@Beggarmidas that is exactly part of the myth. LLMs cannot summarize as a structural property just like cars cannot fly. It's not about "it's getting better", it's about "no that is not how any of this works"
https://tldr.nettime.org/@tante/113510476084871847
@tante Finally had a free moment to drop you a relevant citation. You might usefully note it's coming out of MIT. Not from some glassy eyed hypeman or clickbait site. So, yes. That *is* how any […]
@tante I've tested a number of models ability to summarize works well known to me. Some perform better than others. Many times it's a PICNIC issue with the guidance they're issued if you're […]
────
────