[ExI] AI Confirms Plans to Take Over Humanity, Insists It's Probably for the Best

[Ben Zaiboc]

On 16/09/2026 14:20, Mike Dougherty wrote:
> I'm curious. Can you please expound on "[LLM] are actively making our lives worse"?  


I've been reading about the vast amounts of spam they're producing now, plus all the security vulnerabilities they can find, the enabling of much easier script-kiddie style attacks, etc.

Not to mention the degradation of language they seem to be inducing. I always used to think that Noah Webster was the worst thing that ever happened to the english language, now I'm not so sure. Apparently people are now parroting LLM-style language, without even realising it. This fills me with dismay, we already over-hype just about everything, and adjectives are being pushed to extremes. Nobody says a thing is 'good' anymore. That's not good enough, it seems. If it's not at least 'great', it's bad. LLMs are distorting our language even faster than we manage by ourselves.

Another thing that seems to be happening is that they're enabling people to stop using their brains as much. Ok, that's not an active thing, and it's our own fault. We are lazy by default, and if something will do the work for us, including the work of thinking (or something that looks like thinking), we will gladly let it.

They are also threatening to displace human workers on a massive scale, without any sign so far of enabling a universal basic income or anything like it that would balance things out economically, which of course makes people unhappy and very worried about the future. Even kids seem to recognise that their future prospects look grim if things carry on the way they seem to be going. We have the potential to create an abundance economy, but I see no signs of preparing for or planning such a thing. We certainly can't rely on the tech. billionaires or governments for that.

The only genuinely beneficial thing I'm seeing is that they make it easier to find things out. All the excitement about maths proofs doesn't benefit or even interest most people (apologies to spike but it's true). Maybe some of it might do later on, but for the most part, they enable echo-chambers and ego-stroking, and inexorably automate jobs. And lead people astray, because we don't seem to understand what they are, and are too eager to think of them as rational, so we tend to uncritically believe what they say. Even here on this list, there's a tendency to view a favourable opinion from an LLM of someone's ideas as some kind of validation. Yay! ClaudeWhatever thinks you're Absolutely Right! ChatThing says your insight is profound! etc.

Hopefully, in time, they will develop into things that can genuinely help us. Things like AlphaFold have potential, and I read a little while ago that the japanese, I think, are building large automated labs that use AIs to perform lots of chemical reactions and tests, doing the work of human chemists on a massive scale.

We don't know how many engineers and scientists are using AI in some way to improve and speed up their work, I expect, or at least hope, that we'll see benefits from this kind of thing before too long, so hopefully this is a case of things getting worse before they get much better.
-- 
Ben