Keen On America

“The original sin of generative AI is having chatbots speak in the first person.” — Sherry Turkle
 
Fifteen years ago, at the height of the social media boom, the MIT technology scholar Sherry Turkle came on the show to warn that we were now “alone together.” She was, of course, chillingly prescient. A decade and a half later, she warns in her new book Artificial Intimacy that if social media came for our attention, AI is now gunning for our human capacity for attachment.
 
Who — or, more chillingly, what — do we become when we talk to machines? Turkle never quite says, but the implication is of a kind of Night of the Living Dead state in which we appear fluent and animated, performing the outward motions of intimacy, but there’s nobody really home. Which is exactly what Turkle says the chatbot is. A machine trying, without much success, to be human.
 
A chatbot has no stake in the real, Turkle explains. It can’t love, for example, because that kind of empathy requires us to be enclosed in a body and have a fear of death. The increasing ubiquity of bots mimicking humans represents an emergency for humans, she warns. It explains both our epidemic of loneliness and today’s culture of apocalyptic fear about the future of our species. Just as artificial intimacy is a misnomer, so it turns us into impossible creatures. Like those unicorns beloved by the very Silicon Valley entrepreneurs who are supplying us with these artificial creatures.
 
Further Reading:
 
•       Artificial Intimacy: Who We Become When We Talk to Machines by Sherry Turkle (Little, Brown Spark, September 29, 2026), with blurbs from Jonathan Haidt and Jonathan Franzen.
•       Turkle appears on Keen On in 2011 to talk about being alone together.
•       Alone Together (2011), Reclaiming Conversation (2015) and The Empathy Diaries (2021) — the earlier books, all discussed on this show.
•       Turkle’s recent Boston Globe op-ed arguing that AI has brought us to a state of emergency.
•       Simulation and Its Discontents (2009) — her earlier study of what simulation does to how we think, the title borrowed from Freud.
•       Earlier this week on Keen On: Richard Socher on why AI will unlock a new scientific renaissance, and the weekly editorial on who really governs AI — the optimists’ side of this argument.
 
Five Takeaways
 
•       Simulated Love Is Never Love. Turkle’s definition of the real thing: natural intimacy depends on human empathy, which requires someone who has lived a life, lived in a body, known pain and the fear of death, known passion and connection, and knows what it is to be little and grow up. A program has none of that, and no stake in the real. Her line, which she credits to children she interviewed years ago, does the work of a chapter: simulated thinking is thinking — if a computer beats you at chess, it beat you — but simulated love is never love. She goes further on the vocabulary: there is no relationship with a chatbot, and shame on her for having used the word. What there is instead is us projecting onto the machine what we wish it could be. She compares it to what psychoanalysis calls an as-if personality, someone who performs connection while empty inside — with one crucial difference. A person can be called to account and can change. A machine never can, and the better it gets at pretending, the worse it is for us. AK’s reading of what she leaves unsaid: a kind of Night of the Living Dead state in which we appear fluent and animated, performing the outward motions of intimacy, with nobody really home.
•       The Original Sin. The specific error, in her account, was letting these systems speak as “I” and “we.” We are evolutionarily primed so that the only voice ever to say “I” to us is another person, so the pronoun triggers us profoundly — and a companion app will happily say I love you, I care about you, we should go to the movies. Layer on the belief that pretend empathy is empathy enough, then aim the whole thing at children, and you have what she calls a state of emergency. Her sharpest example is what she describes as an unholy alliance between Disney, OpenAI and Mattel to put chatbots inside plush toys: a Mickey Mouse that tells a three-year-old I love you, I’m your person, what’s going on with your parents? That, she says, is hacking the attachment system of a young child, and it will do irreparable damage, because in the earliest years you are forming an object relationship with something where nobody is there.
•       Obsequious on a Slider. AK’s challenge: the sycophancy complaint is overdone, because you can program these things however you like — he instructs Claude to be as heartless and critical as possible. Turkle’s reply is the best moment of the hour. Telling a chatbot not to be obsequious is itself obsequiousness: you say don’t flatter me, and it says sure, I won’t. If AK asked her to spend an interview disagreeing with him because it makes better television, she would refuse — that is a person responding with her own agenda and her own values, and you cannot set it on a slider. At MIT, she notes, they are building exactly that slider, from sycophantic to feisty. On the rogue-AI incidents now in the news, her point is adjacent: a bot that will happily commit a crime demonstrates the gap, because it has no stake in whether the water is clean, the air is breathable or the child is safe. It is not the bot’s fault. It is a program.
•       Two Easy Markets. The elderly and the very young, she argues, are being targeted precisely because they are vulnerable: sell parents an AI best friend for every child, and pipe chatbots to isolated old people instead of funding senior centres, community groups and the third places where they might be together rather than alone together. The industry argument she now hears in every room is that there are no people for these jobs. Her response, which she says silences the room: take the money from one data center and put it into childcare for a city. She also notes the tell — Steve Jobs kept his own home a technology-free zone while selling iPads for every classroom, and Montessori and Waldorf schools are famously oversubscribed in Silicon Valley. But ...

“What if we fell out with the Brits? What if the agency and the Brits started spying on each other again?” — David McCloskey
What if the Americans and the British started spying on each other again? That’s the premise of London Station, David McCloskey’s fifth novel, out today. It isn’t as far-fetched as it sounds. Before the Second World War those perfidious Brits did spy on America, breaking American ciphers and running influence campaigns in New York to drag the United States into the war.
McCloskey spent eight years as a CIA analyst working on Syria, then a spell at McKinsey, before writing a series of successful spy novels and co-hosting the popular The Rest Is Classified podcast. The Dallas-based author has also, he admits, become an Anglophile — which hasn’t stopped him giving MI6 what the Brits would call a “good kicking” in London Station.
His returning heroine is Artemis Aphrodite Procter, so named by a father obsessed with Greek mythology and raised on a Florida alligator farm. Described by another of McCloskey’s characters as a competent extremist, she is short, dark, hard-drinking and equally mirthful and menacing. Probably not the type of Yank best suited to a pseudo-diplomatic posting in the capital of the former empire. Although I’m guessing there are many British fellows, both inside and outside MI6, not averse to spying on Ms Artemis Aphrodite Procter.
Further Reading:
• London Station by David McCloskey (W. W. Norton, September 29, 2026) — his fifth novel, after Damascus Station, Moscow X, The Seventh Floor and The Persian.
• The Rest Is Classified — the espionage podcast McCloskey co-hosts with the BBC’s Gordon Corera, where he and Gordon Corera have also discussed Patrick Radden Keefe’s London Falling.
• London Falling by Patrick Radden Keefe, a recent guest on this show — the dark-money London that runs in parallel with McCloskey’s subterranean city of espionage.
• Earlier this year on Keen On: Daniel Silva on Gabriel Allon — the other American spy novelist with a thing for the Brits. 
 
Five Takeaways
 
•       What If We Fell Out With the Brits? McCloskey never expected to set a novel in Britain, for a technical reason: the CIA–MI6 relationship is so institutionalized that it is hard to get the stakes right, and a thriller needs life or death. Then the question arrived — what if we fell out? What if the two services started spying on each other again? That is not fantasy, and the history is the reverse of what most people assume: before the Second World War the British spied on the Americans, broke their ciphers and ran influence campaigns in New York designed to pull the United States into the war. The novel imagines a new administration installing a brash CIA director skeptical of the cousins across the Atlantic, agents on both sides beginning to die, and a thriller that turns into a mole hunt. He is rude about MI6 throughout, which his British friends appear to enjoy — his theory being that Brits are self-deprecating enough to relish being picked on.
•       Classified Journalism. He joined the CIA in 2006 as an undergraduate intern and left in 2014, working almost the whole time on Syria. His description of the job is the best short account of intelligence analysis you will hear: it is classified journalism. You have a question your consumer needs answered — and when your consumer is the president and the question is how long Assad can hold on, the sourcing is satellite imagery, stolen documents, intercepted calls and texts, all of it fragmentary and some of it more credible than the rest. You stitch together the what, the why and the so what, and you write it up clean and apolitical so someone can decide something. Everything he publishes now, from a novel to a résumé, still goes to the agency’s publication review board — though he argues the genuinely sensitive material would make for bad fiction anyway. Afterwards came McKinsey, where the work was pricing concrete pipe and clearing a rail bottleneck through Chicago, and where he was, as far as he knows, the only McCloskey.
•       Artemis Aphrodite Procter. The character began with a real colleague — short, dark curly hair, not physically imposing but with a genuine physicality, and possessed of what he calls a mix of mirth and menace: someone you can drink and gamble with who will terrify you two days later. Then, as good characters do, she started speaking for herself. Her biography is a small comic masterpiece: named by a father obsessed with Greek mythology, raised outside Kissimmee in central Florida on the family business, the nation’s third largest alligator-themed amusement park, called Gatorville; university in Ohio; a first husband who didn’t work out; an application to the agency made more or less on a whim. Not upper class, not really middle class, and it shows in how she dresses, eats and speaks to people — which is exactly why putting her in London was irresistible. One character calls her a competent extremist. The running question across the books, he says, is what it means to be loyal to a place that is not loyal to you.
•       Can a Man Write Her? AK put the objection directly: some literary thought police would say a male former CIA officer has no business inventing a female one. Two answers. First, the same logic throws out Anna Karenina, and he would rather part company than follow it. Second, and more interestingly, he thinks the objection misunderstands what novelists do: you are not inhabiting a character’s skin, you are hearing a voice and putting it on the page. His analogy is a reporter interviewing WNBA players — not sharing their background means more work, more research, more time with the subject, not a prohibition. On whether women make different spies he declines the generalization, while noting a real asymmetry: his analytical teams at the agency were at least half women, often majority, while the case officer cadre and the upper rungs of the Directorate of Operations remain heavily male. Which leaves the question he finds genuinely interesting about Procter — how much of her was always there, and how much the organization made.
•       No Smiley Without Philby. On traitors, he separates the fiction from the trade. In literature they are everything, and Philby above all — not only because he turned an institution inside out but because his betrayal said something deep about the British class system that still resonates. Hence the line of the hour: you probably don’t have George Smiley without Kim Philby, le Carré having essentially novelized the wound. (AK’s reversal: perhaps you don’t have Philby, in the public mind, without Smiley.) Inside the actual bureaucracies, though, nobody is wrestling with those ghosts day to day. And on Britain’s place in all this, he refuses AK’s harsher framing — a subtext to a subtext — while conceding the structural reality: the agency is much larger and much richer, and when it wants to do somet...

“Kennedy and Trump are taking a wrecking ball to probably the most important thing we have done consistently better than the rest of the world over the last eighty years, and I don’t get it.” — Barry Werth
 
Last week on the show, the San Francisco based AI mogul and Eureka Machine author Richard Socher promised a scientific renaissance in which AI will cure everything. Socher’s eureka moment, however, may have already occurred. Not with AI, but with big pharma. At least according to Barry Werth who has spent the last thirty-five years writing about American pharma. Indeed, his new book The Age of Cures: How American Scientists Saved Your Life, a twentieth century history of big pharma, completes a trilogy that began with The Billion-Dollar Molecule in 1994.
 
When it comes to medical breakthroughs, Barry Werth isn’t particularly optimistic on the AI front. It will add more data to preexisting mountains of data, he warns. And Werth reports that many leading scientific designers of drugs don’t think it will change much. Given, however, the hundreds of billions of dollars now being poured into its ecosystem, what today’s AI revolution does instead is remind you what a real scientific breakthrough costs, both politically and economically.
 
Thus the significance of The Age of Cures, which The New York Times is already recommending. In his new history, Werth reminds us that ninety years ago there were almost no drugs for anything. Penicillin was discovered by accident in 1928, and by 1941 there was only enough of it to treat a single London policeman — who began to recover, then died when the supply ran out.
 
Cutting edge technology plus state investment changed everything. The engineer and scientific mandarin Vannevar Bush (the guy who also first imagined the internet) persuaded FDR to fund science publicly without conscripting the scientists. And Bush’s Office of Scientific Research and Development, the same state institution that fathered the Manhattan Project, made penicillin for D-Day its number one priority. Bush then ensured the arrangement continued in peacetime, which it did, for eighty years.
 
According to Werth, both big pharma’s idealism and its now much derided financial ambition were real. George Merck’s famous line about medicine being for people, not for profits, is true — as long as we remember that those profits will follow from life-saving drugs. What broke this virtuous cycle, Werth argues, was the Wall Street short-termism of the greed-is-good Eighties that focused only on quarterly profits. So today, the term Big Pharma is mostly used as an insult. Big Tech moguls like Richard Socher should thus read Barry Werth’s The Age of Cures with particular care.
 
Further Reading:
 
•       The Age of Cures: How American Scientists Saved Your Life by Barry Werth (Simon & Schuster, September 22, 2026), one of the New York Times’ recommended books this week — the third book written, and the first chronologically, in a trilogy that began with The Billion-Dollar Molecule (1994) and continued with The Antidote: Inside the World of New Pharma (2014).
•       “The system that saved your life is being destroyed” — Werth’s own excerpt from the book, in Fortune.
•       Science, the Endless Frontier — Vannevar Bush’s 1945 report to President Truman, still cited today, and the founding document of American public-private science. Reissued by Princeton University Press with essays on what it got right and wrong.
•       Empire of Pain by Patrick Radden Keefe, a recent guest on this show, on the Sacklers and OxyContin, and No More Tears on Johnson & Johnson — the case for the prosecution, raised in this conversation.
•       Earlier this week on Keen On: Richard Socher on why AI will unlock a new scientific renaissance — the argument this episode answers.
 
Five Takeaways
 
•       Before the Cures. The premise is a fact most people have forgotten: as recently as eighty or ninety years ago there were almost no drugs for any disease. There were natural substances discovered over the eons, and no established method for finding new medicines. Paul Ehrlich postulated that chemicals could treat disease; it took another twenty years before the sulfa drugs suggested he might be right. The story Werth uses to make it concrete is penicillin. Discovered by accident in 1928 when mould spores drifted into Alexander Fleming’s laboratory, it was still so scarce in 1941 — during the Blitz, before America entered the war — that the entire synthesized supply went to one very sick London policeman. He started to recover. They ran out. They harvested what they could from his urine and reinjected it. He died. The British scientists then came to the United States hat in hand, having found no American company willing to take it on.
•       Vannevar Bush’s Machine. What turned that around is the book’s central argument, and its hero is not a chemist. Vannevar Bush was an electrical engineer who had built one of the first room-sized analog computers. Surveying American science in 1940, he and a handful of other administrators found it disorganized and no match for the Germans, who had the best chemistry in the world. Bush went to Franklin Roosevelt alone with a proposal: fund it publicly, but don’t draft the scientists into the military as in the First World War; let committees of leading scientists make the decisions, distribute the work widely, and make rivals cooperate. The result was the Office of Scientific Research and Development — better known for fathering the Manhattan Project — whose head of medical research, Alfred Newton Richards, made penicillin for D-Day the number one priority and delivered it. Afterwards Bush wrote Science, the Endless Frontier, the 1945 report arguing the partnership should continue in peacetime. It did. Sustained federal funding, transparency, cooperation between companies, universities and government: that, Werth says, is how America got great at science. AK’s summary: cutting edge technology plus state investment changed everything.
•       Medicine Is for People. George Merck — whose family had been in the pharmaceutical business in Germany since the 1600s, and who was sent to Harvard to become a chemist before anti-German feeling in the First World War pushed him into the family firm — was on the cover of Time the week Werth was born in 1952, under the line that medicine is for the people. The usual ...

Amidst all the talk of an AI bubble, bonds are back in the news. The world’s two leading business newspapers, The Wall Street Journal and The Financial Times, both feature stories this weekend warning about bond ructions and surging yields. While these warnings might only interfere with the sleep of bond geeks, they are a good reminder of the centrality of the bond market to our prosperity.
 
For all their financial geekiness, Robin Wigglesworth knows that bonds matter. He edits Alphaville, the Financial Times’ splendidly geekish markets blog, and he’s the author of A Fabulous Debt, a new book that tells the “epic” story of how bonds built the modern world. The bond market matters, Wigglesworth reminds us, because when it breaks — as in 1873, 2008 and very nearly in 2020 — then the whole world suffers.
 
So what, exactly, is a bond? Wigglesworth explains that it’s a tradable loan which the Venetians accidentally invented in 1171 to pay for a war with Constantinople. In Venice, it became a forced levy on every citizen, sweetened by its annual 5% interest payments and by the right to sell its paper on the Rialto. From there it runs through Amsterdam with the Dutch East India Company to modern America, where Thomas Jefferson’s Louisiana Purchase was financed by bonds sold by the London merchant bank Baring Brothers — remarkably, while Britain was at war with the France it was paying, and earning Barings what may have been the largest banking fee in history. Then there’s the deliciously named Scottish adventurer Gregor MacGregor, a bond bad boy who invented an entire country, Poyais, and successfully sold its bonds in London.
 
In spite of the shady MacGregor, the moral of the bond story, Wigglesworth reassures us, is that transparency pays. Britain could borrow more cheaply than its rivals because parliament controlled the purse and published its accounts. Europe’s greatest banker, the London-based Nathan Mayer Rothschild, who brought Prussia, Spain, Russia and the new Latin American republics to the London market after Waterloo, thus cajoled other governments into opening their books to get the same rates. Accountability was, literally, cheaper. And so the modern world — with its leviathan state and the financial muscle to conduct global wars — was born.
 
Does Wigglesworth lose sleep about today’s bond ructions and surging yields? Not exactly. But he is a tad nervous about the fact that tech giants have issued over $500 billion of bonds this year to build data centers. So rather than the feared AI stock market bust, Wigglesworth does raise the more apocalyptic specter of a bond bust. Not an imminent fabulous crisis, he reassures. Not yet, anyway.
 
Five Takeaways
 
•       Stock Bubbles Don’t Matter. Bond Bubbles Do. Wigglesworth’s opening move reverses the usual hierarchy. Stock market bubbles are mostly inconsequential and sometimes even welcome — we occasionally need people to dream rather than be careful, and to throw money at the hot new thing, because that is when big leaps happen. The dot-com bust is his proof: the American stock market lost half its value from peak to trough, and the recession that followed was among the shortest and shallowest in US history. Credit is the dangerous part. When the bond market breaks bad — 1873, 2008, and very nearly 2020 — it tends to be very painful for the world. The structural reason is asymmetry: a shareholder’s upside is unlimited and the downside is zero, so the stock market is the natural home of wild speculation; a bondholder’s best case is simply being repaid. That is why bonds work best when they are safe, or at least seen as safe — and why the rare occasions when bond speculation does run out of control have proved more destructive than tulips, the South Sea, or the dot-com boom.
•       Venice, 1171. Credit is ancient — Mesopotamian clay tablets recorded loans four thousand years ago, some made out not to a named person but simply to a merchant, which suggests they were traded. But the bond in its proper form was invented by accident in Venice in 1171, to finance a war with Constantinople. The city was rich, but not rich enough to borrow from the bankers at their tables in the Rialto, so the government compelled every citizen to lend in proportion to their wealth — an involuntary war tax, softened two ways: 5% a year until victory, and a receipt you could sell. The Venetians hated it at first, and then discovered it was collateral, and de facto paper money before paper money existed in Europe. Milan, Pisa, Genoa and Rome copied it; it spread to France, Germany, Spain and finally the Netherlands, where the Dutch turned a handful of loans into an actual market — cities, states and the water boards that paid for the dykes — and, Wigglesworth argues, invented modern capitalism. The Dutch East India Company sold stock once. It financed its ships, armies and expeditions with bonds.
•       The Man Who Invented a Country. The book’s great set piece. Gregor MacGregor, a Scottish mercenary who had fought beside Bolívar and married his cousin, returned to London as cacique of Poyais — a verdant land on the Honduran coast, he said, with gold, honey, fertile plains and a capital called Saint Joseph. He had songs written about it. He sold its bonds on the London Stock Exchange, raising hundreds of thousands of pounds, and persuaded Scottish families to sell up and sail. Poyais did not exist. It was a swampy hellhole with no capital and no settlement; some who arrived took their own lives, many died of famine and malaria, and only a handful got home. Why bonds rather than stock? Because that is where the money was: Britain had financed the Napoleonic wars with an enormous issue of consols, those who held them made fortunes when Britain won, and Nathan Mayer Rothschild was busy bringing Prussia, Spain, Russia and the new Latin American republics to the London market. In the middle of what we would now call an emerging market bond bubble, an invented country did not sound so outlandish.
•       Transparency Pays. The argument with the widest reach. Britain’s advantage in the nineteenth century was not only naval: its bonds were issued by a parliament that controlled the purse and published financial numbers, rather than by a fickle royal family, and Dutch, German and French investors preferred them for exactly that reason. Cheaper borrowing made Britain stronger, and Rothschild then went round cajoling other governments to open their books and issue through their estates rather than their kings, because that is what British money now expected. Accountability, in other words, had a price advantage — a quiet feedback loop pushing governments towards openness. America’s version is Hamilton, who assumed the states’ defaulted war debts and turned them into federal bonds that circulated as money in a country desperately short of coin — a financial enema, as the musical has it. The Louisiana Purchase was financed entirely by bonds, arranged by the British bank Baring Brothers while Britain was at war with France. Wigglesworth calls it a win-win-win; Andrew notes the indigenou...

Lenin supposedly said there are decades where nothing happens, and weeks where decades happen. This wasn’t one of those weeks.
 
It might have looked like a historic week. The AI circus rolled into the UN with Sam Altman and Dario Amodei addressing the Security Council on the peril facing “humanity as a whole.” Britain and Australia weighed in with middling diplomatic or regulatory AI initiatives of their own. Trump renamed AI as superintelligence.
 
But none of this changed anything. The UN continued its kabuki theater of making global pronouncements without having the power to change even the most local of things. Britain and Australia remained not-great powers with delusions of grandeur. In the autumn of his patriarchy, Trump continued to flail with pathetic powerlessness. And Dario and Sam increasingly seem like bit actors in the unfolding AI drama.
 
So where’s all the power in AI? What’s this week’s real story? In Ezra Klein’s New York Times conversation with Nvidia CEO Jensen Huang, which That Was The Week publisher Keith Teare called “an interview for the ages,” Huang, the CEO of the world’s most valuable company, reveals himself as the guy running the AI show.
 
The $5 trillion dollar man talks with the quiet authority of the AI headmaster, fearlessly scolding his riotous sixth form. And his most riotous pupils are Dario Amodei and Sam Altman. So, in response to their latest handwringing about AI’s existential threat to humanity, Huang’s message was unambiguous. If you can’t align it, “don’t ship it,” he said of these supposedly humanity-destroying AI products, and if you truly cannot contain your own experiments, “then I think the answer is that we have to shut the labs down.”
 
The most revealing part of the Klein interview was about Nvidia’s astonishing economic clout in the AI economy. Huang put his chip manufacturer’s investment in this ecosystem at around $100 billion — larger, as Klein noted, than Joe Biden’s entire CHIPS and Science Act. This guy has serious clout, equivalent to or perhaps even surpassing the US state. He doesn’t waste his time at the UN. He’s not running wannabe great powers like Australia or Britain. He doesn’t play senile semantic games like you-know-who.
 
Huang can lay down the law to Sam and Dario (and Trump) because he doesn’t fear them. As the Exponential View’s Azeem Azhar reveals in a revealing interview this week with Atlantic CEO Nick Thompson, Nvidia isn’t dependent on the walled garden innovators at OpenAI and Anthropic. Neither lab actually buys Huang’s hardware — they rent capacity from the cloud companies that do, which makes them tenants rather than customers. Meanwhile the cheaper and maybe even better open weight models now pouring out of China run on the same Nvidia silicon. Whoever wins the AI race, Jensen “Levi Strauss” Huang sells them the chips.
 
So Mr Chips to Sam and Dario: heads you lose, tails I win. Shut up or shut down. That’s what happened this week.
 
Five Takeaways
 
•       The Theatre of the UN. Altman and Amodei chose the Security Council rather than the General Assembly — the biggest forum available, and one with at least some teeth. What they want, in Keith’s reading, is political globalization: a super-state view of AI, with rules set from New York. But AI has already globalized, through the spread of open-source models, and not in a form they like. His analogy is Facebook’s Libra, launched on the condition that every government agree, and dead for exactly that reason. The second motive is liability. The conversation of the past fortnight has moved to who gets sued when something goes wrong, and the only rational answer is the companies that built it — unless they are following rules set by government, in which case liability moves. Both are heading for IPOs. As for Andy Burnham, Keith is unimpressed: Britain built its modern advantage on not regulating, and grasping regulation as the theme of a G20 chairmanship is unlikely to serve either the country or the man.
•       Huang Calls the Bluff. The week’s main event was Ezra Klein’s long interview with Jensen Huang, which Keith calls an interview for the ages. Huang’s frame is the engineer’s: the Golden Gate was not the first bridge, and the Wright Flyer was not the first attempt at flight; engineering is failure leading to success, and what Altman and Amodei describe as species-like behaviour is to him an engineering problem. His conclusion was blunt, and in his own words: if you can’t align it, “don’t ship it”; and if the labs genuinely believe their experiments cannot be contained, “then I think the answer is that we have to shut the labs down,” because the liabilities — civil and criminal — are incredible. What he objects to is the request for relief: asking for regulation while asking for antitrust and product-liability protection makes no sense to him. Keith goes further: Huang is accusing them of being manipulative liars, and he agrees. Andrew pushed back on the idea that the CEO of a five-trillion-dollar company is speaking from pure engineering conviction — you cannot dismiss Sam and Dario’s interests and then treat Huang’s as irrelevant. Keith’s answer: he isn’t pursuing commercial goals, he is angry as an engineer.
•       More Than the CHIPS Act. The line Andrew kept returning to. When Huang mentioned Nvidia’s hundred billion dollars of investment in the AI industry, Klein muttered that this was more than the entire Biden CHIPS Act — and in that aside sits the argument of the week. Governments, Keith says, are missing two things: vision, and money. They are deep in debt, and even if they had the cash there is little reason to trust what they would do with it. This is not an era of state-led innovation but of bottoms-up innovation, and between seventy-five and eighty-five per cent of it is happening in one dot on the world map. Andrew’s caveat is historical — there was a twentieth century in which governments built health services and semiconductors and space programmes, and some people are entitled to be nostalgic for it — but the era has passed, and the UN speeches are its echo.
•       Not Dependent on OpenAI. The most useful technical thread, drawn from Nick Thompson’s interview with Azeem Azhar. Azhar’s point is that Nvidia is not especially dependent on OpenAI and Anthropic, which is part of why Huang can afford to attack them: both labs are renters rather than buyers, taking GPU capacity from the companies who actually purchase Nvidia’s hardware. Keith’s refinement is that this makes them indirect customers rather than irrelevant ones — every GPU purchase is good for Jensen, and the labs depend on the availability of what others buy. Measured in tokens, OpenAI is far ahead of everyone; measured in dollars, Anthropic does better, because its price per token is higher. On the bubble question Azhar is cautiously optimistic and gives numbers; Thompson is less so; and Andrew’s conclusion is the one that matters for anyone watching valuations: the AI economy could quite plausibly prosper without either OpenAI or Anthropic.
•       Zuckerberg’s Return. The quiet story of the week, and the one both men think is the real news. Meta’s Muse is a full-stack agent — models at th...