Explorez tous les épisodes du podcast ForeCast

Peter Salib is an Assistant Professor of Law at the University of Houston Law Center and co-director of the Center for Law & AI Risk. Simon Goldstein is an associate professor at the University of Hong Kong, a senior editor at AI Frontiers, and a visiting senior scholar at Forethought. They have written together about AI safety, AI rights, and the governance of advanced AI.

They joined Forethought’s Fin Moorhouse to discuss their new paper, Liberalism Forever. The conversation covers:

• Their “null hypothesis”: today’s liberal institutions — free markets and democratic governance — may already hold most of the answers for governing the far future, and longtermists have underrated their power
• Examining whether canonical social-science arguments for markets and democracy still hold under transformative AI, space colonization, and explosive growth, and claiming that most survive (and several get stronger)
• Why they favour reasoning “at the margin” over designing a detailed end-state, and why proposals like viatopia and the long reflection sound liberal but risk illiberalism when actually implemented
• Whether the standard case for markets — information aggregation, allocative efficiency, innovation — still bites once AGI can read off preferences directly
• Why inequality is, in their view, a comparatively boring problem with a known solution (tax-and-transfer), why AI-specific taxes like a compute tax are misguided, and how the growth-versus-distribution trade-off changes when growth rates get very high
• Dyson swarms and space resources: why they doubt there’s any real monopoly worry, since energy would be additive with free entry, and why property auctions might beat egalitarian allocation schemes
• The arguments for democracy that persist post-AGI — public-choice/selectorate incentives and democracy as a commitment device — versus the epistemic arguments, which they think weaken
• Why AGI could make autocracy more competitive (loyal AI bureaucrats and militaries removing the need for a human winning coalition), and what preserving democratic control requires in response
• Handoffs to superintelligent AI and successionism, and their preferred alternative of extending the franchise to AIs with their own values rather than deferring wholesale
• The history of new “technologies of democracy,” from the Federalist Papers to LLMs as impartial arbiters, and where each author thinks their own argument is weakest

You can read a full transcript here.

To see all Forethought's published research, visit forethought.org/research.

To subscribe to our newsletter, visit forethought.org/subscribe.

Will MacAskill is a senior research fellow at Forethought and the author of What We Owe The Future. Tom Davidson is a senior research fellow at Forethought and the author of a series of reports on AI timelines, takeoff speeds, and AI-enabled coups.

In this informal conversation about early-stage research in progress, they discuss:

• Why a philanthropist who takes transformative AI seriously should expect extraordinary investment returns (plausibly 10x to 100x) and how the market doesn’t seem to be pricing this in
• Why AGI shouldn’t be treated as a hard deadline to spend by, and the case that most philanthropic money should be spent during the intelligence explosion rather than before it
• How genuine AI substitutes for human labor would dissolve the hiring bottleneck that makes organizations so hard to scale today
• How relative prices shift when cognitive labour becomes abundant, why the patent system stops making sense in that world, and the window of “crazy philanthropic bargains” open to actors who adapt faster than large institutions
• The prospect of a world where only a small elite has access to superintelligent medical, financial, and strategic advice, and what it would take to avoid this
• Counter-considerations to the main case, e.g., crowding out by new money, the risk that overspending attracts grifters or distorts the field, and whether a richer world will mean worse philanthropic opportunities
• The risk of value drift over a long saving period, and mechanisms for binding your future self (such as irrevocably handing off control of philanthropic funds to independent, constitutionally-bound foundations)
• “Differential intellectual development”: using directable AI researchers to accelerate biodefense, infosecurity, AI interpretability, and neglected armchair fields like population ethics and social choice theory
• Why the case for waiting to give is much weaker for small donors
• Will’s worry that the whole line of reasoning might fail if it turns out scaling philanthropy well is bottlenecked by serial real-world learning that no amount of cognitive labor can compress

You can read a full transcript here.

To see all Forethought's published research, visit forethought.org/research.

To subscribe to our newsletter, visit forethought.org/subscribe.

Avi Parrack is a physics PhD student at Stanford and a researcher at Forethought, where he works on AI, space expansion, and governance. He is the lead author of a recent Forethought report on orbital data centers.

He joined Forethought’s Tom Davidson to discuss:

• What an orbital data center actually is, and why they’re suddenly being taken seriously
• Why the core case rests on cheaper energy from more intense and near-constant solar power
• Why orbital data centers hinge on SpaceX’s Starship bringing launch costs toward $100/kg or below
• Why cooling may actually not be a major issue
• Whether the inability to make repairs dooms orbital data centers
• Vulnerability of space data centers to kinetic attacks, and whether they would cause ‘Kessler syndrome’ debris cascades
• Why model-weight security might actually improve in orbit even as physical vulnerability rises
• If space becomes the cheapest place to scale compute and only SpaceX has the launch capacity, what the means for concentration of power, US–China competition, and the possibility of pausing AI
• Overall cost comparisons with terrestrial data centers
• The longer-run picture: when and how does the post-AGI industrial explosion spread to space?

You can read a full transcript here.

To see all Forethought's published research, visit forethought.org/research.

To subscribe to our newsletter, visit forethought.org/subscribe.

Damon Binder is a senior researcher on the Biosecurity and Pandemic Preparedness team at Coefficient Giving. He has a PhD in physics from Princeton and previously studied existential risks at Oxford’s Future of Humanity Institute.

He joined Fin Moorhouse to discuss his series on the AI industrial explosion. Topics include:

• How input-output tables let you estimate how fast a self-replicating economy could grow if labor were free, yielding a headline result of roughly annual doubling, even under conservative assumptions and heavy regulation
• Why competition makes low-growth restraint unstable
• Why raw-material scarcity isn’t a hard barrier
• Why a fast-growing economy wants cheap, disposable, infrastructure
• Thermodynamic and other physical speed limits to physical growth
• Which biological organisms replicate fastest, and what we can learn from them
• Why physical output matters for hard power
• How Damon uses AI in his research

You can read a full transcript here.

To see all Forethought's published research, visit forethought.org/research.

To subscribe to our newsletter, visit forethought.org/subscribe.

This is an AI narration of "Risk-Averse AIs" by Elliott Thornley and William MacAskill. The article was first released on 24th June 2026.

You can read more of our research at forethought.org/research. Thank you to Type III audio for providing these automated narrations.

Wei Dai is a computer engineer known for his work in cryptography and cryptocurrency systems, and for his long-standing contributions to AI safety, decision theory, and metaphilosophy.

He joined Forethought’s Fin Moorhouse to discuss:

• Do we need to solve ‘metaphilosophy’ before we can trust AIs to answer crucial questions about the long-run future?
• Is it inevitable that the wisdom of the frontier AIs (their philosophical and strategic competence) will lag dangerously behind their raw capabilities in coding, math, and science?
• How do status games distort morally important decisions and conversations, including about the future of AI?
• How worried should we be about AI superpersuasion?
• The concept of “illegible problems”: crucially important issues that aren’t on almost anyone’s radar
• Is philosophical convergence necessary for a good future, or is institutional design enough?
• Wei Dai’s personal intellectual and career journey

To respect Wei’s privacy, this episode’s audio is an AI narration of a transcript of a real conversation, which was edited for clarity.

You can read a full transcript here.

To see all Forethought's published research, visit forethought.org/research.

To subscribe to our newsletter, visit forethought.org/subscribe.

This is an AI narration of "What should go in a model spec?" by James Tillman. The article was first released on 4th June 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 "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.

This is an AI narration of "A draft honesty policy for credible communication with AI systems" by Lukas Finnveden, Mia Taylor, and Max Dalton. The article was first released on 6th 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 "The Saturation View" by William MacAskill. The article was first released on 24th April 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 "Desired behaviour for AI advising humans on important decisions" by Tom Davidson. The article was first released on 17th April 2026.

You can read more of our research at forethought.org/research. Thank you to Type III audio for providing these automated narrations.

Owen Cotton-Barratt is a research consultant working with the Future of Life Institute, and a co-author of several recent Forethought articles on AI tools for epistemics and coordination. Rose Hadshar is a researcher at Forethought. They discuss:

• Whether LLMs are now good enough to start building tools that meaningfully help people track what's true
• What AI-powered reliability tracking could look like
• How AI negotiation systems could work
• Structured transparency and automated arms inspection — verifying compliance without revealing confidential information
• Whether coordination tech is more likely to enable cooperation or collusion
• The vision of a "Sensible Revolution": moving from individual tools to background infrastructure that makes civilisation's decision-making less insane
• Why building the good versions of these tools early matters for path dependency

You can read a full transcript here.

To see all Forethought's published research, visit forethought.org/research.

To subscribe to our newsletter, visit forethought.org/subscribe.

This is an AI narration of "AI impacts on epistemics: the good, the bad and the ugly" by Owen Cotton-Barratt and Rose Hadshar. The article was first released on 12th April 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 "Design sketches: defense-favoured coordination tech" by Owen Cotton-Barratt, Oly Sourbut, Lizka Vaintrob, and Rose Hadshar. The article was first released on 5th April 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 "AI for AI for Epistemics" by Owen Cotton-Barratt and Lukas Finnveden. The article was first released on 1st April 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 "AI should (sometimes) be proactively prosocial" by Tom Davidson and William MacAskill. The article was first released on 29th March 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 "Is labor a luxury in the long run?" by Philip Trammel. The article was first released on 27th March 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 "Concrete Projects in AGI Preparedness" by William MacAskill and Fin Moorhouse.. The article was first released on 26th March 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 "The importance of AI character" by William MacAskill and Tom Davidson. The article was first released on 23rd March 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 "Broad Timelines" by Toby Ord. The article was first released on 16th March 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 "Should We Lock in Post-AGI Agreements Under Uncertainty?" by Fin Moorhouse. The article was first released on 13th March 2026.

You can read more of our research at forethought.org/research. Thank you to Type III audio for providing these automated narrations.

Phil Trammell is an economics postdoc at Stanford University's Digital Economy Lab, working on questions related to economic growth and AI. He discusses:

• Why Piketty's thesis about runaway inequality was likely wrong about the past but right about the future
• How full automation turns capital and labour into gross substitutes
• Why catch-up growth between rich and poor countries could end
• How the privatisation of returns is already concentrating wealth
• Why family dynasties and inheritance become far more important in a post-automation economy
• Whether autocratic regimes can outgrow democracies after AGI
• How to measure whether capital is becoming truly self-replicating — and what the data currently shows

Watch the video version on YouTube.

Read a full transcript here.

To see all Forethought’s published research, visit forethought.org/research. To subscribe to our newsletter, visit forethought.org/subscribe.

This is an AI narration of "Moral public goods are a big deal for whether we get a good future" by Tom Davidson, William MacAskill, Mia Taylor. The article was first released on 23rd February 2026.

You can read more of our research at forethought.org/research. Thank you to Type III audio for providing these automated narrations.

Sam Hammond is is Chief Economist at the Foundation for American Innovation. He discusses:

• How collapsing transaction costs could push towards privatising the nation-state
• The “distributed denial of service” problem for courts and regulators
• Why perfect enforcement of existing laws would be effectively totalitarian
• Estonia's government-as-API as a model for AI-era governance
• Whether 20th-century social democracy was a technological aberration
• Why post-AGI states might look more like Dubai than Denmark
• Mormons, religion, and social scaffolding for a post-scarcity world

Watch the video version on YouTube.

Read a full transcript here.

To see all Forethought’s published research, visit forethought.org/research. To subscribe to our newsletter, visit forethought.org/subscribe.

This is an AI narration of "Design sketches: tools for strategic awareness" by Owen Cotton-Barratt, Lizka Vaintrob, Oly Sourbut, and Rose Hadshar. The article was first released on 10th February 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 "Design sketches for a more sensible world" by Owen Cotton-Barratt, Lizka Vaintrob, Oly Sourbut, and Rose Hadshar. The article was first released on 5th February 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 "Design Sketches: Angels-on-the-Shoulder" by Owen Cotton-Barratt, Lizka Vaintrob, Oly Sourbut, and Rose Hadshar. The article was first released on 29th January 2026.

You can read more of our research at forethought.org/research. Thank you to Type III audio for providing these automated narrations.

Rose Hadshar is a researcher at Forethought. She discusses:

• Why governments might pursue an international AGI government
• The chance that the first AGI developer becomes a de facto world government
• The "Intelsat model" – treating AGI as commercial infrastructure rather than a weapon or science project
• Membership and voting structures that might make a US-led coalition viable
• The "AGI convention" proposal

Watch the video version on YouTube.

Read a full transcript here.

To see all Forethought’s published research, visit forethought.org/research. To subscribe to our newsletter, visit forethought.org/subscribe.

This is an AI narration of "AGI and World Government" by William MacAskill and Rose Hadshar. The article was first released on 26th January 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 "The International AGI Project Series" by William MacAskill. The article was first released on 26th January 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 "What an international project to develop AGI should look like" by William MacAskill. The article was first released on 26th January 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 "A global convention to govern the intelligence explosion" by William MacAskill. The article was first released on 26th January 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 "An overview of some international organisations, with their voting structures" by Rose Hadshar. The article was first released on 26th January 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 "International AI projects and differential AI development" by William MacAskill. The article was first released on 26th January 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 "The UN Charter: a case study in international governance" by Forethought Research. The article was first released on 26th January 2026.

You can read more of our research at forethought.org/research. Thank you to Type III audio for providing these automated narrations.

Tom Davidson and Will MacAskill are both researchers at Forethought.

They discuss:

• What is the industrial explosion?
• Why the case for recursive self-improvement is stronger for physical industry than for software
• How fast the physical economy could grow, the case for weekly doubling times, and limits from natural resources
• Why authoritarian regimes might have a structural advantage in the industrial explosion
• Could a leading country outgrow the entire world to achieve decisive dominance?
• Why does the industrial explosion get ~1% of the attention of the intelligence explosion?

You can read a full transcript here.

To see all Forethought’s published research, visit forethought.org/research. To subscribe to our newsletter, visit forethought.org/subscribe.

This is an AI narration of "Short Timelines Aren't Obviously Higher-Leverage" by William MacAskill and Mia Taylor. The article was first released on 22nd January 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 "Beyond Existential Risk" by William MacAskill and Guive Assadi. The article was first released on 21st January 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 "Is Flourishing Predetermined?" by Fin Moorhouse and Carlo Leonardo Attubato. The article was first released on 21st January 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 "ML research directions for preventing catastrophic data poisoning" by Tom Davidson. The article was first released on 7th January 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 "Viatopia" by William MacAskill. The article was first released on 7th January 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 "Design Sketches: Collective Epistemics" by Owen Cotton-Barratt, Lizka Vaintrob, and Oly Sourbut. The article was first released on 25th December 2025.

You can read more of our research at forethought.org/research. Thank you to Type III audio for providing these automated narrations.

Lukas Finnveden is a Research Analyst at Redwood Research. We talk about making deals with early scheming AIs: paying them to cooperate rather than take over.

You can read a full transcript here.

To see all Forethought’s published research, visit forethought.org/research. To subscribe to our newsletter, visit forethought.org/subscribe.

This is an AI narration of "The first type of transformative AI?" by Owen Cotton-Barratt, Lizka Vaintrob, and Oly Sourbut. The article was first released on 18th December 2025.

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 "Space Debris and Launch Denial" by Fin Moorhouse. The article was first released on 17th December 2025.

You can read more of our research at forethought.org/research. Thank you to Type III audio for providing these automated narrations.

Rose Hadshar (Forethought) and Nora Ammann (ARIA) talk about power concentration, checks and balances, coups, and post-AGI political economy.

You can read a full transcript here.

This is the audio version of a video podcast. Watch the video on YouTube: youtu.be/PB855Xpx1kk?si=v05LVM-2VQ3XiT7N

Prompted in part by Rose's recent article for 80,000 Hours on ‘Extreme power concentration’ — 80000hours.org/problem-profiles/extreme-power-concentration

Timstamps:

• 00:00:00 Intro
• 00:01:26 What is ‘power concentration’? Is power concentration the right framing?
• 00:05:25 When can it be good for power to be centralised?
• 00:09:40 Is ‘checks and balances’ a better framing?
• 00:11:16 How AI undermines existing checks and balances
• 00:15:25 Economic power, meme complexes, and cultural influence
• 00:28:50 AI companies vs governments as centres of power
• 00:31:25 The difficulty of knowing where power actually lies
• 00:40:05 Do humans and AIs concentrate power differently?
• 00:48:38 Should we be trying to imagine a better post-AGI political economy?
• 00:56:28 Concrete actions: transparency & whistleblower protections
• 00:59:07 Human-AI teaming and collaborative intelligence
• 01:01:16 AI agent economies and proving trustworthiness
• 01:13:16 Can we train AI to follow the law?
• 01:16:33 Building resilient coalitions of trustworthy agents
• 01:18:42 Closing reflections

Joe Carlsmith is a writer, researcher, and philosopher. He works at Anthropic on the character/constitution/spec for Claude. Before that, he was a senior advisor at Open Philanthropy.

You can read a full transcript here.

To see all Forethought’s published research, visit forethought.org/research. To subscribe to our newsletter, visit forethought.org/subscribe.

This is a bonus episode to say that Forethought is hiring researchers. After an overview of the roles, we hear from Research Fellow Mia Taylor about working at Forethought.

You can read a full transcript here.

The application deadline has been extended to November 1st 2025.

Apply here: forethought.org/careers/researcher

• (00:00:00) Forethought hiring overview and roles
• (00:03:21) Interview with Mia begins
• (00:03:34) Why Mia joined Forethought
• (00:05:59) Daily work, how work at Forethought was different from expected
• (00:08:38) Research examples
• (00:10:58) Who should and shouldn't apply
• (00:14:11) Disagreements and closing thoughts

• Apply for researcher positions
• Referral form (if you know someone who might be a good fit)

This is a narration of ‘Introducing Better Futures ’ by William MacAskill; published 3rd August 2025.

Narration by Perrin Walker (@perrinjwalker).