A Video Companion to the Field Guide

Altman is the public face of the company that put this technology into hundreds of millions of hands. In the TED conversation he describes growth he says he has never seen in any company, sketches a near future in which the models become something like extensions of ourselves, and takes pointed, sometimes uncomfortable questions about safety, power, and who should hold the moral authority over a technology moving this fast. It is the single best hour for understanding both his charisma and the unease he provokes even among admirers.

He is the field’s great narrator of a soft landing: superintelligence arriving gently, its benefits spread wide. On how, exactly, the spreading happens, he has offered several answers over the years — a cash income, then equity, then a share of computing power — each generous in language, each leaving the ownership of the enterprise itself where it began.

Amodei is the field’s most articulate optimist-with-guardrails — a lead architect of the “scaling” idea who left OpenAI to build Anthropic on the premise that safety and capability have to advance together. In the long conversation he is patient and specific about how the systems work and where they might go. His essay imagines AI compressing decades of biological progress into a handful of years, a “country of geniuses in a data center” turned toward curing disease.

Notably, he has gone further than most builders toward the distributive question: he has written that taxes on AI companies could fund a universal basic income. That is a real acknowledgment that the gains need a channel back to the public.

A chess prodigy turned neuroscientist turned AI builder, Hassabis won a share of the 2024 Nobel Prize in Chemistry for AlphaFold, which cracked a fifty-year grand challenge by predicting the three-dimensional shape of proteins — and which DeepMind released freely for the world to use. On 60 Minutes he is measured and grand at once: he forecasts that AI could help cure disease and usher in what he calls “radical abundance,” and says the moment will need new philosophers to make sense of it.

AlphaFold matters here beyond the science. Given away rather than enclosed, it is a working instance of AI enriching the shared store of human knowledge — the gift economy the Field Guide’s Marcel Mauss described, running at planetary scale.

Less a builder than a broker — of capital, and of the state’s posture toward AI. At TED he argues, against the grain, that the revolution is underhyped: that agentic systems able to plan and act on their own are arriving faster than the public grasps, bringing enormous opportunity alongside sober risks in energy, security, and geopolitics. It is a crisp, confident tour of where the frontier is heading from someone who has spent a decade moving between the boardroom and the halls of government.

Schmidt is where private wealth and public power meet. His influence runs through investment, advisory councils, and the defense-technology push — which is precisely why he belongs in a guide otherwise full of builders.

Half a century ago Hinton bet, nearly alone, that machines could learn the way brains do — and the bet paid out in the neural networks underneath everything in this theater. Then, in 2023, the field’s founding figure walked out of Google so he could say plainly what he had come to fear: that we do not fully understand what these systems are doing, that they may already learn in some ways better than we do, and that nobody has a guaranteed method for keeping a smarter thing under control. On 60 Minutes he lays it out for a general audience with the calm of a man describing weather he can see coming.

Less quoted, and just as important to this room: Hinton’s economic warning. He has said repeatedly that the wealth AI generates will flow to those who already have it, widening the gap between rich and poor — and he has endorsed a universal basic income as a floor beneath the displaced.

In December 2020, Google fired the co-lead of its own ethics team over a paper asking whether language models could be too big — too costly to the climate, too saturated with the biases of their training data, too concentrated in too few hands. The firing proved the paper’s point better than any citation could, and it made Gebru the most consequential dissident in the field. A year later she founded DAIR, a research institute deliberately built outside Big Tech’s walls, funded and governed so that no company can end an inconvenient finding by ending its author.

In the keynote, she goes after the field’s grandest story — the promised utopia of artificial general intelligence — and traces its intellectual ancestry to ideologies that ranked human beings and called the ranking science. Her steady question all the way through: who benefits now, while the promised future does the talking?

The founding image of the accountability movement is Buolamwini at her MIT desk in a white theatrical mask — because the facial-analysis software in front of her could find the mask, and could not find her face. The talk tells that story in nine minutes and names the phenomenon: the coded gaze, bias baked in by whoever happens to be in the room when the machine is taught what a person looks like. Her research that followed — the Gender Shades audit, conducted with Timnit Gebru — measured the gaze precisely: commercial systems that read pale male faces nearly perfectly and failed on dark female faces as much as a third of the time.

What makes her singular is the register: she calls herself a poet of code, and means it. Her spoken-word audit “AI, Ain’t I A Woman?” runs the faces of Sojourner Truth’s heirs — Oprah, Michelle Obama, Serena Williams — past the machines and lets the mislabelings speak for themselves.

Raji is the youngest player on this bill and the one most likely to shape the rules the rest will live under. As a student she worked with Buolamwini on the follow-up to Gender Shades, testing whether the companies named in the audit had fixed anything — and showing that, once watched, they largely had. That finding turned the audit from a protest into an instrument: sunlight, measurably, works. She has spent the years since building auditing into a discipline, from the inside of Google’s ethical AI team to the policy rooms where the instrument might become law.

Her Stanford proposal is the one to sit with: independent auditors — not the companies grading their own homework — with a protected legal right of access to the systems that decide who is hired, housed, policed, and paid.

They call Hinton the godfather; Li is the godmother. ImageNet — the vast labeled picture-set she built when the field thought it a waste of time — is what gave the neural networks their eyes, and she gave it away openly, a dataset run as a commons that ignited the deep-learning decade. In the TED talk she opens 540 million years ago, in an ocean with no eyes in it, and argues that machines are now approaching their own such moment: spatial intelligence, the leap from seeing to doing. She is building toward it herself at World Labs, the company she co-founded.

She belongs on this bill for the other thing she built: at Stanford’s Institute for Human-Centered AI she has spent years insisting, in testimony and in teaching, that the technology bend toward human dignity — and the field’s watchers are, in a real sense, her students. Timnit Gebru wrote her doctorate under Li’s advising. The godmother trained the machines to see, and trained the people who check what the machines see.