Reading List
What I'm reading, citing, or recommending—books through feeds—with a brief note on each.
One of the Santa Fe Institute's founding figures makes the rigorous case against equilibrium economics—and for agent-based models that actually capture how markets, pandemics, and transitions unfold. The most intellectually honest treatment of complexity I've encountered.
Five evolutionary breakthroughs—steering, reinforcing, simulating, mentalizing, speaking—mapped onto five hundred million years of brains. A tempting, and surprisingly sturdy, analogy for the trajectory I keep seeing in agentic AI today.
Logical, creative, and critical thinking, traced through the long history of computation—a corporate philosopher's case for what ChatGPT-era tools are doing to our mental models.
Jarvis's case that print was a five-century parenthesis between two ages of participatory, oral-like information sharing—the internet isn't the aberration. It's the return.
Beyond resilience: systems that don't just survive disorder but get better from it. The frame I keep reaching for whenever 'robust roadmap' starts to sound like the wrong goal.
The book that made me permanently suspicious of any plan that doesn't budget for the event nobody saw coming.
Once content is infinitely replicable at near-zero cost, the article stops being the product—the only way out is to productize the process instead: verification, truth-seeking, sense-making.
A pope trained as a mathematician's first encyclical, on AI and the dignity of the human person—arguing technology must be 'disarmed' from the logic of competition before it can serve rather than govern.
The piece that put a name on something I'd been circling for months: the durable asset in enterprise AI isn't the model, it's everything you build around it.
Planner / Generator / Evaluator as a pattern. Useful well beyond coding agents—it maps cleanly onto how a digital programme should be structured.
Civilization's adolescence: autonomy risk, misuse, economic disruption—serious, but not inevitable, if governance keeps pace.
The Vatican's most complete reflection on AI before Magnifica Humanitas—ancient wisdom and new technology, read together rather than against each other.
The case that powerful AI could compress a century of scientific progress into a decade—if equitable distribution and alignment are treated as first-order problems, not afterthoughts.
(de Gergely Orosz) The most rigorous reporting on how engineering teams actually work—compensation, career ladders, architecture decisions, big-tech culture. The gap between the official story and Orosz's version is usually where the signal is.
Product judgment, AI sovereignty, multilingual accessibility—written with the kind of rigour and ethical seriousness that should be more common in this field. Her thinking on what French-language AI means in practice is a necessary read.
A Wharton professor turning serious AI research into actionable insight—without dumbing it down. The most consistently useful newsletter on what AI actually does to work, learning, and organisations.
One of the more intellectually honest trackers of where AI actually is versus where it's claimed to be. Worth the friction, especially on regulatory gaps and the distance between benchmark performance and real-world reliability.
The signal-to-noise ratio remains the best of any tech aggregator. The comments often outperform the articles they discuss—technically precise, sometimes brutal, always worth the scroll.
The quarterly essays are a goldmine of structural thinking on tech cycles—built as slide decks, dense with charts, but worth more than most written reports ten times their length.
AI, digital transformation, and their human implications—the paradoxes of automation, leadership, and meaning at work.
Satya Nadella's personal blog—longer-form thinking than his social posts, and considerably more careful. The Reverse Information Paradox essay is the best recent example of what happens when a CEO actually does the economic reasoning.