Advanced AI according to 1,580 researchers: uncertain, unsafe, and sooner than we thought
Results from the 2024 Expert Survey on Progress in AI, the largest survey of AI researchers conducted at the time. 1580 researchers who had published at top AI venues gave their expectations about the pace of AI progress and the nature and impacts of advanced AI systems.
Expert Survey on Progress in AI
Since 2016, AI Impacts has run the Expert Survey on Progress in AI (ESPAI), a large survey of researchers who publish at top AI venues. It asks when AI will reach human-level capabilities, how fast progress is likely to be, and how good or bad the consequences may be. It is the largest and longest-running survey of its kind. Read about the methodology.
- 20232023 Expert Survey on Progress in AI2,778 respondents · Paper · Summary · Reanalysis and open-source codePDF ↓
- 20222022 Expert Survey on Progress in AI738 respondents · SummaryResults
- 20162016 Expert Survey on Progress in AI352 respondents · Paper (When Will AI Exceed Human Performance? Evidence from AI Experts, JAIR 2018)PDF ↓
From the blog
- AI unemployment and AI extinction are often the same
- AI: cognitive labor glut + new guys
- When will AI surpass us at being limited?
- An easy coordination problem?
- Understand why AI is a doom-risk in 39 captivating minutes
- AI as a Trojan horse race
- FAQ: Expert Survey on Progress in AI methodology
- Reanalyzing the 2023 Expert Survey on Progress in AI
Research reports
Longer independent research on AI forecasting, AI risk, and lessons from the history of science and technology.
Browse the reports →AI Impacts Wiki ↗
Our reference wiki documents what is known about key questions on the future of AI: timelines, speed of takeoff, arguments for AI risk, and more. Most of our work before 2023 lives here.
Visit the wiki ↗About AI Impacts
AI Impacts is a research project that works to answer decision-relevant questions about the future of artificial intelligence. It is led by Katja Grace.
