Skip to content

Topic

Automated AI Research

AI-driven research in which systems autonomously propose, run and iterate experiments, including evolutionary search, with little or no human selection.

Current stories

buildOne report1 publisher

NVIDIA let an optimizer agent find the four token savers in its SoL-Pi harness for Pi

NVIDIA released SoL-Pi, an MIT-licensed Pi extension with four token-saving harness mechanisms that an optimizer agent found. A dev.to review puts the saving near one third on long sessions, bought with a few lost solves.

Publishers:dev.to

Reality

Evidence45
Adoption
Not scored
Hype gap+20
Incentives
Not scored
Confidence45
leadershipConfirmed3 publishers

Frontier lab scientists ask governments to put auditors inside their own companies

Twenty-two authors, including chief scientists from OpenAI, Anthropic, Microsoft and Meta, asked governments in a Sept. 28 paper to require reporting on how far labs have automated their research and to embed independent auditors inside some firms.

Perspective Coverage

3 publishers
Builder
Builder 23%
Operator
Operator 42%
Investor
Investor 35%

Reality

Evidence60
Adoption15
Hype gap+25
Incentives55
Confidence60
scienceOne report1 publisher

Hinton, Bengio and lab researchers ask governments to prepare for AI that speeds up its own research

Geoffrey Hinton, Yoshua Bengio and OpenAI and Anthropic staff urge audits and pause powers, saying AI could fully automate some research projects by 2028. They say the self-reinforcing loop has not started yet but could compress years of progress into months once it does.

Reality

Evidence40
Adoption
Not scored
Hype gap0
Incentives
Not scored
Confidence40
buildOne report1 publisher

Post-training agents pick their strategy before they run a single experiment

An empirical study of publicly released PostTrainBench trajectories finds the training strategy is fixed at step one and the whole remaining budget goes on local tweaks, with scaffolding and human hints improving execution and leaving that pattern intact.

Publishers:arxiv.org

Reality

Evidence58
Adoption20
Hype gap+12
Incentives35
Confidence55
buildOne report1 publisher

The judge went synthetic first, which tells you which part of your pipeline is next

Latent Space dates five flips from human-made to model-made since 2022. The order they happened in predicts more than the annual cadence the essay claims for it.

Publishers:latent.space

Reality

Evidence42
Adoption58
Hype gap+32
Incentives55
Confidence48