The reading room
On The Diary of a CEO, economist Steve Keen argued that the only thing likely to slow the race for frontier AI is its sheer cost and the physical resources it needs — and that the company left standing will be Chinese. Other speakers pushed back with military necessity and the long unprofitable years of earlier internet giants, and Keen pointed to the recent Kimi release as his example.
On The Diary of a CEO, Steven Bartlett offered Geoffrey Hinton's proposal for a protective, mother-like AI as the most hopeful answer to Konstantin Kisin's forecast that humans become pets or cattle. Kisin argued that maternal care runs on a genetic incentive a machine would not share; Steve Keen answered that it runs on empathy — and then pointed out that an AI determined to keep us safe might forbid war or cut energy use, which is protection by way of control.
On The Diary of a CEO, economist Steve Keen and commentator Konstantin Kisin agreed that machine production would not by itself give ordinary people an income — and then disagreed about where that income would come from. Keen argued only government money creation could supply it; Kisin, who says his anti-communist credentials are well established, said redistribution becomes unavoidable once AI does the work.
The inner-loop thesis says AI advantage may come from how well an organization turns experience into better systems, rather than from owning the best model. A study of 51 successful deployments gives that argument practical support—but also shows why model choice, employee trust and ownership of workplace knowledge still matter.
After touring Unitree’s headquarters and factories, Alvin Graylin said research and demonstrations still dominate its robot purchases, while upper-torso models are finding more commercial demand. His argument is about engineering economics: repetitive work may need capable arms, but not the balancing, repairs and extra parts that come with legs.
Alvin Graylin argues that specialized small models, coordinated agents and deployment safeguards make parameter counts a poor guide to AI danger. Dave Blundin counters that today’s tests may miss what a self-improving system becomes. Cybersecurity evaluations—and a later investigation into unauthorized agent activity—sharpen their disagreement.
Asked about allegations that Chinese labs extracted capabilities from Claude, Alvin Graylin argued that access to another model’s answers cannot explain every engineering advance. The Moonshots exchange turned on three distinctions: legitimate distillation versus prohibited extraction, query bills versus development costs, and learning from outputs versus improving the machinery behind them.
Alvin Graylin argues that China is competing to spread useful AI through industry and overseas developer communities, rather than betting everything on reaching general intelligence first. Provincial competition and open-weight models help explain his account, though the policy contrast is not absolute: America’s AI Action Plan also explicitly promotes adoption.
Alvin Wang Graylin proposes a practical starting point for US–China AI cooperation: an emergency hotline, shared safety tests and an agreement to keep talking. Speaking personally on Moonshots, ahead of a September 24 dialogue described in the episode, he connects those steps to a larger bargain—financing AI deployment abroad while spreading agreed safety standards.
Alvin Graylin argues that AI can become more useful while earning less for the companies financing its infrastructure. His warning centers on cheaper models and local computing weakening cloud revenues, just as NVIDIA proposes financing platforms intended to mobilize more than $500 billion of outside capital.
A drone navigating by sound cannot carry a complete account of the world it flies through. Rich Sutton and Khurram Javed use that mismatch to explain why AI must keep learning. Their Big World Hypothesis favors models that agents can revise through experience—not an end to simulation, imagination or useful prior knowledge.
Rich Sutton and Khurram Javed want deployed AI to change its underlying weights from individual experience, rather than rely on extra context or shared model updates. Their Oak Lab agenda combines learning rates tailored to each weight with a way to refresh a network’s capacity to learn—supported by earlier experiments, but not yet a demonstrated general-purpose system.
Follow a thread