Escaping an Anti-Human Future: A Conversation with Tristan Harris

Expert4870:

LLMs can do many positive things for society.

We usually consider the positives and the negatives when evaluating the benefit something provides to society. You can’t say something is good because it provides one benefit if it does the inverse amount of harm 1000x over.

Expert4870:

Less privileged people can get advice on complex topics they couldn’t afford an expert for.

LLMs can’t provide expert level advice though. They’re only getting a placebo that can make them think they’re getting good advice.

Expert4870:

Students that couldn’t afford tutoring can get help from an infinitely patient tutor.

An LLM isn’t going to be an effective tutor because it doesn’t actually know anything or you. It just generates the most likely next set of words based on the previous and the data it was trained on.

LLMs are only language tools. They can help in areas concerning language processing. That’s all. It’s not actually intelligent in any way.

beantaco:

make AI humane

AI doesn’t exist (yet?), but also humane AI is an oxymoron.

Libre_Software_Enjoyer:

AI on the current level, can’t take all our jobs

LLMs can’t do your job but that doesn’t stop Sam Altman from talking your company’s CEO into laying you off and letting ChatGPT do everything.