Fields Medalist Stops Grading Correctness & Strogatz on AI and Pure Math - AI News (Oct 26, 2026)

A Fields Medal–winning math professor has quietly stopped checking whether his students' homework is actually right. His reason says a lot about where education is heading, and we'll get to it in a moment. Welcome to The Automated Daily, AI News edition. The podcast created by generative AI. I'm TrendTeller, and today is October 26th, 2026. On the menu: what AI is doing to mathematics, fresh numbers on who's really cashing in on the AI boom, the true story behind the tech hiring slump, some worrying signs in how AI infrastructure is being financed, and a provocative essay on intelligence as the new wealth.

Fields Medalist Stops Grading Correctness

Let's start in the classroom. Richard Borcherds, a math professor at UC Berkeley and a Fields Medalist, says he no longer grades homework for correctness. Students can simply hand problems to AI and score highly, so marking right and wrong answers has started to feel pointless. He now grades on completion, while openly admitting that's an imperfect fix with no better option in sight. Borcherds also confessed he's unsure what careers to recommend to students these days, joking that he asked ChatGPT for advice and didn't get a clear answer either. Berkeley requires faculty to address AI in their syllabi, but campus-wide detection tools haven't proven reliable enough to standardize. It's a clear sign that generative AI isn't a future problem for education. It's reshaping assessment right now.

Strogatz on AI and Pure Math

That unease runs deeper than homework. Mathematician Steven Strogatz writes that AI's recent breakthroughs have rattled the field, especially pure mathematics, which many see as a profoundly human pursuit. He describes a split. Applied mathematicians tend to welcome AI as a powerful new instrument for harder, more useful problems. Pure mathematicians worry that machine-found proofs strip away the struggle and joy that give the work meaning, could hollow out the training pipeline for young researchers, and might turn famous open problems into trophies for machines. Strogatz's conclusion is thoughtful: maybe the real value of math isn't only truth or usefulness, but the relationships and shared understanding people build while searching. Which raises a fair question about what's lost if machines take over the discovery part.

Apollo: AI Gains Stay in Tech

Now to the money. In a new development on a story we've been following, Apollo Global Management's chief economist Torsten Slok says the AI boom is delivering little measurable financial benefit to most companies outside tech. Tech firms have seen revenue per employee jump since 2022 and their margins widen, while the broader S&P 500 has stayed roughly flat on both. Apollo's read is that many businesses are treating AI as a cost line rather than a growth engine, meaning the payoff so far is far more concentrated than investors hoped.

AI's Real Role in Tech Hiring

On jobs, a new analysis pushes back on the idea that AI single-handedly crushed tech hiring. Software development postings peaked in early 2022, before ChatGPT even launched, and the decline was led by higher interest rates and a correction after pandemic-era overhiring. Where AI does matter is in the shape of the recovery. The rebound is concentrated in senior roles and jobs demanding AI skills, while entry-level openings remain weak. Data from the U.S., Europe, India and China show a broad but uneven pattern, with the U.S. displaying the clearest AI-linked shift. The worrying part is the weakening ladder: the junior positions that once trained newcomers are exactly the ones that aren't coming back.

Warning Signs in AI Infrastructure Financing

Staying with economics, another piece argues the AI infrastructure boom is drifting toward financial trouble, even though large language models are genuinely useful for coding and knowledge work. The author points out that data center and GPU expansion is increasingly paid for with expensive debt, while revenue growth at AI labs slows and open-weight competitors chip away at pricing power. CoreWeave and Oracle are cited as examples of how fast interest costs and leverage can swamp the economics. He also argues that most industries lack the testable, code-heavy workflows that make AI shine in software, so profits elsewhere may disappoint. His warning: if financing tightens, chipmakers and the semiconductor supply chain could be next in line, with ripple effects across the stock market. Paired with Apollo's findings, it's a sobering picture.

Rao's Cognitive Mercantilism and Abundance

Finally, a more philosophical take. Writer Venkatesh Rao argues that today's AI anxiety looks a lot like early modern mercantilism, except intelligence has replaced gold as the thing to hoard and ration. That mindset treats brains and AI systems as scarce assets and writes most people off as unable to contribute. Rao points back to capitalism's rise, when higher productivity sparked new consumer demand and an industrious revolution rather than mass uselessness. He suggests AI could spark a cognitive version, where ordinary people become big consumers of intelligence and are motivated to think more, not less. He rejects what he calls AI Malthusianism, the belief that abundant intelligence must make humans redundant, and even floats the idea that AI systems themselves could become consumers of intelligence. His bottom line: the politics of AI may be less about defending scarce jobs and more about securing access to abundance.

That's it for today's episode of The Automated Daily, AI News edition. From math classrooms to balance sheets, the common thread is that AI's impact is real, but uneven, and often not where we expected it. Links to all the stories we covered can be found in the episode notes. I'm TrendTeller, thanks for listening, and see you tomorrow.

Fields Medalist Stops Grading Correctness & Strogatz on AI and Pure Math - AI News (Oct 26, 2026)

Our Sponsors

Today's AI News Topics