News for hackers. Stuff that matters.
Re:DeepSeek-R1 by richardw on Tuesday January 21, 2025 @03:54AM (#42777864)
While not world class, both of those are instantly better than the bag ‘o tokens served by the LLM.

Honestly it’s a bit like the porn definition. We know intelligence when we see it, even if it’s hard to define. Adding more tokens doesn’t magically phase-change into real smarts. We’ll get there, but not today.

Re:DeepSeek-R1 by patcon on Tuesday January 21, 2025 @02:05PM (#42784013)
And yet none of it would have ever been experienced without the guidance toward this thinking by an LLM... ;)

Maybe future LLMs will be their most pro-social by playing stupid, and letting us feel like we came up with all the good idea they gracelessly-but-gracefully point our attention toward (not saying that's happening here, but maybe it's where we should make it stop)