It's literally a cult.

“You have to be very careful,” Simone Fischer-Hübner, a computer science researcher at Karlstad University, told Aftonbladet, warning against using ChatGPT to work with sensitive information.

I mean, sending queries to a search engine or an LLM are about the same in terms of exposing one's queries.

If the guy were complaining about information from an LLM not being cited or something, then I think I could see where he was coming from more.

The typical pattern for leaders is to get "second opinions" from advisors who tell them whatever they want to hear, so... maybe asking the equivalent of a magic 8 ball is a marginal improvement?

I would rather have the politicians consult a plain old magic 8 ball than one controlled by Scam Altman.

You can see it on this government policies

It surely can't hurt, if it's to sanity check your highly paid advisors...

I really don't get it. These things are brand new. How can anyone get so into these things so quickly. I don't take advice from people I barely know, much less ones that can be so easily and quickly reprogrammed.

I would genuinely have more trust in a politician consulting I Ching than ChatGPT - at least the book of changes requires some thinking and interpretation, fucking hell...

Sigh, all the tech-bros that convinced the general public that AI are in any way intelligent or that they can think, reason or understand concepts like truth/lies or right/wrong should go on trial for crimes against humanity...

Some of y'all are crazy reactionary. There's absolutely nothing wrong with asking an ai chatbot for an additional opinion. The ai shouldn't be making the decisions, and the ai shouldn't be the only way you look for opinions, but there's nothing wrong with ai being ONE OF the opinions you consider

But it doesn't know anything. At all. Does Sweden not have a fuck ton of people that are trained to gather intelligence?

Except those prompts are retained by OpenAI, and you don't know who's got access to that. They've had chats leak before.

If that's your problem with this story then you're just saying they should run a local instance of chatgpt on their own local computers

Just a moment...

(www.technicalexplore.com)

Most people don’t care whether AI is intelligent, can it think or reason or understand concepts. What matters is that it can give helpful replies, and it does that a lot. In my experience maybe 1 reply out of 10 is seriously flawed, others are either mostly helpful or just tell me something I already knew until I reprompt for more, which, again, also works well most of the time (especially when you allow it to search for more information online). So if you wanted to say it’s dangerous in some ways, this is definitely not the proper way to say it, since neither it being dangerous nor it being right or wrong or helpful or useless has anything to do with intelligence, ability to think, reason, feel, comprehend or whatever.

I mean, I don't really love that he uses AI to begin with, but getting an opinion on top of what you get from your advisors that you hired could be good in theory. Except that most AIs are programmed to be equally sycophantic and they're just plain wrong most of the time.

To me the most irresponsible part is sending your thoughts as the leader of a country straight to a US run company, you just know those prompts are being digested by the NSA directly, and possibly other intelligence agencies.

It's weird for a head of state to consult their mentally challenged imaginary friend?

AI chat bots don't have their own opinions. All they do is regurgitate other opinions, and you have no idea the motivation for how those opinions are weighted.

As i said in other comments, none of that matters. The point is to gather ideas that you otherwise might not have come across. And also as said already, ai generated opinion pieces should never be the only opinions you hear.

What use is an opinion that can neither be explained or defended by the person giving it? How is that useful to a person making decisions for millions of people?

It can do those things. And you need to re-read what's already been said on this comments page, because how it's useful has been explained several times already