Post #2369557
2026-02-15 07:54 UTC
Replies (5)
-
@robchapman@ohai.social 2026-02-15 08:03
@jwcph@helvede.net a box of rocks is less dumb!
-
@simon_brooke@mastodon.scot 2026-02-15 08:05
@jwcph@helvede.net they don't understand anything at all. That's not how they work. They statistically produce a sequence of tokens that resemble answers in their training set that answer questions that are statistically similar to yours. That's literally all they do and it's literally all they can ever do. #LLMs are an evolutionary dead end. They're not a step on the path to #ArtificialGeneralIntelligence. Any understanding or meaning you find in their answers is your own work. #AI #GenerativeAI #LLM
-
@Mndell@mastodon.social 2026-02-15 08:48
@jwcph@helvede.net tried it, LLM response: If your car needs washing… you’ll have to take it there with the car 😉 50 meters is so close that: The engine barely warms up You use almost no fuel The difference compared to walking is practically zero Walking only makes sense if the car is already there or if someone else is driving it. So: get in, drive the 50 meters, wash it, done. 🚗✨ #LLM
-
@chesheer@mastodon.bsd.cafe 2026-02-15 12:00
@jwcph@helvede.net Well, I'm not an AI enjoyer, but Gemini passes the test with flying colors. I wouldn't believe these "tests" on the Internet. Too much is dependent on the model and its actual settings.
-
@finner@appdot.net 2026-02-16 20:11
@jwcph@helvede.net This makes me think about the old saying that went something like "any advanced enough novel technology is indistinguishable from magic". So many people see the output of an LLM and think it's super intelligent, almost magical. But the reality is that every time these things APPEAR to be intelligent, it really is just a magic trick. They aren't ACTUALLY intelligent at all.