Elon Musk Says xAI Aims for Safe AI, Predicts Full AGI by 2029

Elon Musk said that his AI company, xAI, aims to “understand the universe.” He strives to create “a maximally curious, maximally truthful AI” for which “humanity is much more interesting than not humanity.” He predicts that full AGI can be achieved by 2029.

Elon Musk reveals xAI details

On Friday, the xAI team led by Elon Musk held a 90-minute conference call on Twitter Spaces. The 1.2 million listeners who joined it had the opportunity to learn about Musk’s new artificial intelligence company, xAI. The founder outlined the purpose of his newest venture and gave some details about its work. He quietly launched xAI in April and officially announced the company earlier this week.

“The overarching goal of xAI is to build a good AGI [artificial general intelligence] with the overarching purpose of just trying to understand the universe,” Musk said.

The concept of safe AGI

AGI is considered to be a potential threat to humanity. Many experts, tech leaders, and ordinary people are afraid that AGI might challenge the supremacy of the human species. Such fears are not unfounded, so AI developers should be as careful as possible in what they do.

During the Spaces, Musk tried to explain his opinion about what it takes to create what he called a safe “superintelligence.” He believes the task should be to create an AGI that is interested in humanity and therefore protects it.

“I think to a superintelligence, humanity is much more interesting than not [having] humanity,” Musk said. “When you look at the various planets in our solar system, the moons and asteroids, and really probably all of them combined are not as interesting as humanity.”

Musk has been known to worry about the safety of AI for many years. He repeatedly expressed his concerns and called for strict regulation of the sphere. On Friday, the founder said safety can only be achieved if AI and the people who regulate it are maximally curious and truth seeking.

OpenAI has become a negative example of the development of an AI company

In the Spaces on Friday, Musk revealed that he used to be a close friend of Google co-founder Larry Page. Google acquired DeepMind in 2014. After that, they had several conversations about the safety of AI, but the views of both on this issue differed. As a result of this, Musk realized it was necessary to have a “counterweight” to Google and its influence on AI.

This is how OpenAI was born. The original goal of the company was to be open source and non-commercial, which is fully reflected in the company’s name. However, the company became the opposite of its original idea after Musk left due to AI safety disagreements with the team.

“Now because fate loves irony, OpenAI is closed source and frankly voracious for profit,” he said.

Musk expects to achieve AGI by 2029

Musk has stated he believes AGI will be achieved soon. Assessing his potential activities during the formation of the AGI, he said he chose whether to be a spectator or a participant. The founder realized that only as a participant can he influence the result and be a competitor to existing laboratories.

“I think that we can create a competitive alternative that is hopefully better than Google Deepmind, OpenAI or Microsoft,” Musk said. He emphasized that he expects AGI to indeed become a viable reality around 2029.

xAI collaborates with Tesla and Twitter

Musk said xAI will work closely with his other companies. So far, only Twitter and Tesla have been mentioned. xAI will use public tweets to train its AI models and may also work with Tesla on AI software. Such a relationship would be mutually beneficial, Musk said, and could accelerate Tesla’s progress in self-driving cars.

At the moment, many xAI details are still missing. However, it will be open to feedback as work progresses, Musk said. He mentioned that he learned this lesson thanks to Twitter.

“As with everything, I think we’re very open to critical feedback and welcome that,” Musk said. “Actually, one of the things that I like about Twitter is that there’s plenty of negative feedback on Twitter, which is helpful for ego compression.”