METR report finds Anthropic AI accelerating own development
Original title: Why a new report on Anthropic automating AI development is alarming researchers
Models improve twice as fast via AI assistance with heavy redactions limiting public insight.
METR investigated Anthropic over 11 weeks and found its AI models improving roughly twice as quickly as a year ago. The acceleration stems from AI assisting researchers rather than added compute. Three new unreleased models were trained in the past six months and Anthropic redacted most of the detailed report.
Unconfirmed: Anthropic justified redactions as protecting rivals
- Anthropic models improve twice as fast
- Majority of acceleration from AI
- Employees report 12x productivity
- Three new models trained but unreleased
- Extensive redactions applied
From the original article
9 min read at theverge.com
Anthropic CEO Dario Amodei. | Photo: Jason Henry/Bloomberg via Getty Images Anthropic is using AI to build more powerful AI. And it appears to be working remarkably well, with the company's models improving roughly twice as quickly as a year ago, with little indication of slowing down.
By Robert Hart · theverge.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.