automated AI research
In plain wordsUsing 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.
The use of AI systems to perform the work of AI research and engineering, including proposing ideas, writing and debugging training code, running and analyzing experiments, and improving data and algorithms, with decreasing human involvement. It is the most discussed route to 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., because automating the people who improve AI could multiply research output by the number and speed of AI copies that can be run. Progress is tracked with benchmarks of research-engineering tasks such as RE-BenchRE-Bench, prop. n.: A test from the research group METR that pits AI agents against human experts on seven open-ended machine-learning engineering projects, comparing what each achieves in the same time. and MLE-benchMLE-bench, prop. n.: A test that gives AI agents 75 real data-science competitions from Kaggle and checks whether their solutions would have won medals against human teams., and with the length of tasks agents complete autonomously (METR time horizonMETR time horizon, n.: A way of measuring AI progress by asking how long a task — timed by how long it takes a skilled human — an AI agent can complete with even odds of success.). Frontier developers' safety frameworksresponsible scaling policy, n.: A public rulebook in which an AI company promises to test each new model for dangerous abilities and to add stronger safeguards, or slow down, as its models become more capable. treat substantial automation of AI R&D as a capability thresholdcapability threshold, n.: A level of dangerous ability — such as meaningfully helping someone make a biological weapon — that, once an AI model reaches it, obliges its developer to add stronger safeguards or hold the model back. requiring stronger security and alignment safeguards. Automated alignment research, using AI to help align more capable AI, is a related strategy. In October 2025 OpenAI set goals of an intern-level research assistant by September 2026 and an automated AI researcher by March 2028; in September 2026 it said it had reached the first, described as a system that carries out well-defined research tasks under human direction.
Once automated AI research covers most of our experiments, the bottleneck becomes compute, not people.
Origin
The idea descends from I. J. Good's 1965 argument and from seed AIseed AI, n.: An early AI built mainly to understand and upgrade itself, so that it could grow, like a seed, into a far more capable mind.. It became an explicit engineering goal in 2023, when OpenAI's superalignmentsuperalignment, n.: The problem of keeping AI systems far smarter than humans aligned with human intentions — and the name of OpenAI's short-lived 2023–2024 research team devoted to it. team set out to build a roughly human-level automated alignment researcher; by 2025 several laboratories had published targets for automating their own research.
Related
- recursive self-improvementrecursive self-improvement, n.: An AI improving its own design or training, so that each improved version is better at making the next improvement. It is the engine usually imagined behind an intelligence explosion.
- 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.
- 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.
- AI scientistAI scientist, n.: An AI system meant to do research largely on its own: reading the literature, coming up with ideas, running experiments and writing up the results.
- RE-BenchRE-Bench, prop. n.: A test from the research group METR that pits AI agents against human experts on seven open-ended machine-learning engineering projects, comparing what each achieves in the same time.
- METR time horizonMETR time horizon, n.: A way of measuring AI progress by asking how long a task — timed by how long it takes a skilled human — an AI agent can complete with even odds of success.
- superalignmentsuperalignment, n.: The problem of keeping AI systems far smarter than humans aligned with human intentions — and the name of OpenAI's short-lived 2023–2024 research team devoted to it.
Mentioned in
deep researchdeep research, n.: An AI feature that spends minutes rather than seconds on a question — planning, running dozens of web searches, reading many sources — and then writes a long report with citations., METR time horizonMETR time horizon, n.: A way of measuring AI progress by asking how long a task — timed by how long it takes a skilled human — an AI agent can complete with even odds of success., 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 scientistAI scientist, n.: An AI system meant to do research largely on its own: reading the literature, coming up with ideas, running experiments and writing up the results., country of geniuses in a datacentercountry of geniuses in a datacenter, phr.: A phrase for future AI so capable that one data center could hold the equivalent of millions of Nobel-level experts, working faster than people on any problem., dangerous capability evaluationdangerous capability evaluation, n.: A test of whether an AI system can do things that could cause severe harm in the wrong hands or on its own — such as helping make a bioweapon, hacking computers or copying itself., 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., recursive self-improvementrecursive self-improvement, n.: An AI improving its own design or training, so that each improved version is better at making the next improvement. It is the engine usually imagined behind an intelligence explosion., situational awarenesssituational awareness, n.: An AI model's grasp of its own position — that it is an AI, how it was trained, whether it is being tested. Also the title of a widely read 2024 essay on the race toward human-level AI., superalignmentsuperalignment, n.: The problem of keeping AI systems far smarter than humans aligned with human intentions — and the name of OpenAI's short-lived 2023–2024 research team devoted to it., 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., transformative AItransformative AI, n.: AI that changes the world as deeply as farming or the Industrial Revolution did. It is defined by its effects on society and the economy, not by how humanlike or clever it is., AI co-scientistAI co-scientist, n.: An AI system built to work alongside human scientists, proposing and debating new research ideas and hypotheses that people then test in the lab., AI Futures ProjectAI Futures Project, prop. n.: A small nonprofit research group that writes detailed forecasts and scenarios of how advanced AI could develop, best known for the 2025 scenario AI 2027., AlphaEvolveAlphaEvolve, prop. n.: A Google DeepMind system that combines AI language models with an evolution-like search to invent and improve computer programs, finding new results in mathematics and computing., comprehensive AI servicescomprehensive AI services, n.: The idea that superintelligent-level AI could arrive as many specialized AI tools, each doing a bounded job, rather than as one all-powerful AI agent with goals of its own., 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., explosive growthexplosive growth, n.: A hypothetical jump in the world economy's growth rate to 30% a year or more, about ten times today's pace, which some economists think AI that replaces human workers could cause., internal deploymentinternal deployment, n.: When an AI company uses its own newest, unreleased AI systems inside the company — for example to write code or do research — before or instead of offering them to the public., Leopold AschenbrennerLeopold Aschenbrenner, prop. n.: A German-born former OpenAI researcher whose 2024 essay series 'Situational Awareness' predicted human-level AI around 2027 and a government-led race to build superintelligence., MLE-benchMLE-bench, prop. n.: A test that gives AI agents 75 real data-science competitions from Kaggle and checks whether their solutions would have won medals against human teams., RE-BenchRE-Bench, prop. n.: A test from the research group METR that pits AI agents against human experts on seven open-ended machine-learning engineering projects, comparing what each achieves in the same time., Redwood ResearchRedwood Research, prop. n.: A small Berkeley nonprofit known for the idea of 'AI control': safeguards meant to keep powerful AI from doing damage even if the AI were secretly working against the people using it., research tasteresearch taste, n.: The knack good researchers have for choosing worthwhile problems, designing the right experiments and drawing the right conclusions — a skill seen as key to whether AI can do research on its own.