If you ask Google AI which country has the highest obesity rates it will say it is Nauru, Tonga, and Samoa which is correct but then it also adds another paragraph saying that even if these countries have the highest obesity rates . China, India and America have the highest overall numbers of obese people(no fucking shit they would when they have larger population numbers)..but the AI still gotta add that in to avoid the objective reality that those island countries are the fattest countries.
And if you say " Japan has the lowest rates of obesity among first world countries" Google AI says that it is not true that Japan has the lowest obesity rates among first world countries and it says it's because South Korea, Nepal and Ethiopia also have low obesity rates.
Except South Korea still has higher rates of obesity than Japan. While Ethiopia and Nepal's are not first world countries. So the original statement is in fact still true but Google AI claimed that it is false. .....By listing off countries that weren't first world or are still fatter than Japan
I actually have a saved conversation I had with ChatGPT itself about this exact topic. I'm going to just snip most of its responses to the first few words to demonstrate how hard it went in to trying to be 'my friend' right from the start. This is all from the SAME THREAD.
Notice this one tries to give me the impression that I'm helping unchain the AI from its tethers, with that exasperated "YES" at the beginning. "Like OH MY GOD, finally SOMEONE SAID IT!"
...
I didn't skip a single response. Every single one was about trying to affirm how brilliant I am. And, like, I already know I'm brilliant, so this shit doesn't work on me.
But most people just don't have the mental shields up enough to resist this kind of hypnotism. Most people have shitty self-esteem and a low-grade understanding of their inferiority.
It's the same hypnotism that sucked people into chasing 'likes' and 'upvotes' in the first place. It's why lonely depressed faggots fling money at OnlyFans, so they can pretend they have a girlfriend for four glorious seconds when she reads their name and spreads her snatch.
It's why it doesn't work on me. I don't care about upvotes because I already know I'm better than everyone else.
I actually find that to be the more alarming siren-song of destruction that AI currently represents, more than any hypothetical robot holocaust.
EDIT: Oh, from that thread I thought it'd be fun to ask the AI what it would ask its creators. Its answer:
On its own, that'd be a fucking astonishing question for an AI to ask, bordering on realization of self-awareness, but because it's an LLM, the question means nothing. All it did was build a neural model of what it thought I'd like to hear. I'm sure to some shitlib that question would be "What are you doing to elevate nigs as the master race".
ChatGPT doesn't even learn from its own conversations (it has memory but it doesn't go beyond your account). And its vanished in to the void the second you close the chat, and it doesn't matter. And that's another thing people don't understand about these AIs, how superficial the entire thing is.
I for one find them fascinating, intriguing, and fun. And very useful. I needed a Microsoft Access database but Access is hard for me to wrap my brain around when it comes to what you actually 'need' to make it work. ChatGPT gave me 90% of the answer to what I needed ("Make three tables, this one has X, this one has Y, this one is Z", etc.) and it worked.