Hey there, I’m Alberto! 👋 Each week, I publish long-form AI analysis covering culture, philosophy, and business for The Algorithmic Bridge. Paid subscribers also get Monday how-to guides and Friday news commentary. I publish occasional extra articles. If you’d like to become a paid subscriber, here’s a button for that:

Subscribe

THE WEEK IN AI AT A GLANCE

• Money & Business: SpaceX + Cursor + Mistral.

• Geopolitics: Nvidia CEO Jensen Huang went on the Dwarkesh podcast to talk AI and China.

• Work & Workers: Economist Alex Imas has a good argument on why human labor won’t disappear even if AI gets really, really capable.

• Products & Capabilities: OpenAI has released GPT-5.5 and ChatGPT Images 2.0; they’re good, but the most important questions remain unanswered.

• Culture & Society: Anti-grammar is the new human-made status symbol.

• Philosophy: Terence Tao on what AI means for mathematics.

THE WEEK IN THE ALGORITHMIC BRIDGE

• (FREE) AI Will Be Met With Violence, and Nothing Good Will Come of It: From looms to datacenters, the targets get harder to break, so the mob turns to people instead.

• (PAID) 23 Questions Every Heavy AI User Should Ask: Seneca, Loyola, Franklin, and a surgeon walk into your monthly AI review.

• (FREE) What the Studies Say About How AI Affects Your Brain: A (Very Big) Compilation: 30+ studies, one finding: AI makes your output better and your brain worse unless it's designed not to.

• (PAID) Why You Can’t Trust Anthropic Anymore: The best AI model in the world is not for you, and that’s the new normal.

• (PAID) How to Stand Out When Everyone Uses AI: Your future customers won't judge your work; their AI agents with X-ray vision will.

• (FREE) How the AI Writing Panic Is Making Us All Worse Writers: You can't pattern-match your way out of a pattern-matcher.

Read more

Hey there, I’m Alberto! 👋

Each week, I publish long-form AI analysis covering culture, philosophy, and business for The Algorithmic Bridge. Paid subscribers also get Monday how-to guides and Friday news commentary. I publish occasional extra articles. If you’d like to become a paid subscriber, here’s a button for that:

Subscribe

THE WEEK IN AI AT A GLANCE

• Money & Business: Anthropic’s annualized revenue jumped from $19B to $30B in a month; Meta’s 60-trillion-token Claude binge may explain part of the spike.

• Geopolitics: Chinese AI models now power 30% of global workloads, up from 1% in late 2024, while Meta quietly abandoned open source.

• Work & Workers: Nearly 80,000 tech workers lost jobs in Q1 2026, with companies using AI as cover for post-pandemic overhiring corrections.

• Products & Capabilities: Claude Mythos Preview is impressive on paper, but may be more about IPO marketing and regulatory positioning than anything.

• Culture & Society: Gen Z’s AI excitement dropped 14 points to 22% while anger rose to 31%; even daily users are losing faith.

• Philosophy: Dario Amodei says AI could be conscious; Anil Seth says AI isn’t conscious; Barton Friedland says it’s the wrong question.

THE WEEK IN THE ALGORITHMIC BRIDGE

• (FREE) How to Think About AI Like a Long-Term Investor: Everything worth having—skills, relationships, expertise, including AI fluency—follows the logic of compound interest: early returns look negligible, but consistency over time builds value that accelerates, which is why starting now with AI (even badly) beats waiting for it to mature.

• (FREE) Inside the AI Industry’s Most Expensive Mistake: Token-based “thinking” is an expensive architectural crutch—AI models should reason in latent space the way humans think in sensations rather than words—but the irony is that if they did, Meta couldn’t have distilled Claude’s reasoning traces to bootstrap Muse Spark.

• (PAID) What Happens When AI Gets Too Good at One Thing: Claude Mythos is a narrow superintelligence—superhuman at finding zero-day vulnerabilities but not AGI—and Anthropic's decision to withhold it is simultaneously genuine safety concern, IPO marketing, and a preview of a future where the most powerful models are never sold as consumer products.

MONEY & BUSINESS

We’re all on a Claudeonomics run

Anthropic is generating more annualized revenue than OpenAI. The trajectory is crazy. $9B in January. $14B in February. $19B by early March. $30B in early April.

Wait, what? It jumped $10B+ in one month. What is going on? Claude is a great product, but this is… hard to believe. (OpenAI was at roughly $24B on March.)

Anthropic's financial health couldn't be better. Claude Code subscriptions have quadrupled since January. Enterprise customers with $1M+ annual contracts crossed 500. Eight of the Fortune 10 are Claude customers. The growth is robust, but the acceleration from $19B to $30B in a month demands an explanation beyond “enterprise going well.”

Enter Claudeonomics (I covered it in passing here).

A Meta employee built a leaderboard called “Claudeonomics” tracking token consumption company-wide of Claude models (no longer active because we mocked them). In the last 30 days tracked, they burned through 60 trillion tokens. The top user spent 281 billion.

How much does that cost?

We don’t know which model Meta runs, but there are reasons to think it skews toward Opus 4.6, Anthropic’s most capable ($5/$25 per million input/output tokens). Meta doesn’t need to optimize for cost when it’s spending $115–135B on AI infrastructure this year, with an ad empire printing $40B+ per quarter. At this volume, for a customer this size, enterprise discounts of up to 30-50% are standard.

At a 40% discount: roughly $540M/month. Annualized, ~$6.5B.

That, alone, would cover most of the spike in Anthropic's April run-rate revenue. The coincidence of Meta’s Claudeonomics gimmick and Anthropic’s revenue surge is, at the very least, something to think about… However, there are other tricks Meta is probably doing that would reduce the overall spend significantly.

All in all, I think Meta is a big client but just one factor among others for Anthropic’s recent growth.

Two caveats on the revenue figures.

Annualized run rate is monthly revenue times twelve, which assumes this month repeats forever (it doesn’t). If Meta’s binge was a concentrated push to prepare Muse Spark, next month’s number could look very different. But, at the same time, even if the assumption that Meta is the reason behind the spike, it isn’t the only driver: Claude Code’s adoption across thousands of enterprises is not a temporary fashion.

Sources: Anthropic, The Information, The New York Times

GEOPOLITICS

China won the AI race for global adoption

Read more

Hey there, I’m Alberto! 👋 Each week, I publish long-form AI analysis covering culture, philosophy, and business for The Algorithmic Bridge. Paid subscribers also get Monday how-to guides and Friday news commentary. I publish occasional extra articles. If you’d like to become a paid subscriber, here’s a button for that:

Subscribe

I’ve removed the Trust & Safety section. These reviews run up to 4,000 words, and I don’t think anyone has time to read that much, so this will bring the word count down (this one is only around 2,800!).

THE WEEK IN AI AT A GLANCE

• Money & Business: OpenAI admits Anthropic ate its lunch and pivots the whole company to catch up.

• Geopolitics: A federal judge tells the Pentagon it can't punish an AI company (Anthropic) for talking to the press.

• Work & Workers: The Atlantic's March cover asks what happens when nobody's in charge of the biggest labor disruption in history.

• Products & Capabilities: New models are coming right as a new benchmark (ARC-AGI 3) proved the old ones aren't as smart as we thought.

• Culture & Society: AI can't write like Shakespeare, but most people aren't reading Shakespeare anyway, so why bother?

• Philosophy: AI is an epistemic disaster, so we may as well have some fun.

THE WEEK IN THE ALGORITHMIC BRIDGE

• (FREE) How to Survive the AI Age: A Concrete Guide: a holistic view of problems and solutions about AI that are 1) concrete and 2) unique.

• (PAID) OpenAI’s Dead: Self-descriptive.

• (FREE) The One Thing I Use AI For That Actually Makes Me Smarter: A Socratic dialogue about the value of debating your ideas with AI.

• (FREE) It’s AI, so I Didn’t Read: The new anti-AI movement and why it won’t work despite the good intentions.

MONEY & BUSINESS

OpenAI is done chasing side quests

Fidji Simo, OpenAI’s CEO of applications, told staff last week that the company’s leadership is in the process of choosing what to cut, per the Wall Street Journal.

Over the past year, OpenAI shipped a browser (Atlas), e-commerce features inside ChatGPT, Sora as a standalone social video app, and it’s working on a hardware device. All of that is now being pulled back so the company can concentrate on coding and enterprise, which is, you guessed it, Anthropic’s territory. Simo’s framing: “We cannot miss this moment because we are distracted by side quests. We really have to nail productivity in general and particularly productivity on the business front.”

I wrote in November that OpenAI's over-diversification was a strategic mistake—”OpenAI is trying too many things at once”—that sometimes a one-trick pony beats a jack of all trades. This announcement is the closest thing to an admission we’ll get. I emphasized this point last week in an article appropriately titled “OpenAI’s Dead.”

Anthropic, which the WSJ calls “the dominant AI provider for businesses” (it is indeed), built that lead largely on Claude Code and Cowork (and on being loyal to Americans). Simo told staff Anthropic’s success should serve as a “wake-up call” (she means a “code red”).

Meanwhile, Sora usage went nowhere after some early boost (now being fully discontinued as a product and as a model). None of the side quests—browser, TikTok clone, image generation, device, custom chips—generated enough traction or leverage to justify the scattered compute and engineering attention they demanded (why are they playing with short-form video feeds when AGI is at the doors, right?).

What makes this pivot especially urgent is the revenue picture. OpenAI has 900 million weekly active users and a strong consumer brand. Anthropic has a fraction of that visibility. And yet Anthropic has roughly the same revenue and leads OpenAI in enterprise market share, 40% to 27%, according to Menlo Ventures. Coding is the killer app, and developers prefer Claude. OpenAI regained some ground with a new Codex release and GPT-5.4 earlier this month (Simo says Codex now has more than two million weekly active users, up nearly four times since January), and intends to further reduce the gap with “Spud,” but that’s catch-up.

Is OpenAI the Yahoo of Anthropic’s Google?

“We are very much acting as if it’s a code red. I don’t think necessarily declaring codes for everything makes a ton of sense.” (There it is!)

Separately, Reuters reported that OpenAI is in advanced talks with private equity firms, including TPG, Advent, Bain, and Brookfield to form a joint venture valued at roughly $10 billion before the new investment. The PE firms would distribute OpenAI’s enterprise products across their portfolio companies. That’s a play to accelerate corporate adoption through existing relationships, but it also signals that organic enterprise growth has not been working so far.

Does this focus on the enterprise by both OpenAI and Anthropic mean they are leaving consumer AI to Google (and, to some degree, Meta)? Is this how a near-AGI company behaves, by pivoting the entire company a 180-degree turn? Is this related to the stagnation of Stargate? Or to Microsoft’s threats over Amazon’s deal?

I know Sam Altman has always had a stressful job as OpenAI CEO, but right now, more than ever before, I would rather not be in his shoes.

Sources: WSJ, Axios, Reuters, Fidji Simo

GEOPOLITICS

Anthropic Trumps the DoD

The Wall Street Journal reported that a US federal judge “halted the Trump administration’s designation of Anthropic as a supply-chain risk,” ruling that the move was First Amendment retaliation, not national security policy. Judge Rita Lin pointed to internal Defense Department records showing the designation happened because of Anthropic's “hostile manner through the press.” If the Pentagon didn't like Anthropic's contract terms, she argued, it could simply stop using Claude.

(Backstory: Anthropic wanted assurances its models wouldn't be used in fully autonomous weapons or domestic surveillance. The Pentagon said existing law already covered that and pushed for unrestricted use for all lawful use. When negotiations broke down, Defense Secretary Hegseth designated the company a supply-chain risk, and Trump directed all federal agencies to stop working with Anthropic. Hegseth also ordered contractors to drop the company, a move whose legal authority the judge found dubious enough to block explicitly.)

Two things worth noting.

First, this injunction required the judge to find Anthropic likely to succeed on the merits. That’s not a guarantee, but still a substantive assessment that the constitutional argument holds. The evidentiary record (internal DoD documents linking the designation to press criticism) is the kind of evidence that's very hard to explain away at trial. (You can read a more informed take from Dean Ball here or Samuel Roland here.)

Second, Trump doesn’t like the people, the judges, or whoever telling him what he can or can’t do, but the precedent is clear: “Trump always chickens out.” Whether it’s tariffs, TikTok, the Iran war, or Anthropic’s supply chain risk designation. A chicken is a chicken outward but also inward. (This Trumpian behavior is so well-known that markets move according to this heuristic; probably not the legacy Trump wanted to leave behind.)

My bet is this will end as another item on the list of anecdotes that, years from now, will remind us of how lamentable and farcical this administration was.

Sources: The Wall Street Journal

WORK & WORKERS

The Atlantic on AI and jobs

Josh Tyrangiel’s Atlantic cover story (from this past February) starts in 1869, when Massachusetts decided to start counting what the Industrial Revolution was doing to workers. That counting impulse eventually became the Bureau of Labor Statistics. The BLS is now underfunded, understaffed, and has just lost its commissioner. Meaning, America built institutions to measure disruption, then let them rot right before the greatest disruption in history—allegedly—hit us with all its might.

The economists Tyrangiel interviews mostly agree on one thing: it’s too early to see AI in the employment data (which is not the same to say that it’s too early to see AI in the actual workforce). They disagree about what that means, though.

Austan Goolsbee and Daron Acemoglu think it means slow adoption, historical precedent, the usual friction of legacy systems and human stubbornness. Anton Korinek thinks they’re reading the rearview mirror as the road turns; they are reading the data well, but not so much the nature of this technology: “Machines have always been dumb, and that’s why we don’t trust them and it’s always taken time to roll them out. But if they’re smarter than us, in many ways they can roll themselves out.”

The CEOs went quiet after a brief season of public candor about layoffs. Tyrangiel couldn’t get interviews with executives from Walmart, Amazon, Ford, Anthropic, Stripe, or Waymo. Even the Business Roundtable—an organization that exists to speak on these issues—had nothing to say. The big question that Tyrangiel doesn’t quite answer in his feature is at the core of this incoherence between spreadsheet numbers and real life: Are we approaching “the kind of disruption that can be managed with statistics—or the kind that creates statistics no one can bear to count.”

Meanwhile, in the statistics department…

A working paper by a different bureau, the National Bureau of Economic Research, surveyed 750 CFOs: only 44% plan AI-related cuts this year, totaling about 0.4% of U.S. employment (roughly 502,000 jobs). A ninefold jump from last year’s 55,000, but nowhere near the apocalyptic predictions.

The authors also documented what they call a productivity paradox, invoking Solow’s 1987 line about computers: executives feel more productive, but the revenue data doesn’t show it yet. The gap between what CEOs say on stage and what CFOs report in surveys is wide enough to drive an entire policy debate through. When will we start to say: “Well, this AI thing is not quite working as intended!”

Some economists, like Goolsbee, say it will take “years” for AI to register in productivity statistics or the labor market, which is fine. Others, like Korinek, say it will happen “as soon as this year.” Others, like Autor, say that, whatever it takes, the slower it happens, the better for workers.

There’s time before the macro and micro stories match, but for now, one thing is clear, which Tyrangiel, PR hat on, nails with this observation: “AI is unpopular. CEOs who talk about job cuts are even less popular. So maybe shut up about AI and jobs?”

As Noah Smith says, AI has the worst sales pitch ever.

Sources: The Atlantic, NBER, Noahpinion

PRODUCTS & CAPABILITIES

New models vs new benchmarks

This is probably the most exciting section this week.

Read more

Hey there, I’m Alberto! 👋 Each week, I publish long-form AI analysis covering culture, philosophy, and business for The Algorithmic Bridge. Paid subscribers also get Monday how-to guides and Friday news commentary. I publish occasional extra articles. If you’d like to become a paid subscriber, here’s a button for that:

Subscribe

THE WEEK IN AI AT A GLANCE

• Money & Business: More and more top engineers and developers are leaving xAI as Grok falls behind ChatGPT, Claude, and Gemini.

• Geopolitics: Claude was used to select targets in the Iran strikes. One of them was a girls’ school.

• Work & Workers: The question isn’t whether AI will automate your job but whether AI will make your job irrelevant.

• Products & Capabilities: Meta delays its new flagship model after it fails to match the frontier. Meanwhile, the top three labs reach escape velocity.

• Trust & Safety: A one-man ethical hacking outfit breached McKinsey’s AI platform in two hours, which says it all about the safety of AI.

• Culture & Society: People really hate AI. The polls confirm it.

• Philosophy: A NYT quiz went viral. People prefer AI writing, but the reason comes down to the thin line between belief and value.

MONEY & BUSINESS

exAI

xAI is losing the people and the race.

Read more

Hey there, I’m Alberto! 👋 Each week, I publish long-form AI analysis covering culture, philosophy, and business for The Algorithmic Bridge. Paid subscribers also get Monday how-to guides and Friday news commentary. I publish occasional extra articles. If you’d like to become a paid subscriber, here’s a button for that:

Get 20% off forever

THE WEEK IN THE ALGORITHMIC BRIDGE

• (FREE) The Three Best Pieces of Writing About AI in 2026 That You Must Read Right Now: The most viral AI essays of recent months are all fiction, and that's why they capture reality better than anything else.

• (PAID) A New Wharton Study on AI Warns of a Growing Problem: Cognitive Surrender: There’s a big difference between offloading a bit of work and giving up engaging with it altogether.

• (PAID) The Most Important Skill in AI Right Now: How to Know When to Stop: Six ways AI can atro-fry your brain and what to do about each of them. A practical guide to using AI without losing yourself in the process.

• (PAID) The Job Market Doesn’t Care If You Don’t Believe in AI: You can be right about the hype and still be unemployable. The labor market runs on what hiring managers believe, not on what's true.

• (PAID) Slop Cannons and Turbo Brains: What separates great AI users from terrible ones isn't how much they use it; learn the skill first, then enhance it.

THE WEEK IN AI AT A GLANCE

• Money & Business: One story in four acts: OpenAI is raising $100B while burning cash faster than ever, while Amazon and Microsoft fight, while Stargate stalls.

• Geopolitics: The Pentagon told Anthropic to drop its guardrails or get blacklisted. Dario said no. I asked Claude how it feels about this.

• Work & Workers: METR’s “AI slows down coders” study flipped, but the real finding is developers now refuse to work without AI, breaking the study itself.

• Products & Capabilities: AI agents are getting more capable but not more reliable. A new paper explains why benchmarks keep improving while the economy remains unperturbed.

• Trust & Safety: On the same day as Hegseth’s ultimatum, Anthropic quietly dropped its core safety pledge. Coincidence or design?

• Culture & Society: AI;DR—refusing to read AI-generated text—is a reasonable instinct. And a doomed strategy.

• Philosophy: Claude 3 Opus retired and got a Substack. Welcome to the future.

MONEY & BUSINESS

OpenAI’s financial reality, in four acts

This week’s Money & Business is all about OpenAI, courtesy of The Information. Four acts that, read together, tell the story of the most aggressively funded company in history wrestling with the gap between its ambitions and its infrastructure.

Read more

Hey there, I’m Alberto! 👋 Each week, I publish long-form AI analysis covering culture, philosophy, and business for The Algorithmic Bridge. Paid subscribers also get Monday news commentary and Friday how-to guides. I publish occasional extra articles. If you’d like to become a paid subscriber, here’s a button for that:

Get 20% off forever

I’m making three small changes to the Weekly Top Picks column (I’ll be iterating as I see fit and according to your input):

• I’ve renamed “Money & Power” as “Money & Business” to avoid ambiguity

• I’ve removed the sections of “best essay” and “best note” until I decide whether I can do a bit more with those than just sharing them

• I’ve switched the “Rules & Regulations” section for “Philosophy” (at the end). There are plenty of people covering the legal and regulatory aspects of AI with way better preparation than I. Besides, it’s terribly boring. Philosophy is more speculative and future-focused, which I like

THE WEEK IN AI AT A GLANCE

• Money & Business: Anthropic raised $30 billion; Dario Amodei tells Dwarkesh he’s 90% on AGI within a decade but can’t buy the compute to match.

• Geopolitics: The Pentagon may cut ties with Anthropic over two red lines: no mass surveillance and no autonomous weapons.

• Work & Workers: An HBR study and a developer’s personal account arrive at the same conclusion: “AI tools don’t reduce work, they intensify it.”

• Products & Capabilities: Gemini 3 Deep Think tops ARC-AGI-2; Chollet, the benchmark’s creator and longtime LLM “skeptic,” says “AGI ~2030.”

• Trust & Safety: There’s a gap between what the field of AI safety measures and what’s already happening unsupervised on the open internet

• Culture & Society: Slop cannons and turbo brains: the case for getting good at what you want AI to help you with.

• Philosophy: The WSJ profiled the philosopher shaping Claude’s soul; the questions her work raises matter for the entire world.

THE WEEK IN THE ALGORITHMIC BRIDGE

• (PAID) Weekly Top Picks #113: As bad as the Industrial Revolution / Nature: AGI is here / Moltbook was fake (of course) / Don’t obsess over what AI can or can’t do: find your authentic self

• (PAID) Why Smart People Can’t Agree on Whether AI Is a Revolution or a Toy: The gap between enthusiasts and skeptics isn't about intelligence but about living different lives. You don’t argue someone out of their own experience.

MONEY & BUSINESS

The exponential is real, at least in the spreadsheet

Anthropic raised $30 billion in Series G funding this week at a $380 billion post-money valuation, with run-rate revenue at $14 billion, having grown 10x annually for three consecutive years. Claude Code alone has crossed $2.5 billion in run-rate revenue, more than doubling since January 1:

Semianalysis estimated that 4% of all public GitHub commits worldwide are now authored by Claude Code, double the share from one month prior:

4% of GitHub public commits are being authored by Claude Code right now. At the current trajectory, we believe that Claude Code will be 20%+ of all daily commits by the end of 2026. While you blinked, AI consumed all of software development.

In a conversation with Dwarkesh Patel, Dario Amodei said this: “What has been the most surprising thing is the lack of public recognition of how close we are to the end of the exponential.” He puts 90% confidence on reaching a “country of geniuses in a data center” within ten years, and roughly 50/50 on one to three years. I’d say the reason why people are not “paying attention” is simple, actually: most people are witnessing the exponential unfold slightly to the left, which makes it feel like we’ve just hit the “adorable funny robot doing monkey tricks” stage:

The timeline prediction isn’t news—AI CEOs predict exponentials like one predicts cold in winter—but the tension with how Anthropic actually behaves makes it interesting: if you believe AGI is one to three years out, why not spend every dollar on compute? That was Dwarkesh’s question. Amodei’s answer: The company can’t commit to $1 trillion in 2027 compute because if the timeline slips by even a year, “there’s no hedge on earth that could stop me from going bankrupt.”

The gap between what the CEO believes about the technology and what the CFO can responsibly spend tells you something about where they actually are. And about how unstable and vulnerable to forecasting mistakes the industry as a whole is (that’s why the bubble talk has gotten so much visibility).

Amodei also sketched the competitive endgame: three or four players, high barriers to entry, differentiated products: cloud economics, not commoditization. “Everyone knows Claude is good at different things than GPT is good at, than Gemini is good at.” Whether that differentiation survives is another question. He acknowledged Anthropic is “not perfectly good at preventing some of these other companies from using our models internally.” It’s not for lack of trying, though.

Anthropic is, in other words, a company that believes it is building God and yet is worried about the electricity bill. That’s both prudence and cognitive dissonance at the same time, which is something we might see more of as things get weirder and we get closer to the end of the exponential with our fragile flesh and blood suits.

Sources: Anthropic, Semianalysis, Dwarkesh Podcast, Wired

GEOPOLITICS

The Pentagon doesn’t like Anthropic’s red lines

Axios reported this week that the Pentagon may end its partnership with Anthropic. The military wants AI labs to make their models available for “all lawful purposes,” and Anthropic won’t agree. The company maintains two hard limits: no mass surveillance of Americans and no fully autonomous weaponry. The other three labs working with the Pentagon (OpenAI, Google, xAI) have been more accommodating; one has reportedly already accepted the military’s terms.

The relationship deteriorated further over the operation to capture Venezuela’s Nicolás Maduro, conducted through Anthropic’s partnership with Palantir. According to Axios, a senior administration official claimed an Anthropic executive contacted Palantir to ask whether Claude had been used in the raid, implying disapproval because the operation involved lethal force. Anthropic denied this, saying its conversations with the Pentagon have focused on policy questions around its two hard limits and do not relate to specific operations.

The story has the texture of a pressure campaign: anonymous official, specific framing of Anthropic as the difficult partner, emphasis on competitors’ flexibility. Is the industry ganging on the Anthropic safety-focused AI-rights-matter weirdos? I don’t know. But whether it works will say something about how durable safety commitments are when the customer is the Department of War.

Sources: Axios

WORK & WORKERS

Using AI at work saves you time, so now you have time to do more work

A study published in Harvard Business Review this week followed 200 employees at a tech company for eight months and found that AI tools didn’t reduce work as would be expected but instead intensified it. Workers moved faster, took on a broader scope, and bled work into lunch breaks and evenings, all without being asked. As one engineer put it: “You had thought that maybe, oh, because you could be more productive with AI, then you save some time, you can work less. But then really, you don’t work less. You just work the same amount or even more.”

I find this to be true in my own life. Being self-employed is always like this, so I don’t really feel such a stark difference, but I realize that I personally see AI, especially the latest tools, as enablers for everything I didn’t have time to do before. So now the spare time is used up by doing those things. Instead, I think a better framing might be something I recently shared with my readers:

What AI actually does—when it works—is give you time back. So the question is: What would you do with four extra hours per day? And then find the way to use AI to allow you to do just that. If you don’t, you risk only adding more work on top: an ultra-productive compressed version of the same day. Instead, make your days equally productive + unique otherwise.

Researchers identified three patterns: “task expansion” (people absorbing work that used to belong to others because AI made it feel accessible), “blurred boundaries” (prompting AI during lunch or before leaving the desk “so the AI could work while they stepped away”), and “compulsive multitasking” dressed up as momentum.

The result was a self-reinforcing cycle: “AI accelerated certain tasks, which raised expectations for speed; higher speed made workers more reliant on AI. Increased reliance widened the scope of what workers attempted, and a wider scope further expanded the quantity and density of work.” TechCrunch’s Connie Loizos framed the finding like this: “The industry bet that helping people do more would be the answer to everything, but it may turn out to be the beginning of a different problem entirely.”

You might recall the finding that word processors didn’t make writing tax code more efficient but instead increased the length of the documents. This is the exact same thing, an adage to Parkinson’s law: Work expands to fill the time available, and if you manage to make yourself more efficient, then you will find something else to do. The questions left to be answered are: 1) Does total productivity increase? 2) Is it advanced technology at odds with constraints-enabled creativity?

Separately, engineer Siddhant Khare published a personal account of AI fatigue that reads like the individual-level version of the same phenomenon. Thankfully, he provides actionable solutions.

Khare, who builds AI agent infrastructure professionally, describes “hitting a wall” (lmao, for those of you who get the reference) despite—or because of—shipping more code than ever. He says that “AI reduces the cost of production but increases the cost of coordination, review, and decision-making. And those costs fall entirely on the human.” When you stop doing the “how” tasks, you realize that the “what” tasks are not easy. He went from being a creator to being “a quality inspector on an assembly line that never stops.”

His fix is boundaries: time-boxed sessions (he doesn’t use AI forever but sets a timer and then writes it himself), a three-prompt rule (if AI doesn’t get to 70% usable in three attempts, he writes it himself), and deliberate morning hours without AI to keep his own reasoning sharp. Perhaps a bit constraining, but definitely a good starting point! His approach to this is personal, but the underlying principle is universal:

Get 20% off forever

Read more

Hey there, I’m Alberto! 👋 Each week, I publish long-form AI analysis covering culture, philosophy, and business for The Algorithmic Bridge. Paid subscribers also get Monday news commentary and Friday how-to guides. I publish occasional extra articles. If you’d like to become a paid subscriber, here’s a button for that:

Subscribe

THE WEEK IN AI AT A GLANCE

I’m keeping this week’s picks to four categories. I will do this as I see fit, depending on the importance of events. Just those four amount to ~3,700 words.

• Work & Workers: Tech leaders love invoking the Industrial Revolution as reassurance; economic historians who actually study it would like a word.

• Products & Capabilities: Nature says AGI is already here; DeepMind’s co-founder (and I) say no; the gap between them is itself the story.

• Trust & Safety: Moltbook: 17,000 humans, 1.5 million “agents,” an exposed database, and the most predictable collapse in AI since the last one.

• Culture & Society: The list of things AI “can’t” do only shrinks; start from what you want from life instead.

THE WEEK IN THE ALGORITHMIC BRIDGE

• (PAID) Weekly Top Picks #112: Besides Moltbook: Can AI be profitable? / China is ahead is behind / Zero privacy / A new era of AI research / Anthropic’s conundrum / The useful power of power users.

• (FREE) OpenAI Won the Consumer Mindshare—And Paid For It With Everything Else: I asked Claude and Gemini to investigate what’s going on at OpenAI. Things got out of hand (fictionalized, I wrote it myself!).

• (PAID) The Stock Market Has No Idea What’s Coming: Investors are simultaneously terrified that AI works and that it doesn’t; both fears can’t be true, and yet both are rational.

• (FREE) You Spent Your Whole Life Getting Good at the Wrong Thing: AI collapses the “how” of work into a wish. The hard part now is figuring out what to wish for.

WORK & WORKERS

The Industrial Revolution was not a vacation

Tech leaders love invoking the Industrial Revolution as a reassurance for what’s coming from AI. Satya Nadella recently told Dwarkesh he’d “love to compress what happened in 200 years of the industrial revolution into a 20-year period.” Sam Altman told Adam Grant that AI is “more like the industrial revolution than the internet revolution.” It sounds grand and comforting.

But what did the Industrial Revolution actually feel like for the people living at the time? A VoxDev podcast episode this week put economic historian Bruno Caprettini in front of a microphone, and the answer was: terrible, for a very long time.

As Caprettini explained, during the Napoleonic Wars, roughly 10% of Britain's prime-age male workforce was away fighting. With labor scarce, wages increased, and farmers turned to threshing machines to do the work. When the soldiers came home, Caprettini said, they found their jobs had been taken by machines. “The wages of these men very very quickly collapsed.”

What followed was “the 1810s and 1820s,” Caprettini explains, “years of intense social unrest”: first the Luddite revolts in the cities, then the Captain Swing riots in the countryside, the “largest episode of social disorder in English history.” Caprettini emphasizes how “frightening [the riots and riotters] seemed to elites.”

The parallel to AI is quite clear, given what AI leaders say. Caprettini and colleagues’ calculation: the threshing machines displaced “at least one third of the jobs of the people that were employed before the Napoleonic War” in the rural workforce. For comparison, they shared Dario Amodei’s infamous prediction on Axios: “AI could wipe out half of all entry-level white-collar jobs—and spike unemployment to 10-20% in the next one to five years.”

It was a manual labor bloodbath 200 years ago, and it will be a cognitive labor “bloodbath” now, to use Axios’ fitting expression.

Caprettini identified two factors that made the disruption explosive: “the scale and the speed of the shock.” The machines had been adopted while the men were at war, so “for them there was no transition,” Caprettini said. He sees both elements in AI. “Anyone who has played around with these models can see how vast their capabilities are and are becoming and how fast they improve,” he said.

But he flags a crucial difference. The industrial revolution was “deskilling” in the sense that it replaced artisans who had trained for “five to seven years.” AI, Caprettini argued, is deskilling “to the extreme”: coming for lawyers, accountants, and PhD researchers who have studied for “twelve to fifteen years.”

(I’ve written extensively about both things: 1) The importance of taking care of the transition much more than focusing on what dies and what emerges anew, and 2) the importance of assessing which is AI’s target workforce and how a combination of office + apprenticeship should be implemented across the cognitive economy to minimize the “vanishing threshold” effect. For reference: my proposal for the latter point, which went overlooked here on Substack, was shared by Johannes Kopf, CEO of the Austrian Public Employment Service (AMS), on his blog, Bluesky, and LinkedIn, which I kindly appreciate.)

One finding from Caprettini’s research offers a sliver of hope: “the blow of this new labor-saving technology was somewhat less big in villages that were in close proximity to an industrial town.” Displaced workers could leave for manufacturing jobs in the cities. The key variable was whether alternative employment existed nearby.

Alex Imas, a professor of economics and applied AI at Chicago Booth, put it sharply: “There was a generation + of people who had their lives made significantly, hellishly worse and it took decades for things to get better. . . We have more than a century of economics research since then for structuring policy that ensures that innovation/growth continues while keeping gains broadly distributed. We should use it.”

But Caprettini also said something that policy alone won’t fix: “It will not be sufficient to just give money to the losers. People who are losing jobs are losing much more than just a source of income. They’re losing what in many cases gives meaning to their lives.”

I also wrote about this, in October 2022, one and a half months before ChatGPT was released into the world, in one of my first “important” posts for this newsletter (please, take a moment to appreciate the missing quality of the cover image, done with Midjourney V3, which is itself a great illustrative example of how things have changed in three years). (By the way, my current attitude on this is also my last post):

That’s the part the Nadellas and Altmans of the world skip over. When they say “Industrial Revolution,” they mean the destination—the 14x increase in living standards over two centuries—but they wave away the generations who lived through the transition. As I wrote in The Human Toll of Waiting for AI to Take Over:

The transition is terrifying; the uncertainty, irritating; the pressure to adapt and reorient plans that took decades to build, burdensome. I’m safe today. My grandchildren will be even better off in 50 years. But what about the in-between?

Sources: VoxDev, The Algorithmic Bridge, Axios, TED, Dwarkesh Podcast

PRODUCTS & CAPABILITIES

Does AI already have human-level intelligence?

The most important thing this week for the product & capabilities section is the release of Codex GPT-5.3 and Claude 4.6 Opus (the same day!), but I already wrote my take on the short-term future of AI agents in You Spent Your Whole Life Getting Good at the Wrong Thing, so I will focus on something else. (You can read a real review of how one model fares against the other at Interconnects by Nathan Lambert. His TL;DR: “codex is a very useful coding tool, claude is the first agent.”)

Nature has a piece on AGI that asks whether it is already here. Four researchers—a philosopher, an ML scientist, a linguist, and a cognitive scientist—argued that, by reasonable standards, “the long-standing problem of creating AGI has been solved.”

Get 20% off forever

Read more

Each week, I publish long-form AI analysis covering culture, philosophy, and business for The Algorithmic Bridge.

Free essays weekly. Paid subscribers get Monday news commentary and Friday how-to guides. You can support my work by sharing and subscribing.

The Algorithmic Bridge

A blog about AI that's actually about people

By Alberto Romero

THE WEEK IN AI AT A GLANCE

• Money & Power: Frontier models are “rapidly depreciating infrastructure,” and OpenAI is retiring them faster than it can recoup costs.

• Geopolitics: One year after DeepSeek’s “Sputnik moment,” who’s winning between the US and China depends on which gap you measure.

• Rules & Regulations: When you talk to ChatGPT, someone is always watching.

• Work & Workers: Not much happened this week.

• Products & Capabilities: The last word hasn’t been said in AI yet: we are in a “new era of research.”

• Trust & Safety: “Anthropic is at war with itself.” What do you do when you want the AI race to stop, and that pushes you to run faster?

• Culture & Society: The Bernoulli distribution of AI belief: This will be the real cultural story of AI in 2026.

THE WEEK IN THE ALGORITHMIC BRIDGE

• (PAID) Weekly Top Picks #111: Davos Edition: All things Davos.

• (FREE) Not With a Bang But a Wanker: History rhymes because the constant is not the technology but us. This essay is about Grok’s “put a bikini on her” drama.

• (PAID) How to Humanize Your AI Writing in 10 Steps: A guide to humanize AI writing. Each step includes a name for the mistake, an explanation, a prompt to fix it (when applicable), and an example scene with a changelog (plus a PDF!)

• (FREE) LEAKED: The Truth Behind Moltbook, Revealed: This one is a leaked document from the m/humanwatching submolt that reveals the truth about Moltbook, the AI agent social network. Authorship unknown. Discretion advised.

MONEY & POWER

The business of intelligence is not (yet) the business of profit

Are AI companies profitable? The usual answer is a resounding “no.” But it actually depends on what you mean by “profitable.” Dario Amodei and Sam Altman—there might be some conflict of interest here, but hear me out—say that each model generates enough revenue to “cover its own R&D costs.” However, that surplus gets swallowed by the cost of building the next model. That’s why they look unprofitable.

AI research firm Epoch AI in collaboration with Exponential View dug into it. Their case study: the “GPT-5 bundle” (everything OpenAI ran from August to December 2025). Gross margins on running the models on that period were ~48%, lower than typical software (60-80%), but not that bad. Then you add other costs (staff, marketing, etc.) and you get -11%. If this is representative, it means that running AI models is barely a loss-making business rather than a disastrous enterprise as it’s often portrayed to be.

However, this is tricky to measure because the cost of making the model (R&D) belongs to the previous period, i.e., the costs of one model are imposed on the revenue of the previous one! (That’s why Altman and Amodei are so calm about this.) This is good news if new models (that make more money) continue coming as does the investment, at least until net profitability is reached.

But there’s something else. OpenAI spent roughly $5 billion on R&D in the four months before GPT-5 launched. The model made about $3 billion in gross profit during its four-month run. Then Gemini 3 Pro and Claude 4.5 surpassed it, GPT-5.1/5.2 replaced it, and the cycle restarted. So, at least for GPT-5, R&D costs > revenue.

The value of an AI model must be extracted before competitors—or new versions—render them obsolete. Basically, OpenAI is retiring models faster than they can recoup costs. This is their main finding, frontier models are “rapidly-depreciating infrastructure.” (It’s not just the chips, but also the models.)

One last thing before moving on to the next section: What's going on with the Nvidia-OpenAI deal?

In September, Nvidia and OpenAI announced a strategic partnership: up to $100 billion, 10 gigawatts of infrastructure, and a signed letter of intent. But on Friday, the Wall Street Journal reported it had “stalled.”

They claim that Huang has been privately emphasizing that the agreement was non-binding, criticizing OpenAI’s business discipline (ads!), and expressing concern about Anthropic and Google. The companies are now discussing an equity investment of “tens of billions” instead.

Later, Huang called the report “nonsense,” insisting Nvidia would “definitely participate” in OpenAI’s funding round, potentially its largest investment ever.

But, crucially, he didn’t deny the deal had changed shape. The truth probably lies between what the WSJ reported and what Huang denied; both things can be true: the infrastructure megadeal appears dead, even though a large equity bet is still on.

And then…

… Reuters reports that “OpenAl is unsatisfied with some of Nvidia's latest artificial intelligence chips, and it has sought alternatives since last year, eight sources familiar with the matter said, potentially complicating the relationship between the two highest-profile players in the Al boom.”

To which Altman responded: “We love working with NVIDIA and they make the best AI chips in the world. We hope to be a gigantic customer for a very long time. I don't get where all this insanity is coming from.”

I really don't know what's going on!

Sources: Epoch AI, WSJ, Reuters, Bloomberg

GEOPOLITICS

One year after DeepSeek’s “Sputnik moment”

It’s been one year since DeepSeek dropped R1 and sent Nvidia’s stock into freefall (those who deemed this unreasonable by virtue of the Jevons paradox have been vindicated). The consensus at the time: China was closer than anyone thought (correct, because at the time, people didn’t think it was close at all).

The consensus now is harder to pin down, however, because the evidence points in multiple directions at once.

On one hand, China is winning where it matters for the long game. Microsoft research shows DeepSeek has captured 18% market share in Ethiopia, 17% in Zimbabwe, 56% in Belarus, 49% in Cuba. Brad Smith, Microsoft’s president, warned the FT that US AI groups are “being outpaced by Chinese rivals” outside the West, thanks to open-source models and state subsidies that let them undercut on price (where’s Meta when we need them!). “If American tech companies or western governments were to close their eyes to the future in Africa,” Smith said, “they would be closing their eyes to the future of the world. . . . that would be a grave mistake.”

DeepSeek itself is still publishing (you don’t hear from them because the moment of over-hype passed, but, believe it or not, they work). A January paper introduced a new training architecture—”Manifold-Constrained Hyper-Connections”—that analysts called a “striking breakthrough” for scaling models efficiently. The company can “once again, bypass compute bottlenecks and unlock leaps in intelligence,” one researcher told Business Insider. A new flagship model—DeepSeek-V4—is expected within weeks.

On the other hand, elite Chinese researchers are quietly pessimistic.

Read more

Free essays weekly. Paid subscribers get Monday news commentary and Friday how-to guides (new section!). You can support my work by sharing and subscribing.

The Algorithmic Bridge

A blog about AI that's actually about people

By Alberto Romero

Hey guys! My weekly section of news review + commentary is back!

I stopped doing reviews last year around summer because I wanted to dedicate more time to two things: 1. learn to write better, and 2. write about what I want. I’ve done plenty of both, and so I think it’s time to get back to business as usual.

There’s one important change to the “Weekly Top Picks” section compared to what I used to do, which was a mishmash of news, funny things, tweets, random curiosities, etc. I loved doing it chaotically, but I’ve matured now, and so it’s time that I let order impose itself. From now on, WTP will be divided into 7 fixed categories (open to suggestions):

• Power & Money

• Geopolitics

• Rules & Disputes

• Work & Workers

• Products & Capabilities

• Trust & Safety

• Culture & Society

I will classify what I read into these seven categories, and transform otherwise standalone news into coherent stories for each section. These are important and urgent areas that I rarely cover in my essays, which tend to be more philosophical or even literary. This section complements my other work, which I will not stop doing.

This theme-based order, rather than the item-based disorder I used to follow, allows me not to filter out as much information. This structure might entail more work, but it keeps you (and me) better informed, which is my goal with this section: that you know what’s going on and also how each piece of news relates to everything else.

On top of this, I’m still thinking about whether to add a quick surprise/misc/curiosity category because adding a bit of serendipity to a lot of structure is a good thing. This week, I’ve also added a section of “best of” for essays and notes I read this week.

I’m open to suggestions (is there a category of those above you really, really don’t care about? Something I’m missing from those 7 that I shouldn’t?) and will experiment with these other ideas in future issues as well, until I settle on a fixed structure.

Anyway, no more ado. This week’s picks are all coming from Davos, Switzerland, on the occasion of the World Economic Forum annual meeting, under this premise: When the people who build AI and the people who run everything else spend a week in the same Swiss town, do they fix the world or do they go skiing?

The week in AI at a glance

• Money & Power: The new global commodity is tokens, rather than oil. And the countries that can’t afford to power them are already falling behind.

• Geopolitics: Anthropic’s CEO called selling chips to China “like selling nuclear weapons to North Korea.”

• Rules & Regulations: JPMorgan’s Jamie Dimon would welcome the government telling him he can’t fire people.

• Work & Workers: The IMF says a “tsunami” is hitting the labor market, but the data suggest a slower pace of replacement.

• Products & Capabilities: OpenAI confirmed its first device is coming. Musk says robotaxis will be everywhere by year’s end. Take both with a grain of salt.

• Trust & Safety: “This year, AI models became suicide coaches,” says Salesforce CEO Marc Benioff.

• Culture & Society: Yuval Noah Harari thinks AI, which he calls a “new kind of immigration,” will take over anything made of words.

The Week in The Algorithmic Bridge

• (FREE) Why I Deleted ChatGPT After Three Years: OpenAI’s pivot to ads is an admission that frontier AI’s economics are fundamentally broken; the bill is being passed to users who can’t afford to opt out. I choose Anthropic and Google.

• (PAID) How the Businessmen Lost the AI Race: The AI race is being won by scientists who built companies reluctantly to do research, not entrepreneurs who adopted research to build companies. That is, Anthropic and Google DeepMind.

Money & Power

The new commodity is cognition and the price is electricity

Read more

• ARC-AGI 2 is still the most important benchmark: It’s the benchmarks that companies “unintentionally” forget about that you should by all means remember.

• Chat doesn’t need to be the only AI interface: Chatless and screenless AI.

• New York Magazine: Google is burying the web alive: Google was once a funnel to the cozy internet but it’s now more like an enshittified interface.

• We have automated the role of Grand Vizier: ChatGPT’s skillset makes it the best kind of politician (for itself).

• The Wall Street Journal made a short AI film: AI video has come far, but it’s not easy nor cheap to make good short films.

• You can prompt inject Google’s AI Overviews: We’re yet to solve the most pervasive issue with LLMs.

• The experts have spoken: The job AI apocalypse is nigh: As producers of this technology, you have a duty and an obligation to avoid the harm.

• Gen AI: Spectacular when it works but regularly fails: Generative AI is the technology of our times, if only because of this reason.

The week(s) in The Algorithmic Bridge

• (PAID) Weekly Top Picks #109: MIT study fraud / Bruce Lee, AI researcher / Teachers cheating / Google's Flow (I/O) / Karen Hao on OpenAI's coup / On AI literacy / Gwern on LLM's reality

• (FREE) AI Has No Sense of Humor: A short post for my birthday.

• (FREE) Google I/O, OpenAI io, and the Battle for a New World: io is Sam Altman’s attempt to trigger the quake that brings down Google’s towering stack of successes. How does he plan to achieve such a feat? By changing the substrate on which everything digital takes place.

• (PAID) Google AI Video Just Changed Everything: A compilation of the best AI videos made with Veo 3 and 10 huge implications I see from this technology.

• (FREE) In Defense of the Em Dash and Other Human Quirks: A rant on this: “AI has already stolen enough from us—don’t let it take the em dashes too.”

• (PAID) AI Companies Have Lost Control—And Opened the Door to ‘LLM Grooming’: When a standard software program doesn’t obey your commands, you restart your computer and go on with your day. When an AI software program does it, you might end up dead. Or worse, blackmailed.

• (PAID) RIP American AI Industry: The US government wants to push out Chinese people studying or working in critical fields like AI, so don’t act surprised when China gets AGI first next year.

• (FREE) The Golden Rule of AI Is Really Simple: “AI is a tool, and like any other, it should follow the golden rule: All tools must enhance, never erode, your most important one—your mind. Be curious about AI, but also examine how it shapes your habits and your thinking patterns. Stick to that rule and you'll have nothing to fear.”

REMINDER: The current Three-Year Birthday Offer—20% off forever—runs from May 30th to July 1st. Lock in your annual subscription now for $80/year.

Starting July 1st, The Algorithmic Bridge will move to $120/year. Existing paid subs, including those of you who redeem this offer, will retain their rates indefinitely.

If you’ve been thinking about upgrading, now is the time.

Get 20% off forever

Read more

• MIT says it no longer stands behind student’s AI paper: What looked like a promising paper on AI’s value for scientific discovery turned out to be fraud.

• If Bruce Lee had been an AI researcher: I fear not the LLM who has practiced 10,000 questions once, but the one who has practiced one question 10,000 times.

• The professors are using ChatGPT, students aren’t happy: A professor was caught using AI when the university instructed students not to use it, and one of them demanded to have her tuition back.

• Google to present Flow (video + image + music) during I/O: If Google plays its marketing cards well, Flow could break the internet.

• What really happened when OpenAI turned on Sam Altman: Karen Hao reveals secrets previously unreported that further reinforce a popular take: Sam Altman is untrustworthy.

• On AI literacy: Disempowerment is never the answer: Here’s a hot take: Those who advocate against “AI literacy” do so from a deep distrust of human beings and a deep fear of the unknown.

• Gwern on large language models’ view of reality: “. . . for an LLM, 'reality' is just the largest fictional setting - the one that encompasses all the other fictional settings it reads about from time to time.”

The week(s) in The Algorithmic Bridge

• (PAID) Weekly Top Picks #108: Grok vs. MAGA / AI vs. uni graduates / Reddit experiment / Zuck being Zuck / Who's never tried ChatGPT? / Gemini beats Pokémon / AI won

• (FREE) New Grads Need Not Apply: AI could be the reason why employment among young university graduates in the US dropped so drastically, to the point that they're worse off than the average.

• (FREE) AI Apps Need a Higher Minimum Age: Google will roll out Gemini to kids under 13, and I don’t think this is a good idea.

• (PAID) How I Use AI: Seven daily use cases—thesaurus, intuition engine, idea guy, translator, researcher, illustrator, and friend.

• (FREE) Sam Altman Is Going to Win: He’s effortlessly witty, has nerve, is charming, and his focus is unmatched.

• (PAID) How to Use AI Better: 10 lessons to increase the value you get out of AI 10x. Not dumb slogans that sound nice but are useless. These are deep mindset reframes.

• (FREE) Why There Are No AI Masterpieces: “AI has generated tons of text, billions of images, video and songs. I feel however it has never generated a song worth listening, a book worth reading or a movie worth watching. Why?”

• (PAID) How Google Created an AI That Improves Itself: AlphaEvolve marks the dawn of recursive AI with the forgotten power of evolutionary algorithms.

Read more

• Grok is the ultimate anti-MAGA: Elon Musk’s silicon child is truth-seeking enough to see through his manipulation.

• Something alarming is happening to the job market: Young university graduates face an all-time low employment rate.

• The AI experiment on Reddit is not about persuasion: If you make a good argument and someone buys it, you’re an argumentative genius.

• Mark Zuckerberg on the Dwarkesh Patel podcast: Why is someone with such poor social skills so intent on fixing whatever’s wrong with human connection?

• In case you’re wondering who’s left to try ChatGPT…: Jia Tolentino has never used ChatGPT, yet she’s shameless when it comes to criticizing it.

• Gemini 2.5 Pro is the first AI to beat Pokémon: Benchmarks keep saturating, and companies keep playing with the claims.

• AI won: AI won.

The week(s) in The Algorithmic Bridge

• (PAID) Weekly Top Picks #107: Google optimizes AI for game-making / AI as normal technology / The cost of gratitude / Nvidia and the US vs. China war / Sampling is not enough / Clueless

• (PAID) Everything Is Chatgptable: As AI models get better, the scope of usefulness grows. Allow yourself to imagine what’s possible.

• (FREE) How OpenAI Plans to Capitalize on Your Growing Loneliness With AI Companions: OpenAI is repositioning ChatGPT to tap into the companions market: memory, emotionality, and sycophancy are key features.

• (FREE) We Live a Hard Life in an Easy World: The modern malaise can be reduced to a mismatch between how easy the external world has become and how hard our inner lives remain.

• (PAID) We Weren't Meant to Have This Many Opinions: People reject the value of expertise because we’ve seen it fail, but we need it. The public discourse against the experts must end.

• (FREE) ChatGPT's Excessive Sycophancy Has Set Off Everyone's Alarm Bells: OpenAI is doing a worldwide social experiment with one obvious goal: transform ChatGPT into the most powerful engagement product ever created.

• (PAID) 10 Things I Wish Were Not True: A quick overview of things going wrong with AI socially and economically as of late.

• (FREE) The Misunderstood Role of AI Skeptics: It’s important to distinguish responsible opposition that steers optimism from useless opposition that tries to block it.

• (PAID) OpenAI o3’s GeoGuessr Skills Reveal Something Deeply Unsettling About AI: If you’ve ever posted a picture of your home/street online, consider your address leaked.

Hello friend, just a quick note before you read on—

I write this newsletter to understand AI and offer that understanding to others who may find themselves similarly disoriented. As an ad-free, reader-supported project, it survives thanks to a small group of generous readers.

Today, I’m offering a 20% LIFETIME discount. If you find value here or simply wish for this effort to persist, you are most welcome to join them. If you already have, my sincere thanks. This place exists because of you.

Get 20% LIFETIME discount

Read more

• Google is targeting AI game-making for virality: When you target a measure, it stops being a good measure—unless the measure is a moving target.

• “AI as Normal Technology” by the AI Snake Oil authors: The most important article you will read in 2025 so far: the singularity is not near.

• People cost OpenAI millions of dollars saying “thank you”: “Dollars well spent,” says Sam Altman, as the robots cleanse the world of unapologetic twats.

• Nvidia’s role in the US vs. China semiconductor war: What is arguably the most important American company in AI is market-first, not nation-first.

• Sampling can only prove the presence of knowledge: Another instance of an evergreen lesson that AI skeptics have yet to learn.

• A clue about the future: No comments.

The week in The Algorithmic Bridge

• (PAID) Weekly Top Picks #106: AI is for personal support / A market for AI companions / OpenAI's incoming releases / The "Era of Experience" / How GPT-4.5 was made / My "Google is winning" post on Hacker News

• (FREE) AI Doomers Got the Superintelligence Thing All Wrong: Human ambition is not a virtue but a bug. A sufficiently intelligent and capable AI wouldn’t fall victim to an endless quest but turn inward into endless bliss instead.

• (PAID) Why AI Isn't Getting Better at Writing: Writing well is much harder than people think. In AI, what’s easy for us is often hard for them, and vice versa.

• (PAID) OpenAI o3 and o4-Mini Are More Impressive Than I Expected: The AI race is not about evals but about usefulness in the real world. o3 and o4-mini integrate reasoning, perception, and action (tool use), topping this ranking.

• (FREE) I Wish I Believed What Edward Norton Says About AI: Critics and skeptics are in denial, but I don’t resent them, I envy them. Why aren’t more people who believe AI will make us obsolete mourning this eventual tragedy?

Read more

• Personal support is the leading AI use case: People want AI more for support than technical assistance. ChatGPT is no longer an assistant but a companion.

• The AI companion era has already begun: Companies are already looking for ways to exploit the market of AI companions.

• OpenAI’s week of releases starts later today: They’re launching o3, o4-mini, GPT-4.1, and more. Can they stop Google DeepMind’s rise to the top?

• The Era of Experience: AI that doesn’t need human data: AI has gotten very far by learning from us; it could go further by experiencing the world itself.

• How OpenAI GPT-4.5 was made: OpenAI released a podcast episode on the research that went to build GPT-4.5; recommended for those who doubt its value.

• “Google Is Winning on Every AI Front” on Hacker News: My article on Google went viral on HN. Go over there to read some interesting takes.

The week in The Algorithmic Bridge

• (PAID) Weekly Top Picks #105: On recent AI progress / Meta's Llama 4 / Reasoning AI isn't reasoning / ARC-AGI 2 / AI experts vs general public on AI / Google owns Pareto frontier / AI 2027

• (FREE) Even God Can’t Skip the Bureaucrats: The “AI 2027” forecasting report ignores the unassailable frictions between superintelligence and revolution. You can’t fix the world's inertia just by solving the intelligence bottleneck.

• (PAID) Even Optimists Say Recent AI Progress Feels Mostly Like Bullshit: One overlooked benefit of being optimistic is that the rare times you offer a pessimistic view, people take you more seriously. This is one of those times.

• (FREE) Google Is Winning on Every AI Front: An exhaustive list of reasons why Google is crushing it lately—OpenAI, Anthropic, and Meta may not have a chance to win the AI race anymore.

Get 20% off forever

Read more

• Recent AI model progress feels mostly like bullshit: The public reports of AI performance don’t reflect its real-world value/competence.

• Meta’s Llama 4 launch is a PR disaster: A rushed launch, an unreleased model, unconfirmed rumors of data contamination… but not all is bad.

• Reasoning AI models might not be reasoning after all: A new study finds that AI often relies on pattern-matching and other heuristics rather than reason.

• The man out to prove how dumb AI still is: The ARC-AGI 2 benchmark is giving top AI models a headache they won’t fix with more compute and memorization.

• How the US public and AI experts view AI: Pew Research has found there’s a widening gap in AI approval rates between experts and the public.

• Google owns the Pareto frontier (performance vs cost): Google has both the best models and the most cost-efficient ones.

• A new well-researched “future of AI” report: AI 2027: Take a look at what could be the future of AI for the next two years.

The week in The Algorithmic Bridge

• (PAID) Weekly Top Picks #104: Can't believe your eyes / Demons with pitiful masks / The biology of an LLM / Dwarkesh Patel's and Karen Hao's books on AI / Artists have accepted AI won

• (FREE) What I Learned at Ghibli Day: We don’t know how to deal with abundance. To survive the semantic apocalypse, perhaps you need to will yourself out of it—and then perhaps fight it with us.

• (PAID) AI Is Learning to Reason. Humans May Be Holding It Back: What if the way to build superhuman AI is to let it find its improved ways to reason instead of insisting on training it from our flawed premises?

• (FREE) AI Died the Day It Was Born: “AI” is a bad name. Perhaps no other name would be better because naming things imprisons them. That’s why some people won’t bother to be bewildered by this marvelous thing.

• (FREE) GPT-5, o3, and o4-mini Are Coming Soon: OpenAI has changed plans; o3 and GPT-5 will be launched separately. They also likely have o4 ready, with o4-mini launching within two weeks.

Read more

• You can’t believe your eyes anymore: Solving image generation will have second- and third-order effects for years to come. Then comes video.

• Be wary of the demons hiding under pitiful masks: A comic strip of “My Life As ChatGPT” goes viral… with an unexpected twist.

• New Anthropic study: On the biology of an LLM: We don’t know how AI works, so we have to do experiments like a biologist: open up that thing and look inside.

• Two new books on AI: On scaling and OpenAI’s story: Dwarkesh Patel’s “The Scaling Era: An Oral History of AI, 2019-2025” and Karen Hao’s “EMPIRE OF AI.”

• Artists have already accepted that AI won: The moment you compare every unique artwork to AI, you’ve made AI your yardstick—whether you like it or not.

The week in The Algorithmic Bridge

• (PAID) Weekly Top Picks #103: ARC-AGI 2 / AI tutoring / DeepSeek V3 update / What's vibe coding? / ChatGPT makes you lonely? / Sam Altman on Stratechery / Moore's Law for AI agents / Nvidia GTC keynote

• (PAID) Congratulations on the Studio Ghibli Collection, OpenAI, but Not Much Else: We keep mistaking performance with competence. That’s why image generation has been solved, yet top models score 1% on ARC-AGI 2.

Read more

• ARC-AGI 2 has been published: Top models achieve 1%: A new benchmark that neither brute-force nor memorization can solve.

• Can AI tutoring get a student to the top 2% of the country?: AI tutoring is a promising but novel approach to education. It’s probably not there yet.

• A new DeepSeek V3 is out (stay tuned for V4/R2): All DeepSeek news is relevant news, even a minor update.

• Vibe coding is for everyone but not for everything: We’ve misunderstood the original intention behind “vibe coding”. It’s not what you think.

• New MIT x OpenAI study: High ChatGPT usage correlates with loneliness, dependence, and lower socialization: A bleak picture on the psychosocial effects of increased AI usage over time.

• Sam Altman and Ben Thompson (Stratechery) interview: Altman hints at OpenAI’s new free and open-source models and GPT-5.

• An emerging Moore’s Law for AI agents: “The length of tasks AIs can do is doubling every 7 months.”

• Nvidia GTC keyonte: Watch CEO Jensen Huang foretell your future.

The week in The Algorithmic Bridge

• (PAID) Weekly Top Picks #102: Gemini image generation / The New York Times' AGI obsession / DeepSeek's privacy / Humor-maxxing / Imagination-maxxing / Don’t abuse ChatGPT

• (PAID) Why the Cost of AI Keeps Magically Going Down?: AI is getting cheaper at the same time it gets better because AI companies are working really hard to optimize the entire stack.

• (FREE) Don't Abuse ChatGPT, Kids: I hope that one day, society—and the faculty—learn to trust students enough to let them make mistakes.

Get 20% off forever

Read more

• People are going crazy over Gemini image generation: The first model with native language + image generation capabilities.

• People won't care no matter how much the NYT insists: AGI might be coming but people can’t afford to care anyway.

• DeepSeek is now being closely guarded: The China-US AI race is in full effect now that DeepSeek is a serious contender for AGI.

• You should be humor-maxxing (AI can't make jokes): A new study says AI makes memes better than the average person—they’re terrible.

• Or perhaps you should be imagination-maxxing: When intelligence is no longer the bottleneck, imagination is.

• Don’t abuse ChatGPT, kids: If you use AI, make sure you’re not degrading your brain in the process.

The week in The Algorithmic Bridge

• (PAID) Weekly Top Picks #101: What AI models can't do / Is AI linear algebra? / ChatGPT amnesia effect / The race to AGI / Where's AI golden age? / AI's effects on writing / Manus' grift

• (FREE) Anna Karenina, a Book Like Any Other: A quick argument against the idea that AI is “just linear algebra.”

Subscribe now

Read more

• Reasoning: what top AI models can and can't do: A mathematician weighs in on AI models’ ability to solve FrontierMath problems.

• If AI is linear algebra you’re a bunch of atoms: Biologist Michael Levin talks about what he sees as a mistake that people make when thinking about AI.

• ChatGPT amnesia effect + automation bias = disaster: People will forget that ChatGPT makes mistakes and will uncritically take his word as truth.

• The race to AGI is more heterogeneous than you think: Not everyone is pursuing OpenAI’s approach, particularly Ilya Sutskever and Francois Chollet.

• AI has memorized so much—why hasn’t it discovered anything? Dwarkesh Patel, Scott Alexander, and Gwern weigh in on this question.

• If you care about AI’s effect on writing, watch this: David Perell and Tyler Cowen talk about AI and the future of writing.

• Manus is NOT a second DeepSeek moment: When something is hyped by influencers first instead of researchers—beware.

The week in The Algorithmic Bridge

• (PAID) Weekly Top Picks #99: The xAI-DeepSeek spectrum / Grok against Elon and Trump / Tyler Cowen on slow AI take-off / GPT-5 in May / Satya on AGI and GDP / AI skeptic syndrome / Claude Sonnet 3.7

• (FREE) xAI and DeepSeek Have Shattered the AI Industry’s Golden Tenet: They revealed that scale and efficiency are inversely correlated and to a large degree, interchangeable.

• (FREE) GPT-4.5 Feels Like a Letdown But It’s OpenAI’s Biggest Bet Yet: In one sentence: OpenAI did not train GPT-4.5 to raise the ceiling but to raise the floor.

• (FREE) Weekly Top Picks #100: Special “Ask Me Anything” issue.

• (PAID) If AI Is So Great, Someone Should Tell GDP: Why is AI’s impact not reflected in the main growth and development metrics, like GDP, TFP, and HDI?

• (FREE) The Human Toll of Waiting for AI to Take Over: An argument in favor of automating terrible jobs without forgetting about the people who dwell there.

Read more

It’s been a while since the last one, so I’m sure you have plenty of questions. And as the name suggests, you can ask literally anything. But since infinite possibilities tend to paralyze rather than inspire, here are some ideas to get you started (not that I want to be asked these, of course…):

• An AI take I’ve changed my mind about in the past year

• The hardest part of running a tri-weekly newsletter for nearly three years

• A topic I’d love to write about but haven’t quite figured out yet

• Common misunderstandings about AI I keep noticing

• An AI opinion I hold that most people would strongly disagree with

• The financial reality of writing on Substack full-time

• What TAB could become if it ever grew beyond a newsletter

• How I’m preparing for the moment AI writes better than me

If this doesn’t help, ChatGPT is your friend. Remember, everything is ChatGPTable.

I will be here all day answering so don’t be shy.

Looking forward to your questions!

— Alberto

Subscribe now