METR says evidence cannot rule out drastic AI R&D acceleration within a couple of years
Evaluation nonprofit METR says current public evidence is not enough to rule out "drastic AI R&D acceleration in the next couple of years," in a capability update published on 20 November covering public models up to Claude Opus 5.5 and GPT-6 Astra. It defines that as capabilities improving so fast that humans can no longer meaningfully understand, assess and react to catastrophic risks from new systems.
Among its findings, frontier models usually match or outperform skilled humans on METR's coding and optimisation tasks; when it recruited its own human baseliners, it typically failed to find people who beat the models. Where many domain specialists had attempted a task, models placed at a very high percentile but not at the top.
Models do worse on tasks that are costlier to turn into reinforcement learning environments, but appear to be improving at similar rates on cheap and costly tasks. On tasks that require choices without clear feedback, AI agents appear to have worse judgment than human experts, and model judges overrated outputs from similarly strong or stronger models.
On METR's combined task suites, the length of tasks models can complete half the time is doubling roughly every 3.6 months, similar to a trend OpenAI recently reported on internal research tasks. Its composite capability index shows no strong acceleration or slowdown across the public frontier, though some developer-specific trend changes appear. METR stresses it remains "significantly uncertain" about how fast relevant capabilities are improving.
Experts who spoke to MIT Technology Review called the added open-ended, harder-to-verify tasks a step in the right direction, but noted that many completion times in the new suite are estimates rather than measured by people doing the tasks, and that it is still far from clear whether, or how quickly, automated AI research might arrive.
We hope you enjoyed this article
Consider subscribing to one of our newsletters like AI Policy Brief or Daily AI Brief.
Also, consider following us on social media:
More from AI Safety
Nov 20 OpenAI says new safeguards cut reasoning-extraction attacks on GPT-6.2 Astra to 0.2% Nov 20 Australia and China agree to hold talks on AI after Albanese meets Li Qiang Nov 20 Trump says he and Xi agree AI needs safety rules Nov 20 Dragos seeded two million AI prompts with secrets and found breaches at three providers Nov 20 Protesters rally against Anthropic in New York and San Francisco on its stock market debutAI Policy Brief
Weekly report on AI regulations, safety standards, government policies, and compliance requirements worldwide.
Industry analysis
2025 Global Business Services Agenda: Gen AI Takes Center Stage
This industry analysis by The Hackett Group explores the transformative impact of generative artificial intelligence (Gen AI) on global business services (GBS) in 2025. The study highlights the shift from exploration to acceleration of Gen AI initiatives, with 89% of executives advancing these projects to improve customer satisfaction, innovate products, and reduce costs. The report also discusses the challenges and strategies for successful Gen AI adoption, emphasizing the need for a technology-enabled operating model and the importance of reskilling the workforce.
Read moreYou may also like
OpenAI says its researchers now use $2,400 a day of AI agents each
METR challenges Anthropic's claims about AI safeguards
Epoch AI finds no model meets its employee standard, but Fable 5.2 matches contractors on data center research
New coding benchmark puts Claude Opus 5.5 on top and opens a 35-point gap that DeepSWE hid
OpenAI details how it contains frontier models, including a pause after a DNS exploit
Daily AI Brief: the AI news that matters, in your inbox.