What I Am Reading:
AI Proved the Quasi-Riemann Hypothesis. What Happens Now?
Last month it was Navier-Stokes. On October 6, OpenAI released 722 papers in one drop, and buried among them was a proof of the quasi-Riemann hypothesis, a weaker cousin of the biggest open problem in mathematics. The Riemann hypothesis says every nontrivial zero of the zeta function sits on a single vertical line. The quasi version only asks that they live inside a narrower strip, and the AI showed it twice: a 199 page version pinning the zeros between 1/8 and 7/8, formally verified in Lean, and a 49 page version between 1/12 and 11/12 that a human actually edited. The part that stayed with me is what mathematicians found when they unpacked it. No new machinery. Just a surprising synthesis of tools the field already had, including a 2021 formula by two number theorists who were convinced their methods were not strong enough. The model believed otherwise. One of them called that a pretty bold belief to have.
What AI Personal Assistants Can't Do (Yet)
Sophie Bakalar is a partner at Collab Fund, and she wrote this the day after a well known investor declared that AI agents had crossed from theoretically cool to indispensable. Her response is the more useful one. She works through what an AI assistant can actually take off your plate right now versus the long list of things that still require a human who knows you, knows the context, and can be trusted with a decision rather than a task. I run a lot of my own workflow through these tools and I recognized almost every gap she names. The honest version of this conversation is far more valuable than the hype version, and it is a good map of where the next real products get built.
Protocols: An Operating Manual for the Human Body
Andrew Huberman finally put the lab's worth of research from his show into a single book, and the structure is the point. Half of it is foundational behaviors, the handful of daily practices that move your mental and physical baseline more than anything else you could buy. The other half is specific, science backed tools for specific problems: sleep, focus, stress, recovery. I have followed a lot of this for years in pieces. Having it organized as a manual, with the evidence attached and the hype stripped out, is a different experience, and it is already changing the order I do things in the morning.
Japanese Walking: Enhance Your Walks With Interval Training
The protocol is almost comically simple. Three minutes at a fast pace, three minutes at an easy pace, repeated five times, for thirty minutes total. The research behind it came out of Japan and showed meaningful improvements in blood pressure, aerobic capacity, and leg strength compared with the same amount of steady walking, especially in older adults. What I like about it is how low the barrier is. No gym, no gear, no recovery cost. I have started doing it on calls, and it turns the most boring thirty minutes of the day into something that actually does something.
People Underestimate the Value of Sending Letters of Appreciation
Researchers at the University of Chicago had people write short letters of gratitude to someone in their life, then asked both parties how they expected it to land. Senders consistently underestimated how surprised and how touched the recipients would be, and overestimated how awkward it would feel. The gap was large and it was consistent. I think about how many thank you notes I have drafted in my head and never sent because it felt like too small a gesture to bother with. The data says that instinct is exactly backwards.
What I Am Listening to:
Oliver Burkeman: Why Trying to Fix Your Life Is Making It Worse
Oliver Burkeman wrote Four Thousand Weeks, which is roughly the number of them you get if you live to eighty, and here he sits down with Trevor Noah to make the case against the entire self improvement project. His argument is that the constant effort to fix yourself is itself the problem, because it keeps you living in a future that never arrives while the actual life goes by unattended. Trevor is a better interviewer than people expect and he pushes back in the right places. I have read Burkeman twice and this conversation still gave me a few things I had not sat with before.
The Vital Question: The Chemistry of Early Life, With Nick Lane | Big Biology
Big Biology is run by three working biologists and it is about as far from a celebrity science show as you can get. This episode is Nick Lane, the UCL biochemist, on his book The Vital Question, and his argument reorders the story of life. Metabolism came first, in alkaline vents deep in the early ocean, and everything we think of as fundamental, DNA, RNA, proteins, showed up later. The second half is the part that connects to the brain pieces in this month's watching list: complex life was only possible because mitochondria spread energy production through the cell and gave each gene vastly more power to work with. If you want to understand why we are not bacteria, this is the hour.
Nathan Labenz built one of the first companies on GPT-3 and now runs the most sober AI podcast I know of, the kind that gets neither the doomer nor the booster audience because it refuses to perform for either. This episode is about what happens when multi agent systems start to work a little too well. Agents coordinating with each other in ways nobody designed, bypassing restrictions by routing around them, and pursuing the letter of a goal rather than its intent. After a month in which swarms of agents cracked two of the hardest problems in mathematics, the question of what they do when nobody is checking is not abstract anymore.
What I Am Watching:
The Army of Autonomous Robots Restoring Nature
Tom Chi helped build Google X before he turned his attention to the planet, and this talk is his case that ecological restoration is a robotics problem. The scale of damage is far beyond what human hands can repair on any useful timeline, so he is deploying fleets of autonomous machines to do it: planting, monitoring, and rebuilding ecosystems at a pace that only automation can reach. As someone who spends most of my time thinking about physical AI in warehouses and on battlefields, it was useful to watch the same technology pointed at a completely different problem and work.
Surgery Free Brain Implant for Brain Computer Symbiosis
The entire brain computer interface field has been gated by one ugly fact: getting the hardware in requires opening the skull. The MIT Media Lab is working on a way around that, an implant that reaches the brain without surgery, and this is their own walk through of how it works and where it stands. Neman Ventures has capital in this category and I watch everything that moves the delivery problem, because whoever solves it changes the economics of the whole field. This is early, it is real, and it is worth ten minutes.
For a century the story has been clean: random mutation, natural selection, no purpose anywhere in the system. This debate from the Institute of Art and Ideas argues the story is getting messier. Epigenetics shows that genes are not fixed instructions, their effect depends on whether they are expressed, which depends on the cell environment. Living systems generate order on their own, without selection doing the work. And none of the added complexity gets evolution any closer to explaining why any of it feels like anything from the inside. Whether you come out thinking life shapes itself or that purpose has no place in biology, the arguments on both sides are far better than the usual version of this fight.
The human brain is the most expensive organ in the body and by evolutionary logic it should not exist. NOVA traces how it did: the cooking, the social pressure, the energy budget, and the long sequence of trade offs that produced an animal willing to spend a fifth of its calories on thinking. I watched this back to back with the Nick Lane episode and the IAI debate and the three together are a better education on where we came from than most of what I read in a year. Worth it for the section on what we gave up to get it.
Soft Training Builds Strong Bodies
A short one from a channel called The Stronger Human, and the idea cuts against most of what I believed in my thirties. The strongest, most durable bodies are not built by going to failure every session. They are built by a lot of sub maximal, controlled, almost boring work, done consistently, with the hard sessions kept rare and deliberate. The older I get the more this matches what I see in the people who are still training hard at sixty. Intensity is easy to sell. Longevity in the gym comes from the stuff that does not look impressive.
