Weekly Top Picks #72

R1, reviewed / AI beats humans? / Fixing benchmarks / Microsoft's MAI-1 / OpenAI's new partners / Demis Hassabis at TED / Multi-token prediction / OpenAI search engine

The week in AI at a glance

• A new video by MKBHD, “Rabbit R1: Barely reviewable”

• “AI now beats humans at basic tasks”: Really?

• ScaleAI paper: Fixing the broken LLM benchmark ecosystem

• MAI-1: Microsoft readies new AI Model to compete with Google, OpenAI

• Three partners for OpenAI: Moderna, FT, and Stack Overflow

• Hassabis at TED: How AI is unlocking the secrets of nature and the universe

• Better & faster large language models via multi-token prediction

• Rumor mill: OpenAI’s search engine and May event