arxiv.org/abs/2610.34073 been sayin this; straightforward naive RAG search over a corpus reliably outperforms complex memory systems, with summarization, fact extraction, dreaming etc mostly hurting rather than improving effectiveness

Show Post

The Reinforcement Learning Conference has, as of today, switched to rolling submissions! top tier review process + fast turnaround + desk-rejecting slop + getting to present accepted papers at RLC! rlj.org/news/rlj-now...

Show Post

7. So our argument is that LLMs reduce Barzun's House of Intellect into summarized statistical relationships that can be manipulated and remixed in the same ways as the oral tradition, but via processes that are automated and work at scale. Monuments of unageing intellect now have the wings of song.

Show Post

1. A quick thread to summarize my and Cosma's new piece on LLMs as Homeric songs at the Economist. The piece has its origins in bactra.org/weblog/feral... and this fantastic article by Murray Shanahan et al, which we couldn't link as we would have liked www.nature.com/articles/s41...

Show Post

If my job was manual labor I could do from home, I’d already be back full-time just to keep the baby entertained.

Show Post