It's just gonna pollute the internet with even more bullshit. Language models don't really understand topics, they just put together words that are likely to appear with each other. Biases are inherent to this design.

As long as LLMs aren't tailored to bias certain views, it may just lift humanity.

Press x to doubt on this.

Nothing about a glorified auto-fill will uplift humanity. We already see the opposite happening right now. LLMs output is already making the internet worse because its being flooded with auto-fill bullshit. And that in turn is making LLMs even worse over time, because instead of just being garbage out, now the garbage goes in and cimes out worse than the first time.

And its impossible for LLMs to not have bias, and ones claiming to not have bias are probably the worst of all.

LLMs are not the aspects of AI that are going to have the biggest impact. It will be AI based tools for evaluating data in the aid of specific tasks that has the big effect

I'm already seeing people come into software dev support forums asking "ChatGPT said you could do this but it's not compiling" and people replying that no, that's not possible and them arguing about it because ChatGPT said it.

Once Elon Musk unleashes his "uncensored" AI chat bot, we're going to be flooded with made-up misinformation, it's going to be a bloodbath.

Well I don't have a dog in the AI fight. I did sufficiently couch my comments as a thought experiment. I could have postulated the opposite scenario I suppose. Or none at all. I do see that there are some strong and confident predictions as to the outcomes.

As soon as the AIs start saying it would make the most sense to equally distribute resources and having 10 people hoarding all the wealth is bad for the economy they're going to get some adjustments real fucking fast.

I somehow envy your optimism.