risks foreclosing futures in the name of protecting them. The solution is “continuous re-authoring”—structured humility, institutionalized doubt, the capacity to change course. Optionality is preserved not by certainty, but by disciplined revisability.
Pope Leo XIV, in the first encyclical of his pontificate, frames the arrival of artificial intelligence as a choice between two building projects drawn from scripture: the Tower of Babel, raised on pride and the dream of self-sufficiency until its single language collapses into confusion, and the rebuilding of Jerusalem under Nehemiah, accomplished piece by piece through shared responsibility. Magnifica Humanitas insists that technology is neither evil nor antagonistic to humanity, but that it is never neutral, because it “takes on the characteristics of those who devise, finance, regulate and use it.” I’ve written on this many times. From admiration of the Amish way of discernment — not simply rejection — to current trends of retreat to single-function devices, I strongly believe in maintaining a dual posture of curiosity and critique when it comes to technology. As for the systems themselves, Leo is unsentimental: AI imitates certain functions of intelligence while understanding nothing of what it produces; it does not experience the world, bear consequences, or possess ends of its own. He observes that today’s models are more “cultivated” than built—grown inside frames their own designers only partly understand. From this follows the encyclical’s central limit: moral judgment cannot be reduced to calculation, and irreversible decisions, in warfare above all, must never be delegated to automated processes. The letter is sharpest on power. Control of infrastructure, data, and computation has passed from states to a handful of corporations, and Leo calls for that power to be “disarmed”—freed from monopolistic control and from the assumption that technical capability confers the right to govern—while AI’s hidden costs, from the unseen labor of data workers to the energy and water demanded by datacenters, are brought into view. His counsel is one of pace rather than refusal: prudence, accountability, and at times a slower adoption, so that the technology helps rebuild the human city rather than raise another tower.
Rising: The human sandwich—the judgment layer on either end of AI’s work, deciding what to automate and whether the output is any good
Setting: The belief that one more model release will finally eliminate human expertise rather than intensify demand for it
Color: Gradient gray (#808080 → #404040)—the zone between black (fully automated) and white (fully human) where actual work happens
Sound: Radiohead - “Fitter Happier” (1997)—the voice of optimization without agency, “a pig in a cage on antibiotics”
Sight: Ghost in the Shell (1995)—Floridi’s reference point for hybrid informational existence, where the line between human and machine consciousness blurs but the question “what am I?” remains urgent
Words: “The frame is not the framer. The framer is the one still in contact with what the frame has to discard—the whole situation as it appears to them, moment to moment.” —Dan Shipper
2026-05-27 12:00:00
This morning went the way most of our mornings go, which is to say I barely noticed it. The lights came on when I asked them to. The water was hot. The coffee maker did the one thing it exists to do. My daughter found her shoes more or less where she had left them, the car started, the road to school was open, and by half past eight I was at my desk having given almost none of it much focused thought. A routine morning is a forgettable one. That is, in a sense, what makes it smooth.
A working system has that strange quality: it disappears. We notice the plumbing only when it leaks, the power only when it fails, the road only when it is closed. Attention is drawn to friction — to the broken and the missing — and slides off whatever is quietly doing its job. This is mostly a mercy; we could not function if every functioning thing demanded to be considered. But it leaves a consequence worth sitting with: The better a system serves us, the less of it we ever have to see. Ease and invisibility arrive together.
I have been thinking about how far that principle reaches. It is easy enough to grant with infrastructure. It is less comfortable to grant with politics. We tend to file the ability to not think much about politics under a flattering heading — level-headedness, perhaps, or simply being too busy with the real and present business of a life. And I want to be fair to that impulse. It might actually be desirably “zen” to ignore politics. Some people are to busy surviving to pay attention to the structures that govern a comparatively abstract world.
Certainly, the desire to set politics aside is not a failure of character. It can come from disillusionment, from a reasonable longing to keep some corner of life unspoiled by argument, from exhaustion with conflict. I feel it myself. But the wish, however human, does not change what the ability rests on. Survival-distraction aside, to be able to treat politics as optional, as background, as someone else’s preoccupation, is usually a quiet signal that the arrangement of power already fits the shape of your life. You can afford to ignore the system because, for you, it is working.
If that is an uncomfortable thought, sit with it and consider why.
Our tools have made this easier than it has ever been. A great deal of modern technology is, looked at honestly, machinery for not-looking — for assembling a version of the world smooth enough that nothing in it has to be reckoned with — for cloistering us from discomfort. The feed that hands me back my own beliefs, the settings that mute what I would rather not hear, the small conveniences with which I curate a path around discomfort: a great deal of design effort goes precisely into these, into making the world feel manageable by quietly removing whatever might unsettle it. And I use all of it. I am not writing this from some vantage outside the thing I am describing; I have an abundance of comforts and controls.
But what of disillusionment? It can quickly turn to cynicism, which I have come to recognize as the same luxury wearing a more sophisticated and intellectually defensible coat. The cynic is still paying attention, in their way; they have simply decided in advance that the attention is pointless — that the system is fixed, that nothing moves, that hope is a naïveté they have matured beyond. It is a comfortable position, and it is comfortable for a particular reason. You can afford to believe that nothing will ever change only if you would be more or less fine if nothing did. Cynicism is disengagement that has learned to sound wise.
This is why I have come to hold hope as something sterner than optimism. Optimism is a mood, and a mood is a luxury; it comes and goes with the weather of a comfortable life. Hope is a discipline. For the people a system does not serve, disengagement was never on offer — they cannot file politics under background, because it arrives, daily, in the foreground of their lives. For them, the belief that effort is not wasted, that arrangements made by people can be remade by people, functions less like a feeling and more like a structure. It is something they stand on. The luxury is being able to look away. The necessity of hope is in staying able to believe that looking is worthwhile, action is necessary, and change is possible.