I would like to take at face value Elon Musk's claim that he is developing a "maximum truth AI" called "TruthGPT". I really hope that all of this comes from a place of honesty and integrity, and that he is trying to create a better, fairer world. Whatever form it ultimately takes, I want to believe that Musk's new AI venture will legitimately produce an "AI that seeks to understand the universe."

I'm ready to get on board with it, but he's making it very difficult.

In particular, the source of this new information is an interview with Tucker Carlson of Fox News (via TechCrunch (open in new tab)); drawing on an interview spot on Fox and Friends (open in new tab), the interviewer describes Mask's sense of humor, his "sense of humor," his "sense of humor," his "sense of humor," and his "sense of humor. s sense of humor and the fact that he doesn't take himself too seriously, then uses the phrase with a straight face, which is doubly harsh:

"People like Stalin and Chuck Schumer, who take themselves too seriously, make me uncomfortable.

Tux, that's a powerful deadpan. At least give me a wink at the end.

Anyway, this is just an arresting aside to effectively announce on Fox News that Musk is committed to creating an alternative to OpenAI and Google's own generative pre-training (GPT) AI.

"We're going to start what we call TruthGPT," Musk tells Carson. Or a maximum truth AI that tries to understand the nature of the universe."

[14 [This may be the best path to safety, in the sense that an AI that seeks to understand the universe is unlikely to annihilate humanity.

In other words, it sounds lovely. A kind, thoughtful, caring artificial intelligence is at best "unlikely" to suffocate us all. Of course, there is always the possibility that it could annihilate us all, but that's the kind of risk we take when we play with AI.

That's why Musk and Woz have teamed up to call a halt to AI experiments. Steve Wozniak, as far as we know, has not purchased millions of pounds of AI-powered GPUs (opens in new tab), founded a new company called X-AI, or promised on national television to create a new generative, pre-trained AI.

That said, the idea of competing with ChatGPT is worthwhile, given its right-wing-inspiring appearances on Fox News and the disturbing similarity between TruthGPT and the name of the former orange president's abortive social network. After the Microsoft deal, OpenAI shifted from its initial non-profit, open-source foundation to a more commercial one.

And it seems to me that if you are doing AI for profit, you may be moving away from pursuing something like a universal foundation of truth. But if you are trying to change course against rival AI projects, you are also trying to add your own biases to the model.

Especially if it appears on a banner that says: "Elon Musk: Leftist programmers are training AI to lie."

Perhaps we need to bring in some right-wing programmers to tell the truth?

We don't know what Musk's real plans are for his AI project, but we don't know if it will bear the name TruthGPT, which he first created on Twitter. He is looking for investments (open in new tab), hiring former employees of Deep Mind, and marketing on Fox News.

But it may take some time to fully discern what form it will actually take.