'We didn't vote for ChatGPT': Prime Minister of Sweden admits he uses AI chatbots for 'second opinions' and an awful lot of people are not happy about it

In a recent interview with Swedish site Dagens industri (via The Guardian), Ulf Kristersson, the Prime Minister of Sweden, said he regularly consults with AI chatbots including ChatGPT and the French service Le Chat to get a "second opinion" on decisions. Unsurprisingly, that turned out to be an admission he probably should have kept to himself.

"I use it myself quite often," Kristersson said in the interview. "If for nothing else than for a second opinion. What have others done? And should we think the complete opposite? Those types of questions."

That doesn't seem great, and I say that as a citizen of a country with a former Prime Minister who famously consulted mediums, took part in seances, and saw dead people. On the other hand, while there's certainly some crossover between psychic mediums and AI-powered chatbots, I don't recall King's mother ever suddenly going off as MechaHitler and calling for a second Holocaust. Important distinction there.

I'm not the only one who finds the whole thing incredibly dodgy: Experts in the field expressed concerns about Kristersson's AI dalliances too.

"The more he relies on AI for simple things, the bigger the risk of overconfidence in the system," Virginia Dignum, a professor of responsible artificial intelligence at Umeå University, told Dagens industri. "It is a slippery slope. We must demand that reliability can be guaranteed. We didn't vote for ChatGPT."

Writing in Expressen (Google translated), consultant and "AI enthusiast" Jakob Ohlsson called Kristersson's approach to AI "amateurish," saying that it's good the Prime Minister is curious about new technologies but adding that "he is inputting his political thoughts into a language model he does not understand, owned by a company he does not control, whose servers are located in a country whose democratic future no one can be completely sure of anymore."

And while a representative for Kristersson later said that "sensitive information" is never a part of the PM's AI usage, Ohlsson said that's not good enough because "a competent analyst can piece together the government’s strategic thinking from small clues," all of which end up in the hands of US tech companies "that are already becoming more powerful than many states, in a USA whose future political development is anything but predictable."

Writer and lecturer Signe Krantz wrote in Aftonbladet that "Kristersson has fallen for the oligarchs' AI psychosis," and warned that "chatbots would rather write what they think you want than what you need to hear." Krantz also asked why the PM would use "random number generators" for advice rather than "his large and well-paid staff of experts."

"It can take longer for an expert to fix the AI's mistakes than it would have taken to do the job from scratch," wrote Krantz. "It's just as embarrassing every time someone lends their thinking power to AI. When the Prime Minister does it too, it's a different matter. It can pose a security risk."

Not everyone is against Kristersson's chatty ways: In a separate Dagens industri editorial (Google translated), Tobias Wikström wrote that the Swedish PM "is absolutely right about AI technology," with a number of caveats including that "of course, the Prime Minister understands that one cannot trust the answers that the internet's enormous search capabilities provide, whether they are AI-produced or not." Which might make one wonder why one is bothering in the first place, but I digress.

People who make money—a whole lot of money—promoting the cure-all tonic of AI have in recent years poured great effort into convincing us all that the AI-powered future is both inevitable and awesome, as long as we get behind it with sufficient enthusiasm (and money).

But reality suggests something rather different: A torrent of slop, suckers, security holes, and a bottomless money pit that's swallowed countless billions of dollars, with countless more to follow. Will artificial general intelligence be a real thing someday? Maybe! But it sure isn't a real thing now, nor is its creation plausibly imminent, no matter what the high-priced hucksters say. Right now it's not much more than Eliza with internet access, and that's probably not something for the leader of any nation to be casually kicking around ideas with.