In recent days we've learned that Elon Musk wants in on the AI game, and now we learn why: He's worried that ChatGPT and other chatbots are being trained to be too politically correct, and this will lead to dystopia, he says, and so he wants to create his own "TruthGPT."
Every time Elon Musk talks these days it feels like he's at war with just about everyone. But not, apparently, Fox News or Tucker Carlson. Once again digging in his heels and showing us how he's fully in bed with the Murdoch/Fox morass, days after giving a combative interview to the BBC about his handling of Twitter, Musk wanted the Fox News audience to know about his plans to build his own AI that will be allowed to be racist, or something.
"I'm worried about the fact that [ChatGPT] is being trained to be politically correct, which is another way of saying untruthful things," Musk said, playing into Carlson's narrative of several years about censorship and liberals. "Certainly the path to dystopia is to train AI to be deceptive."
Musk said that his TruthGPT would be "a maximum truth-seeking AI that tries to understand the nature of the universe," and, I guess, will attempt to be anti-woke?
Musk said he knew that he was "starting late" in the AI game — and he was one of the original co-chairs of OpenAI almost a decade ago, so he knows just how late. But, true to form, Musk has launched a new company, based in Reno, alongside the company he launched last month to be the parent of Twitter — they're named X.AI and X Corp., respectively.
We all know that whipsmart AI talent is gonna want to live in Reno!
Should it be unsettling that Musk seems to be conflating his own culture-war crusade with the AI race? Four months after he tried he make big waves with his Twitter Files revelations — which centered on purportedly liberal leadership within Twitter who had knowingly suppressed mostly bullshit stories around the 2020 election because they looked a lot like election interference/bullshit — he's telling the world that it needs an AI to be built under his watch that will seek out the "truth"? Would that be, like, Donald Trump's truth, or Vladimir Putin's truth, or just Elon Musk's truth? Something like Truth Social, perhaps?
Barron's is already saying that this should be further cause for concern for Tesla investors — with Musk likely to have put up his own capital, possibly by selling more Tesla stock, to get yet another new company off the ground, not to mention adding this new distraction.
Meanwhile, Musk is speaking at a marketing convention in Miami today about his plans for Twitter, and a group called Stop Toxic Twitter paid to fly the banner you see below over the conference at Miami Beach's Fontainebleau Hotel on Monday.
On Fox News, Musk echoed the sentiments of the open letter he signed on to a few weeks back, along with a host of AI engineers and others, about the need to pause or slow down the AI race. Some surmised that this could be selfish on his part, wanting to launch his own thing, but it also echoes sentiments that NYT columnist Ezra Klein discussed in a recent opinion piece, that many in the AI field want to be regulated, and the government is way behind on this.
"It's not fun to be regulated. It's sort of arduous to be regulated," Musk said to Carlson. "Some people may think I'm some revelatory [sic?] maverick that defies regulators on a regular basis. This is not the case." Musk said he's used to regulation, with SpaceX and Tesla, and that AI definitely needs a regulator right now in order to stand a "better chance of advanced AI being beneficial to humanity."
Also, Musk seems to be philosophizing — poorly — when he says, of his "truth"-seeking AI, "I think this might be the best path to safety in the sense that an AI that cares about understanding the universe is unlikely to annihilate humans because we are an interesting part of the universe."
Personally, if I were an AI being trained on the "truths" of Truth Social, Twitter, and Tucker Carlson's show, I'd think about annihilating humanity! But that's just me. And maybe that's not what Elon means. Interesting choice of platform for your announcement, though, buddy.
