Tonight, commented about how the writing about how AI "has hallucinations" is an inaccurate characterization, I find that the person who shared the post with a bit, actually has done AI research. And is someone that I've talked to in the past before.

often the world can become "very small" in the sense that if you're in the right kind of spaces, you can just stumble across people who really do know about a topic, and will just treat you like a real person even if you don't have really the kinds of "education" required. As long as you aren't being some sort of dick about things.

this is an odd... "related" article...Hofstadter's butterfly - Wikipedia

hallucinations is inaccurate. AI (of the LLM variety) is essentially an extremely prolific sociopathic BS-er - saying whatever suits it in the moment without any regard for truth. It's amoral. It has no real intelligence nor real context.

How great that you've found people to discuss things with.

sef_lopod: I find "hallucinations" to be a fair, relatively "neutral" term. Saying, "sociopathic BS-er" and 'saying whatever suits it in the moment," is giving far more agency to AI than is fair. Amoral, I'll accept, as morals aren't even a "concept" that applies.

I guess, it's whether you want to give it "personhood" that isn't warranted.

mimipal: I spend a fair amount of my "social time" over on "The Fediverse". I still check in here. And sometimes feel something is worthy of bringing over here.

I'm glad you do still check in here. And that you post what you think is worthy.

There's a few people I like to keep track of here. But it's kind of quiet overall.

making the point that if LLM were to be regarded as an intelligence/person then they've been building the worst possible one.

but hallucinating is much more a property of a sentient person (sensing reality but also getting it wrong). So you would be the one granting it personhood. BS-ing is distinct from lying in that there's no regard for the existence of truth, just whatever seems to work in the moment.