Gregory's Server@bhawthorne frustratingly, "AI", in this case LLMs, has been used to provide verbose webspam to maximise ad impressions. I know LLMs aren't popular in much of the Fedi, but ChatGPT gives a helpful and concise answer to the original question. There's mileage in LLMs to leapfrog all this web junk we have to wade through today.
|
12 comments
@donnodubus fair question. My feeling is that over time, we graduate to authority figures. If you want legal advice, you don't try and determine which statutes to read, you ask someone who's synthesized and absorbed them and can hopefully regurgitate relevant parts, ideally with interpretation along the way. If I had a chef on hand, I'd probably ask about hazelnut roasting! Although current LLMs are less than reliable, and there's a moral (and legal?) question around their training, of course  @andrewgretton If you consider how knowledge and truth are actually arrived at, it's quite clear that LLMs are fundamentally incapable of achieving that kind of intelligence. No matter how advanced they get. The best LLMs are simply better at defuzzing the source materials they've fuzzed. Which leaves the obvious observation that they're roundabout serving up the content they've stolen. Content we could just read directly... @donnodubus for sure, and the reference source materials are more authoritative. However, consider the utility of an LLM that's "right" most of the time. If it can tell me - more quickly and mostly accurately - how to roast hazelnuts, the average internet user will prefer that experience to Googling/DDGing/etc which today is a toxic experience due to webspam etc. It's utility over ethics; maybe Napster all over again? And we know what "won" for years until iTunes and Spotify won. For a while. @andrewgretton well the webspam problem is driven by AI, so that's a self-fulfilling prophecy. Convenient that these new capital ventures are being sold as the solution to the problem they made... @donnodubus very good point. Albeit depressing 😬 There's probably a good amount of money in a genuinely useful and entirely new search engine these days. I'm not sure Kagi and co are it, though, despite their innovative approaches. @bhawthorne I hear you and respect your choice. For me personally, LLMs have passed the utility threshold already such that even when I know that X out of Y answers are plain wrong or even potentially dangerous, it *still* has sufficient value to continue using, as long as I go in eyes open. And my assumption - a shaky one, perhaps - is that the tech will improve, along with the hardware that runs it.  @andrewgretton My problem with LLMs is that they are trained to provide confident answers that imply they are correct, even when they are dangerously wrong, so I have no way of knowing whether a given response is correct. I’ve spent 60 years learning how to ask and answer questions to further knowledge. LLMs make that quest harder, not easier. @bhawthorne @andrewgretton And if you know that "X of Y" answers are wrong or even dangerous, what are you going to do with each answer you get? Assume it's not part of 'X' and YOLO? Otherwise you have to do the research to find out whether the answer is in 'X', which means you've done more work overall than if you'd just not used the LLM in the first place. @bhawthorne @andrewgretton while I get your point about the _potential_ for LLMs to give you garbage, but asking ChatGPT, I get this: "Roast hazelnuts at a temperature of around 350°F (175°C). This allows them to toast evenly without burning. Keep an eye on them and shake the pan occasionally for even roasting. They typically take about 10-15 minutes. You'll know they're done when the skins crack and they become fragrant" which seems to jive with both your, and my intuition that it's 350. |
@andrewgretton Why should anyone use an LLM to answer a question like this as opposed to using the authoritative source that the LLM is giving only a fuzzed impression of?