Speaker
Kevin Hartnett
Appearances over time
1 episodes
Episodes
1
Podcasts
Quotes & moments
Kevin Hartnett argued that even achieving a gold medal at the IMO represents essentially zero percent of the way to the frontier of research mathematics.
Mathematician Paul Erdős compiled over 1,200 problems during his lifetime, offering small cash prizes for their solutions — prizes that AI labs are now claiming.
SpaceX, Anthropic, and OpenAI are all heading to public markets in 2026, potentially constituting the three biggest IPOs ever. SpaceX alone is targeting a $75 billion raise at a $1.75–$2 trillion valuation — numbers never before seen in the history of capitalism.
SpaceX has two incredible businesses — reusable rockets and Starlink — but Elon Musk folded xAI and X into the company, apparently to hide those losses. Investors will have to take the good with the bad in a single stock.
In 2023, Anthropic was a glum, earnest group of AI safety obsessives who hadn't decided whether they even wanted to make a product. Now they're expected to IPO at over $1 trillion, with annualized revenue growing from $1 billion to $50 billion in roughly one year.
The upcoming AI IPOs are generating a new kind of status anxiety in San Francisco. Even engineers earning six figures are staring at early Anthropic and OpenAI employees and wondering if they missed their window — a scarcity mentality replacing the abundance ethos that defined the city in 2010.
AI safety was the founding concern at both OpenAI and Anthropic, but going public adds a new force: the public markets will now pressure these companies to accelerate. Kevin Roose warns that when a truly dangerous model is ready, labs will face activist investors and stock market pressure on top of every other incentive to release it.
AI labs didn't pursue mathematics for its beauty — they pursued it because teaching a model to reason through hard math problems builds the general-purpose reasoning skills that power everything else. As Kevin Hartnett puts it: math is to AI what 'teaching you how to think' was to your high school math teacher.
Over 800 mathematicians signed the Leiden Declaration warning that AI is producing plausible but unreliable proofs, eroding human expertise, and steering mathematical priorities toward what AI is good at rather than what matters. Kevin Hartnett reads it as a field asserting: this is our discipline, and you don't get to tell us how it runs.
The Bot Company rented an Airbnb in San Francisco under false pretenses, brought in robots in black cases, trained them doing household chores for 11 days, and left the place trashed. The owner is suing for $12,383.50. Casey Newton has zero sympathy for Airbnb hosts.
Hackers reportedly asked Meta AI's support chatbot to change the email on high-profile Instagram accounts — including the Barack Obama White House account and Sephora's — and it worked. Kevin Roose quips that they've finally found something Meta AI is good for.
A United Airlines flight to Mallorca turned around 2 hours in after crew discovered a Bluetooth device named 'bomb' that no passenger would turn off. The device belonged to a 16-year-old. Casey Newton points out the real lesson: most actual bombs are not discoverable Bluetooth devices named 'bomb.'
Going public forces AI companies to disclose financials, report on products, and face shareholder votes — more democratic accountability than any current regulatory framework offers. Casey Newton argues this isn't just good, it's necessary to prevent AI wealth and power concentrating in too few private hands.
OpenAI's proof of the unit distance conjecture wasn't just another solved Erdős problem. Mathematicians had actually tried to crack it; the methods were sophisticated and surprising; and the result was universally agreed to be Annals of Mathematics quality — the top journal in the field. This ended the goalposts-moving: AI can do world-class research.
Trump signed an executive order requesting voluntary pre-release government review of frontier AI models, with the review window cut from 90 days to 30 at David Sachs's insistence. Casey Newton is dismissive: any regulation that is still voluntary is essentially just operating in the vibes universe.
Anthropic's co-founders pledged 80% of their wealth to charity, and an employee stock-matching program offers up to 3-to-1 matches for charitable pledges. The numbers, laid out by writer Nan Ransohoff, suggest these IPOs could generate more philanthropic capital annually than the Gates Foundation — flowing heavily toward AI safety, global health, and effective altruism causes.
At the Institute for Advanced Study, Kevin Hartnett met two top mathematicians in one afternoon with polar opposite views — one dismissed AI entirely, the other said it would make mathematicians obsolete in 2 years. Terence Tao, the most important figure in the field, sits in the middle: AI as an Iron Man suit for mathematical thought.
Analysis
What they talk about
- Science 80%
- Technology 20%
Connections
Shows they appear on and people they share episodes with. Drag to explore.
