Tesla, SpaceX, and Twitter CEO Elon Musk is planning an entry into the rapidly expanding world of AI chatbots, with an AI project he calls “TruthGPT,” to counter the political correctness of Microsoft and Google-backed AIs.
Musk, who helped found ChatGPT developer OpenAI but later resigned his board seat over potential conflicts of interest with Tesla, said he had doubts about the organization, due to its close ties to Microsoft and the fact that its creators have trained it to be “deceptive” through political correctness.
“The intention with OpenAI was to do good, but it’s not clear if it’s doing good. It’s not clear at this point,” said Musk.
“I am worried about the fact that it’s being trained to be politically correct, which is simply another way of saying untruthful things. That’s a bad sign. Certainly, a path to AI dystopia is to train an AI to be deceptive.”
“I’m going to start something which I call TruthGPT, or a maximum truth-seeking AI that tries to understand the nature of the universe.”
“This might be the best path to safety,” said Musk. “An AI that cares about understanding the universe is unlikely to annihilate humans, because we are an interesting part of the universe – hopefully.”
Musk said he helped create OpenAI as a “counterweight to Google,” but that he failed to sufficiently focus on it, resulting in the company becoming a creature of Microsoft.
“I put a lot of effort into creating this organization… but I kind of took my eye off the ball, I guess. They are now closed source, and for profit, and closely allied with Microsoft,” said Musk. “In effect, Microsoft has a very strong say, if not directly controls OpenAI at this point. So you really have an OpenAI and Microsoft situation, and Google Deep Mind. [Those] are the two heavyweights in this arena.”
Early analysis of OpenAI’s ChatGPT revealed its strong left-wing bias. There were numerous cases of it refusing prompts that would generate conservative viewpoints, such as requests for a critique of drag queen story hour or praise of Donald Trump. When asked to answer questions on the political compass, ChatGPT gave answers that placed it squarely in the left-liberal quadrant.
OpenAI later sought to overcome the allegations of bias, with one founder saying the bias was a result of a “mistake” in the design process. OpenAI later appeared to fix some of ChatGPT’s biased responses, though significant traces remained.