ForeCast

Matthew Adelstein, a moral philosopher and writer on AI, animal welfare, and humanity’s long-run future, explores whether artificial intelligence should inherit society’s most consequential decisions. They discuss AI presidents, digital citizens, reversible handoff, moral philosophy beyond human comprehension, democratic legitimacy, and whether goodness can survive competitive pressures in a post-AGI universe.

Simon Goldstein, an AI-safety researcher and editor, and Peter Salib, a law professor focused on AI governance, debate whether liberal institutions can steer a transformative future. They discuss whether markets and democracy still work under AGI and space colonization. Topics include marginal reforms over utopias, market info-aggregation, inequality via tax-and-transfer, monopoly risks in space, and enfranchising digital minds.

Will MacAskill, senior research fellow at Forethought and author of What We Owe The Future, discusses when and how philanthropists should time giving in light of transformative AI. He explores why waiting for the intelligence explosion could yield massive returns, how AI dissolves hiring bottlenecks and reshapes prices and patents, and the risks of concentrated access, overspending, and value drift.

Avi Parrack, a Stanford physics PhD and Forethought researcher on AI and space governance. He explains what orbital data centers are and why near-constant solar power makes them compelling. He walks through Starship-driven launch cost thresholds, cooling and repair tradeoffs, debris and attack risks, latency and interconnect limits, and the geopolitical stakes of scaling compute in orbit.

Damon Binder, a PhD physicist and senior researcher on biosecurity and pandemic preparedness, outlines his analysis of an AI-driven industrial explosion. He explains input–output methods for growth limits, energy and machinery as core bottlenecks, why cheap disposable infrastructure could win, thermodynamic and biological analogies to replication speed, and the geopolitical instability of restraint.

They discuss the risk of misaligned AIs attempting takeovers and the idea of making AIs risk-averse to deter rebellion. The conversation covers targeting CARA risk aversion, numeric examples of small payments that can buy cooperation, and practical ways to train and test risk aversion. They also explore payment mechanisms, credibility strategies, limits and dangers, and real-world steps for companies.

Wei Dai, a computer engineer known for cryptography, cryptocurrency, and AI safety work, joins to explore whether we must solve metaphilosophy before trusting AIs. They discuss AIs that are brilliant at coding but poor at philosophy, risks from AI persuasion and status-driven distortions, the idea of “illegible problems,” and whether institutional fixes can substitute for philosophical convergence.

James Tillman, researcher and writer on AI policy and model specs. He outlines four ways to judge specs and why predictability, safety prohibitions, and future-proofing matter. He discusses evaluability, transparency tradeoffs, risks of rigid defaults, and how specs can coordinate society. Practicality, meshability, and trainability are highlighted as core concerns.

This is an AI narration of "Will We Really Put Data Centers in Space?" by Avi Parrack and Fin Moorhouse. The article was first released on 22nd May 2026. You can read more of our research at forethought.org/research. Thank you to Type III audio for providing these automated narrations.

This is an AI narration of "Stickiness in AI Behavioral Design" by James Tillman. The article was first released on 13th May 2026. You can read more of our research at forethought.org/research. Thank you to Type III audio for providing these automated narrations.