Original Reading Library

Read the original arguments behind better AI futures. Search authors and titles, follow curated preparation prompts and keep your own source-linked notes.

• Should we aim for flourishing over mere survival? — MacAskill, Forethought
• Against Maxipok: existential risk isn't everything — MacAskill & Assadi
• How hard to achieve is eutopia? — MacAskill & Moorhouse
• Will morally motivated actors steer us towards a near-best future? — MacAskill & Moorhouse
• The trajectory of the future could soon get set in stone — MacAskill, Forethought
• Effective altruism in the age of AGI — MacAskill
• The crucible: how I think about the situation with AI — Cotton-Barratt, 10 min
• Forethought: a new AI macrostrategy group — 4 min
• Better Futures (full series) — Forethought
• Is Flourishing Predetermined? — Moorhouse, 1 min
• What We Owe the Future — MacAskill, book
• Preparing for the Intelligence Explosion — MacAskill & Moorhouse, Forethought
• Will AI R&D Automation Cause a Software Intelligence Explosion? — Eth & Davidson, Forethought
• Three Types of Intelligence Explosion — Hadshar, MacAskill & Davidson, Forethought
• The Choice Transition — Cotton-Barratt & Raymond
• Which questions can't we punt? — Vaintrob, Cotton-Barratt & Hadshar, Forethought
• Measuring AI Ability to Complete Long Tasks — METR
• Is there a half-life for the success rates of AI agents? — Toby Ord
• AI 2027 — AI Futures Project
• How quick and big would a software intelligence explosion be? — Davidson & Houlden, 41 min
• Carl Shulman on the Dwarkesh Podcast (two parts) — about 6h audio
• The case for multi-decade AI timelines — Erdil, Epoch AI
• Full automation of AI R&D probably yields a large speed up even without a software-only singularity — Greenblatt, Redwood Research, 2026\
• Rerunning the Time of Perils — MacAskill, 11 min
• The first type of transformative AI? — Vaintrob, 2 min
• Are Short AI Timelines Really Higher-Leverage? — MacAskill & Taylor, 17 min
• AI-enabled coups: a small group could use AI to seize power — Davidson, Finnveden & Hadshar, Forethought
• Gradual disempowerment — 80,000 Hours problem profile, after Kulveit et al.
• My favourite version of an international AGI project — MacAskill, Forethought
• Design international AI projects with DAID in mind — MacAskill, Forethought
• Should we make grand deals about post-AGI outcomes? — Moorhouse, Forethought
• AuditBench: Evaluating Alignment Auditing Techniques on Models with Hidden Behaviors —
• Thoughts on AGI and world government — MacAskill & Hadshar, Forethought
• Intelsat as a Model for International AGI Governance — Hadshar & MacAskill, Forethought
• Gradual Disempowerment (full paper) — Kulveit, Douglas, Ammann, Turan, Krueger & Duvenaud
• Disempowerment spirals as a likely mechanism for existential catastrophe — Raymond & Cotton-Barratt, 6 min
• The AI Adoption Gap: Preparing the US Government for Advanced AI — Vaintrob, 20 min
• Idea: the intelligence explosion convention — MacAskill, 11 min
• Cooperation, Conflict, and Transformative AI: A Research Agenda — Clifton, Center on Long-Term Risk
• Gaming the Future — Web book
• What sort of post-superintelligence society should we aim for? — MacAskill
• Bootstrapping to viatopia — MacAskill
• Moral error as an existential risk — MacAskill
• The value of moral diversity — Taylor, Forethought
• Understanding the moral status of digital minds — 80,000 Hours problem profile
• Taking AI Welfare Seriously — Long, Sebo and co-authors
• Research priorities for AI welfare — Eleos AI (also in the optional list).
• No Easy Eutopia (full essay) — MacAskill & Moorhouse, 58 min
• Convergence and Compromise (full essay) — MacAskill & Moorhouse, 63 min
• Persistent Path-Dependence (full essay) — MacAskill, 31 min
• Propositions Concerning Digital Minds and Society — Bostrom & Shulman
• The Saturation View — MacAskill, Forethought
• Moral public goods are a big deal — Davidson, MacAskill & Taylor, 21 min
• Otherness and control in the age of AGI — Carlsmith, essay series
• AI character is a big deal — MacAskill & Davidson, Forethought
• AI impacts on epistemics: the good, the bad and the ugly — Cotton-Barratt & Hadshar, Forethought
• AI Tools for Existential Security — Vaintrob & Cotton-Barratt, Forethought
• Concrete projects to prepare for superintelligence — MacAskill & Moorhouse, Forethought
• How to make the future better (other than by reducing extinction risk) — MacAskill, Forethought
• Inspect — UK AI Security Institute
• Towards Understanding Sycophancy in Language Models — Sharma et al
• AI should (sometimes) be proactively prosocial — Davidson & MacAskill, 11 min
• Desired behaviour for AI advising humans on important decisions — Davidson, 2 min \+ draft spec
• What should go in a model spec? — Tillman, Forethought
• Stickiness in AI Behavioral Design — Tillman, Forethought
• A draft honesty policy for credible communication with AI systems — Finnveden, Taylor & Dalton, Forethought
• AI for AI for Epistemics — Cotton-Barratt & Finnveden, 11 min
• Some tools for collective epistemics — Cotton-Barratt, Vaintrob & Sourbut, 1 min
• Design sketches: defense-favoured coordination tech — Cotton-Barratt et al., 29 min
• Meaning Alignment Institute — AI-powered democratic representation and full-stack alignment.
• Effective Institutions Project — institutional decision-making, including US-government AI readiness.
• Formation Research — lock-in and secret-loyalties research.
• Modeling Cooperation — AI development competition dynamics.
• Simon Institute for Longterm Governance — advising states negotiating international AI governance.
• Forethought — visiting and research roles open periodically
• post-AGI civilizational equilibria workshop —
• creativecommons.org/chooser — generates the line for you

Interactive tools load with JavaScript. Original sources retain their language and attribution.