Shortly after San Francisco–based startup OpenAI introduced ChatGPT last November, it took the world by storm with its ability to turn simple prompts into answers and essays, albeit sometimes factually incorrect ones. Once Microsoft unveiled its new Bing with A.I. tech related to ChatGPT, the bot even developed an unsettling, sometimes “unhinged” personality. Elon Musk, who cofounded OpenAI as an investor in 2015, didn’t seem amused. Earlier this month, Musk criticized OpenAI for going against its initial purpose of being an open-source nonprofit that would “counterweight” Google. And in response to OpenAI founder Sam Altman’s tweet about people’s contradicting uses of A.I. last December, he said that “training A.I. to be woke—in other words, lie—is deadly.”
The danger of training AI to be woke – in other words, lie – is deadly
— Elon Musk (@elonmusk) December 16, 2022
Now Musk is reportedly working on building a team that will rival OpenAI. The world’s richest person (who recently reclaimed that title) has been in talks with A.I. researchers recently about creating a lab that would work toward developing a rival to ChatGPT, The Information reported Monday, citing people with direct knowledge of the matter.
While Musk has not confirmed any specifics regarding his new passion project, he has reportedly tapped Igor Babuschkin to lead the new A.I. initiative. Babuschkin worked at DeepMind, a subsidiary of Google’s parent, Alphabet, focusing on A.I. innovations. He also previously worked at OpenAI, according to his LinkedIn.
Babuschkin seemed to confirm the tie-up with Musk on a project in an interview with The Information, adding that the work was still in its infancy and that Musk aimed to build a more reliable language model rather than one with fewer restrictions.
“The goal is to improve the reasoning abilities and the factualness of these language models,” he said.
This isn’t Musk’s first run-in with A.I. inventions. The self-proclaimed free-speech absolutist cofounded OpenAI in 2015 as a nonprofit research organization, an answer to Big Tech companies that were making money from A.I. innovations. He eventually left its board in 2018, which OpenAI said was to avoid a conflict of interest with Tesla’s expansion of its own A.I. capabilities.
Since the buzz was rekindled last year, Musk hasn’t been shy to point out the shortcomings of ChatGPT and, specifically, the guardrails that OpenAI has in place. Earlier this month, when a user tweeted about ChatGPT refusing to answer with a racial slur after being prompted to do it in a hypothetical scenario, Musk replied to that thread, calling it “concerning.”
Concerning
— Elon Musk (@elonmusk) February 6, 2023
Musk has also been critical of Twitter’s content moderation in the past. Since he took over the social media company, Musk has cut many employees overseeing Twitter’s content moderation unit during layoffs last year and earlier this year.
Musk could not be reached for comment, while messages to Twitter and Tesla were not returned. Igor Babuschkin did not immediately return Fortune’s request for comment.
