(1/5) With all the ongoing hype around LLMs, an interesting research paper - arxiv.org/pdf/2506.21734 and its associated Github repo - github.com/sapientinc/HRM kind of went a bit under the radar. It is still early to tell if this is significant, as no reports of independent verification yet.

Yet another generative AI use-case no-one actually asked for.

Right on, these days, you don't sell out - you get "sponsored" to "grill their [sponsor's] CEO" :) What most of of these proxy metrics don't factor in, is the abuse potential later down the road in performance reviews. Any dev worth their salt will *not* be completely honest in answering them.

Absolutely relatable, still have to look it up ocassionally!

If computers still looked like this I bet the internet would still be good