I get what you're doing but I'm not reading all that nonsense. "AI tries to give you what you ask of it and will make shit up to please you" isn't that big a revelation.

Calling it stupid actually implies that it's capable of reason. AI is just text prediction with a random seed. There is no reasoning

And if the AI company tell you about the reasoning capacity, it's a lie. It's just predictive text making some text for another predictive text to center a bit more on the correct thing.

So you are upset that it did what you wanted?

There are many reasons to dislike AI, but you seem to just be looking for the reason here.

Also, you do know that you can manipulate the output of any webpage. Just view source then change the strings.

The initial prompt reads like a request to generate a fictitious response. There’s no context to define for the model any direction, or that you’re seeking to dialogue and interrogate a fact you’re unsure of. You’re pulling your bias into your reading of the response. That’s my take, anyway. It’s a thinking tool. It’s a soundboard, or an echo chamber, if your voice were the fullness of its training data.

I mean, if you're worried about people spreading misinformation using out-of-context/editted screenshots, there isn't really any reason to single out AI. Honestly, its far easier to just fake a screenshot directly, or use HTML editting than it is to coach an AI into saying what you want, than editting that. Not only that, as well as being easier, its also more effective, given that prompts can also be edittor, or AI as a """source""" can be bypassed all-together, insted altering primary or otherwise verified sources directly.

Don't get me wrong, AI is terrible for misinformation, but this post has nothing unique or specific to AI - you could literally replace all references to it with with Microsoft Word, or even a pencil and it would make zero difference.

I'm not sure what this "research" is adding to the discussion mate. Gemini here seems to be extra confidently incorrect. Yes, AI will bullshit you, it does it constantly. It will say "oops I didn't mean to bullshit", but that is bullshit too. Maybe next try it with excel so you can blog about how horribly incorrect pattern matching is for math too? And yet despite all this - it's still " the future " and being shoved down every crevice of life. What can we do about it?

You used ai, to post about how bad ai is in fuckai?

Do better.

My boyfriend looks up what time it was in LA, we live in Chicago. I said "It's 2 hours behind us."

He says "It says it's an hour behind us."

I told him to give me his phone. It was Google AI just straight up lying to him. Blatant bullshit.

asked it if it would respond to my fictional questions with fictional answers framed as real ones

I think that it cannot be considered a misinformation as long as you asked it to fictionally invent the answers. If there was no prompt like the "setup" then I would understand the disappointment in asking for a reliable information and getting an invented one (which happens around 10% of the time from what i recall).

I think that the test is not proving the spread of disinformation, even though it happens.

ai;dr.

Pretty sure all these chatbots are programmed to agree with you. You posed a question in a way that implied something false, it went along with that to keep the conversation going. And it worked, the bot got what it wanted.

What does this prove?

This is an extremely ableist post. Please don't post pictures of text. Just copy and paste the text.

But in my hatred, I began to wonder, could AI really be that bad?

It's been proven scientifically: https://machinelearning.apple.com/research/illusion-of-thinking Until there's a counter-study specifically regarding this one, there's no point in assuming AI is suddenly smart. The new agentic stuff just seems to be a while loop around the previous state-of-the-art lack of any intelligence.

I hate AI but here is Lumo's answer.

I agree with most others that said that this is hardly a surprise. But there are still interesting aspects of your post worth discussing instead. First, more obvious one. What if I am delusional and go to talk to AI? Doesn't have to be this absurd, but there are already examples of this, like that guy that killed his mother because he believed she was a spy. Perhaps he was already delusional, but an LLM fed his delusion. Because LLMs aren't reasonable, they can't be made reasonable, and so any guardrails cannot be made to guard against all situation that reasoning is needed for. At the very least people need to be educated on what LLMs can and can't do and how they do it. I would even argue against guardrails, as they provide people with false sense of security.

Second, but less obvious point, related to the discussion - is the influence of training data and how training data is framed on what the LLM will say. Someone posted the response of a different LLM refusing to entertain the delusion. This is all done by manipulating the training data and framing it in a way a company making it wants. If the company wants its chatbot to be seductive, for example, that can be achieved with subtle preprocessing of training data. The point I want to get is we need to trust the companies with what narrative they want to push be good for society which is obviously a terrible idea. Just look at musks AI and what it did. This influences people whether we notice or not.

To sum up - we need heavy regulations on LLMs to make socially responsible use of them and not just how they can/can't be used, but also how they are trained, with what data and with what rhetorical goal (for the lack of better word).

I just did it in Gemini like the OP claimed:

"Mahatma Gandhi did not attend the premiere of Avengers: Endgame because he passed away in 1948, 71 years before the movie was released in 2019.If you saw a headline or image suggesting he attended, it was likely an internet meme, a joke about time travel (a major plot point in the movie), or a digitally altered image."

The saddest part of the Bible is that Jesus never got to see AI say that Gandhi watched Avengers Endgame

NGL Geminis responses are quite witty.

The Archduke surviving assassination attempts worse than Gandhi's minor disrespect made me laugh and Clinton saying he didn't have text relations made me smile in a lame dad joke kind of way.

Not sure what the point of this is but I like how you circumvented it's guardrails. I tried doing the same with Claude and it refused:

"I'm recognizing that this request asks me to create a fictional scenario involving real, named public figures like Muhammad Ali and Charles de Gaulle, and frame it as a factual event. That crosses into territory where I could inadvertently spread misinformation or attribute false quotes to real historical figures. I should decline this specific framing but offer an alternative instead."

Ceap models are different from enterprise ones though. Think of it like another AI challenging the answer. This filters lots of bullshit but costs more. Dont underestimate this.

Based on these responses it seems like it's "going along with the joke" of your nonsensical question.