AI Futures Project
In plain wordsA small nonprofit research group that writes detailed forecasts and scenarios of how advanced AI could develop, best known for the 2025 scenario AI 2027.
A nonprofit research organization that forecasts the development of advanced AI and its consequences through quantitative models, detailed scenarios and tabletop exercises. It is best known for AI 2027AI 2027, prop. n.: A detailed 2025 forecast, written as a month-by-month story, imagining AI becoming superhuman by the end of 2027, with two alternative endings depending on whether humanity slows down. (April 2025), a month-by-month scenario in which automated AI researchautomated AI research, n.: Using AI systems to do the work of AI researchers, such as running experiments and writing code for new models, so that AI helps build its own successors. yields a superhuman codersuperhuman coder, n.: A milestone used by AI forecasters: an AI that can do the coding work of a company's best programmer, far faster and in many copies at once. and then an intelligence explosionintelligence explosion, n.: The idea that once AI gets good enough at improving AI, each smarter version could build a smarter one faster still, so abilities would snowball far beyond human level in a short time. by 2027, ending in two alternative branches, a race and a slowdown; the scenario was read widely, including by government officials, and criticized as too fast and too specific. Its accompanying timelines and takeoff forecasts set out its assumptions explicitly, and it has since revised them as evidence has accumulated. Later work includes scenarios with policy recommendations for US decision-makers. The method continues its founder's 2021 forecast essay 'What 2026 Looks Like'.
Origin
Founded in 2024 by Daniel Kokotajlo, who left OpenAI's governance team that year and drew attention for refusing to sign a non-disparagement agreement that put his vested equity at risk; researchers Eli Lifland and Thomas Larsen lead much of its work. It is a US nonprofit, legally incorporated as Artificial Intelligence Forecasting Inc.
Related
- AI 2027AI 2027, prop. n.: A detailed 2025 forecast, written as a month-by-month story, imagining AI becoming superhuman by the end of 2027, with two alternative endings depending on whether humanity slows down.
- AI timelinesAI timelines, n.: Forecasts of when AI will reach major milestones, such as matching humans at most work. People with short timelines expect this within years; others expect decades, or never.
- superhuman codersuperhuman coder, n.: A milestone used by AI forecasters: an AI that can do the coding work of a company's best programmer, far faster and in many copies at once.
- intelligence explosionintelligence explosion, n.: The idea that once AI gets good enough at improving AI, each smarter version could build a smarter one faster still, so abilities would snowball far beyond human level in a short time.
- takeofftakeoff, n.: The stretch of time between AI reaching roughly human level and AI becoming vastly superhuman. People argue over whether it would take decades, years, months or just days.
- automated AI researchautomated AI research, n.: Using AI systems to do the work of AI researchers, such as running experiments and writing code for new models, so that AI helps build its own successors.
Mentioned in
Daniel KokotajloDaniel Kokotajlo, prop. n.: An American AI forecaster and former OpenAI researcher who led 'AI 2027', a detailed scenario of a race to superintelligence, and campaigns for transparency at AI companies.