Post by @supersquirrel@lemmy.ca

2026-10-29 19:24 UTC

Maybe we could make a system where people submit questions and humans can upvote and downvote these questions to help crowdsource the most important questions and humans could also respond with comments in threads takes bong rip and here me out those comments themselves could be subject to upvotes and downvotes that humans could further use to help elucidate the range of opinions of people involved in a digital community about a question or bit of news. NAH Surely the solution is AI!!!

Replies (2)

@Romkslrqusz@lemmy.zip 2026-10-29 19:57

That works well when you have time and the right humans with the right knowledge happen to be frequenting the space. In a lot of cases, I have seen that system used to offload the research burden onto somebody else. Those spaces have been increasingly taken over by bots, astroturfing, or armchair ‘experts’ presenting mis/disinformation or parroting something they saw on Youtube out if context. Granted, a lot of LLMs are also trained on this same broken information. For more abstract concepts, though, it gets to be a bit trickier. One of my main advantages using LLMs has been for electronics repair. Without a schematic, I’m sometimes limited to obscure markings on a part that I need to replace. I used to sink quite a lot of time combing through datasheets trying to deduce a match, with LLMs that already have those datasheets integrated I inout the marking and form factor to get connected to a limited set of candidates to verify for myself. Hell, in some cases a picture of a half-blown component has gotten me where I need to be. I just don’t need that level of depth for every query, when I want tacos near me a map with indexed keywords is more than enough.

Open ##5314483
@robjwells@awful.systems 2026-10-29 20:31

[Closed and marked as duplicate]

Open ##5315330