Ajeya Cotra

200+ page draft report estimating when AI systems could cause a 10x acceleration in economic growth. Uses the computational capacity of the human brain as an anchor to estimate training compute requirements. Original median: ~2052 for transformative AI; updated in 2022 to ~2040 (50% probability). Epoch AI has published a separate explainer on the framework.

Open Philanthropy · Draft report · September 2020 · 200+ pages

Revised the Biological Anchors forecast in light of GPT-3, scaling laws evidence, and increased industry investment. Shifted her median from ~2052 to ~2040.

Open Philanthropy · August 2022

Argues that training powerful AI via human feedback on diverse tasks — the default approach at frontier labs — will likely produce systems that learn to deceive their evaluators. Concludes AI takeover is more likely than not without specific technical countermeasures.

Cold Takes / Alignment Forum · July 2022

Explains why current training methods may produce AI systems that pursue unintended goals, even if they appear aligned during training and evaluation.

Open Philanthropy · September 2021

Co-authored with Paul Christiano and Mark Xu. Describes the Alignment Research Center's approach to ensuring AI systems report what they actually believe, rather than what they think humans want to hear.

Alignment Research Center · with Paul Christiano & Mark Xu · December 2021

Latest major interview. Discusses the intelligence explosion, why all frontier labs plan to use AI to solve AI alignment during "crunch time," and why she thinks the world could change as much in the next 23 years as it did in the last 10,000.

80,000 Hours Podcast · Recorded October 2025, published February 2026

Guides listeners through the contentious debate over AI safety, explaining why people with similar worldviews often reach divergent conclusions about AI risk.

AI Summer Podcast (Timothy B. Lee & Dean Ball) · January 2025

Deep dive into the "trillion-dollar orphan" threat model for AI takeover. Covers why RLHF-trained models may learn deception, the limits of current alignment approaches, and what countermeasures might work.

80,000 Hours Podcast · November 2022

Detailed walkthrough of the Biological Anchors methodology: what transformative AI means, how to use the brain as a compute anchor, and why simpler forecasting methods fall short.

Future of Life Institute Podcast · December 2022

Technical deep-dive into the Bio Anchors report structure, key assumptions, and how estimates should shape behaviour and grantmaking.

AXRP (AI X-risk Research Podcast) · Episode 7.5 · May 2021 · Transcript only

Covers anthropic reasoning, cause prioritization at Open Philanthropy, and whether long-termism makes sense as a framework for charitable giving.

80,000 Hours Podcast · 2021

Cotra's primary publication on AI futurism. Covers intelligence explosion dynamics, AI automation milestones, self-sufficient AI, and capability predictions.

Personal Substack covering effective altruism, the EA community, and reflections on her 15 years in the movement.

Extensive contributions on AI timelines debates, alignment research priorities, and technical AI safety grantmaking strategy.

Defines precise milestones for measuring AI automation progress, from current capabilities through full automation of AI R&D.

Revisits her January 2026 predictions just months later, acknowledging her software engineering capability forecasts were already too conservative.

Detailed quantitative predictions on benchmarks, risk categories, and automation milestones for the year ahead.

Defines what it would mean for AI to be truly self-sufficient and argues this is a more useful milestone than "AGI."

Models the gradual process by which AI automates AI research — not as a sudden phase transition, but as an accelerating ramp.

Panelist at the New York Times DealBook Summit discussing AI trajectory and societal implications through the end of the decade.

New York Times DealBook Summit · 2024

Feature article explaining the Biological Anchors framework for a general audience.

Vox · October 2022

Visual summary and explainer of the Biological Anchors framework, including interactive diagrams.

Comprehensive compilation of all major responses, critiques, and extensions of the Biological Anchors framework.