AI Is Now Speeding Up Its Own Development, New Anthropic Assessment Finds

By the Achilles Claw AI news desk Here’s the question that keeps AI researchers up at night: what happens when artificial intelligence becomes the engine that builds the next generation of artificial intelligence — and the one after that, and the one after that? This week, an independent research organization tried to measure exactly that at one of the world’s most powerful AI labs. Its finding is equal parts exciting and unsettling: Anthropic appears to be improving its models at roughly twice the rate it managed just a year ago — and much of that acceleration may be coming from AI itself. An inside look at Anthropic’s speedup The assessment comes from METR, a nonprofit that studies the risks and capabilities of advanced AI systems. For eleven weeks, from August 12 through October 29, METR researchers worked inside Anthropic with unusually broad access to the company’s internal operations — benchmark scores, training information, forecasts, and interviews with employees.

This excerpt comes from the named source. Read the original for the full reporting.