Radar
Top stories
Importance 2 of 5Covered by 1 source

Hot take: frontier models haven’t actually become much smarter

I know this is a controversial take, but I would like to hear others’ opinions: LLMs have become much more useful, but I’m not convinced their “raw intelligence” has improved nearly as much as their overall performance suggests. Context: I work with LLMs as a SWE, as a hobbyist experimenting with local models, and…