The event showed how Musk initially appears to be more focused on answering deep scientific questions, rather than competing with OpenAI, Google and Microsoft to create AI products for consumers. Musk said he saw his venture as an alternative to larger AI companies, but said xAI was still “embryonic” and would take time to catch up with OpenAI and Google.
The new wave of AI tools has surprised many AI experts with their capabilities, and some leaders in the field have moved up their timelines for how many years it will take for an AI to be generally smarter and more capable than any human. But the current crop of technologies work by predicting what image or word will say next, and there is no evidence that they are capable of solving unanswered scientific questions. Many AI researchers say that human-level AI could be decades away, if it ever happens at all.
Musk said questions about the nature of dark matter or unresolved aspects of how gravity works could be among the mysteries his company is trying to explain. He spoke at length about the “Fermi Paradox” – a theoretical question that asks why humans haven’t discovered aliens yet, suggesting that technological civilizations usually destroy themselves – or are destroyed by an external force – before traveling to other solar systems.
Jimmy Ba, a professor at the University of Toronto and a well-known AI researcher who joined Musk’s company, said the goal is to create an AI that can help with a number of major problems facing humanity.
“How can we build a general-purpose problem-solving machine to help all of us, humanity, overcome the most challenging and ambitious problems out there, and how can we use these tools to empower ourselves and empower everyone,” Ba said.
Musk, who has criticized existing chatbots for being politically correct, said his company’s AI could offend people because it would speak “the truth”.
Musk has been outspoken about AI for years, famously saying in 2014 that inventing super-intelligent computers would be like “summoning the demon” and could pose an existential threat to humanity. He co-founded ChatGPT maker OpenAI in 2015, but left the company in 2018 after disagreements with the other executives. In recent months, he has complained that OpenAI and other AI companies have been scraping Twitter data to help train their bots.
Tesla, the publicly traded electric car company Musk owns, already has a robust AI team that has been working for years to give its cars self-driving capabilities. Musk has said that his new AI venture will work closely with both Twitter and Tesla. Investors have criticized Musk for using Tesla resources to help his other businesses, particularly Twitter.
“We’re going to work with Tesla on the silicon front and maybe also on the AI software front,” Musk said, referring to Tesla’s custom silicon data chips. “Obviously, any relationship with Tesla has to be an arm’s length transaction because Tesla is a public company with a different investor base.”
Training the “large language models” that form the backbone of modern AI tools requires massive amounts of data and stacks and stacks of computers to crunch all that information, an extremely expensive process. Tesla has significant data processing capacity, and has used it in the work to build self-driving technology.
At the Twitter event, Musk said his company would need computing power, but suggested it wouldn’t be as much as other companies currently use. The team would remain relatively small, said Igor Babuschkin, a former Google and OpenAI researcher who joined xAI.
Musk initially delayed the start of the event to wait for more listeners to join. He said he adjusted Twitter’s algorithm so the space appeared to more people. Around 34,000 people listened when it started.