What I Am Reading:
Why your personal brand is your most valuable asset
Darius Foroux makes a case I have watched play out in my own career: as the economy shifts from need to want, who you are becomes more valuable than what you do. When products and skills get commoditized, the differentiator is trust, and trust attaches to people, not companies. The piece is a useful reminder that reputation is not vanity. It is the compounding asset that opens doors before you walk through them. I have seen this firsthand in venture, where the quality of your name determines the quality of your deal flow.
How To Plan Your Day Like Marcus Aurelius
Two thousand years ago the most powerful man in the world kept a private journal of reminders to himself on how to live. What strikes me about Marcus Aurelius is not the philosophy in the abstract but how operational it was. He used the morning to prepare for difficulty and the evening to review where he fell short. That structure, intention at the start and honest reflection at the end, is something I have tried to bring into my own routine, and it holds up remarkably well across two millennia and a completely different life.
The Society of Thought Hiding Inside Every Reasoning Model
This is one of the more fascinating findings I have come across this year. When you ask a frontier reasoning model a hard question, it does not simply think longer. It spontaneously generates an internal debate among distinct perspectives that argue, question, and reconcile, what the researchers call a society of thought. Nobody trained the models to do this. It emerged on its own when they were rewarded purely for getting the right answer. The implication is profound: robust reasoning may be inherently social, even inside a single mind, and the path to better AI may run through composing richer societies rather than building one larger oracle.
The Longevity Drug That May Be Quietly Undercutting Your Workouts
Rapamycin is the molecule longevity researchers love most, the one drug that reliably extends lifespan in mice across labs and doses. So this new study published in the Journal of Cachexia, Sarcopenia and Muscle gave me pause. In older adults following an exercise program, the rapamycin group performed worse on functional measures like the chair-stand test, and showed elevated inflammation markers, suggesting the drug may blunt the gains you would otherwise get from training. It is a useful corrective to the hype. The biology of aging is full of tradeoffs, and the intervention that helps a sedentary mouse may quietly work against an active human.
The Quantum Clock Is Already Ticking on America's Autonomous Arsenal
The threat here has a deceptively benign name: harvest now, decrypt later. An adversary does not need a working quantum computer today to compromise encrypted military data today. It only needs to intercept and store the command and control traffic now, then decrypt it once the hardware matures. The piece argues the Pentagon is building a generation of autonomous drones and munitions on cryptographic foundations a future quantum computer will break, and that a drone designed in 2026 to fly into the 2040s cannot ship with encryption expected to fail in the 2030s. This is the part of the quantum story that actually keeps me up at night, and it sits right at the intersection of two areas I spend a lot of time thinking about.
Why World Models Are AI's Next Frontier
Large language models predict the next word. World models try to understand how the physical world actually works: space, physics, cause and effect. The distinction matters because true general intelligence requires the ability to revise your internal model when conditions change, something LLMs struggle with the moment they leave the distribution of their training data. Yann LeCun believes in this so strongly that he left Meta to pursue it, predicting that within three to five years nobody in their right mind will use LLMs of the type we have today. Whether or not that timeline holds, world models are where physical AI and robotics get genuinely interesting, and where I think a lot of value gets created over the next decade.
Magazine Architect
Morgan Housel built this one around a single phrase he heard and could not shake. A magazine architect is someone who designs a house to look impressive in photographs rather than to actually be lived in, and the metaphor extends to almost everything: careers, companies, public personas optimized for how they appear rather than how they function. The most successful people he knows are almost the opposite, willing to look unimpressive in the moment in exchange for substance that compounds. It is a short read and a sharp one, and it lingers longer than its length suggests.
What I Am Listening to:
BI 237 Ehud Ahissar: Consciousness and Perceptual Dualism
Brain Inspired remains my favorite show at the intersection of neuroscience and AI, and this episode is a good example of why. Ehud Ahissar from the Weizmann Institute spent years studying how rodents sense the world through whisking, and it led him to a genuinely strange and compelling theory of consciousness. He proposes that we communicate through a non-physical digital process while we experience the world through a physical analog one, and that these map onto opposing loops of brain circuitry. It is dense, it is original, and it is the kind of idea you will not encounter anywhere near the mainstream conversation about consciousness.
Follow Up on Bioregulators: Answering Your Questions With Nat Niddam and Erin Ryan
Nathalie Niddam runs one of the most technically serious longevity podcasts out there, and this bioregulator episode is a deep one. Bioregulator peptides are short chains that appear to signal specific tissues to repair and regulate themselves, and the research history behind them, much of it out of Russian gerontology labs, is fascinating and almost completely unknown in the mainstream wellness world. Whether or not every claim holds up, the level of detail here is far beyond what you get from the typical health show. This pairs especially well with the rapamycin piece above, because both are really about the same question: which longevity interventions actually do what they promise.
COMPLEXITY: Can Machines Ever Truly Understand? With Alison Gopnik and John Krakauer
The Santa Fe Institute podcast is where I go when I want the smartest people in the world arguing about hard problems without dumbing anything down. This conversation tackles a question I keep circling: what would it actually mean for a machine to understand something, as opposed to imitate understanding. Gopnik's work on how children learn through innovation rather than imitation draws a sharp line that current AI systems have not crossed, and Krakauer pushes on what intelligence even is. It is humbling listening, the kind that makes you realize how much we still do not know about the thing we are all racing to build.
What I Am Watching:
Conan O'Brien Delivers the Harvard Commencement Address
Conan O'Brien returned to Harvard, where he graduated in 1985, to deliver the commencement address to the class of 2026. It is exactly what you would hope for: very funny, and then quietly moving in the places that matter. His core message, that the things you fear will define you rarely do, and that reinvention is not a failure but the entire point, lands harder coming from someone who has been publicly knocked down and gotten back up more than once. Worth the time even if you have heard a hundred commencement speeches.
Historian Emily Kate Genatowski spent a year living with an AI-powered robot roommate and came back with five lessons that cut through almost all the noise in the humanoid robotics conversation. The most useful thing about her account is how unglamorous it is. The future of robots in the home is not science fiction, it is practical, messy, and already partly here, full of quirks and mundane friction alongside the genuinely useful moments. As someone investing in physical AI, I found her ground-level view more clarifying than most of the polished demos that go viral, precisely because she is honest about what does not work yet.
Daniela Rus: How AI Will Step Off the Screen and Into the Real World
Daniela Rus runs the largest computer science lab at MIT, and she is one of the clearest thinkers on where AI and robotics actually converge. Her argument is that the next chapter of AI is physical, machines that engage dynamically with the real world rather than just generating text on a screen. She introduces liquid networks, a class of AI modeled on the neural processes of simple organisms, that can run far more efficiently than today's massive models. For anyone trying to understand why physical intelligence is the frontier worth watching, this is a grounded, technically serious case from someone actually building it. It connects directly to the world models piece in this month's reading list.
