In an Oxford study, LLMs correctly identified medical conditions 94.9% of the time when given test scenarios directly, vs. 34.5% when prompted by human subjects (Nick Mokey/VentureBeat)

https://www.techmeme.com/feed.xml Hits: 14

Summary

— Also: another paper that seals the deal — The Apple paper on limitations in the “reasoning” of Large Reasoning Models, which raised challenges for the latest scaling hypothesis, has clearly touched a nerve.

First seen: 2025-06-14 22:00

Last seen: 2025-06-15 11:01

Read Full Article More from this Source

In an Oxford study, LLMs correctly identified medical conditions 94.9% of the time when given test scenarios directly, vs. 34.5% when prompted by human subjects (Nick Mokey/VentureBeat)

Summary

Related News

Anthropic details how it built its multi-agent Claude Research system, claiming significant improvements in internal evaluations over single-agent systems (Anthropic)

New York State passes a bill mandating safety and transparency requirements for frontier AI models; it awaits Governor Hochul's signature (Maxwell Zeff/TechCrunch)

Statista: the global influencer marketing industry is projected to grow 36% between 2024 and 2025, reaching $33B, as brands tighten overall ad budgets (Bloomberg)

Filings: Bengaluru-based fintech Cred, which offers rewards for paying credit card bills and more, raised ~$72M at a $3.5B valuation, down from $6.4B in 2022 (The Economic Times)

Amazon plans to invest ~$13B in Australia from 2025 to 2029 to develop its data center infrastructure, as demand for cloud computing and AI grows in the country (Ainslie Chandler/Bloomberg)