"I'm in favor of AI regulation because... Advanced AI is a risk to the public," he said today. But when asked for specifics, Tesla's CEO left empty-handed.
Elon Musk
Elon Musk today called for regulation of AI to minimize the risk of "potentially catastrophic consequences," but Tesla and SpaceX CEOs were dismissive of details.
At the Viva Tech Conference at a live event in Twitter space, Musk sidestepped questions about specific AI regulations he wants lawmakers to pursue, instead focusing on the importance of AI to his business, such as Tesla's Autopilot technology. The Autopilot system relies on artificial intelligence, and Musk has said that if it doesn't succeed, Tesla's valuation is "basically zero."
"We could have a potentially catastrophic outcome," Musk said. "The most likely outcome (of AI) is positive, but we need to minimize the likelihood of problems with digital superintelligence." I'm in favor of AI regulation because... Advanced AI is a risk to the public.
In response to questions from the audience about the specific regulations he supports, Musk only said "there should be regulatory insight into LLM," referring to large language models like ChatGPT.
When the European Parliament passed the Artificial Intelligence Bill, the EU this week became the first major government agency to advance comprehensive AI legislation, which outlines regulatory procedures for reviewing and monitoring AI systems that could endanger the public, such as Autopilot, and infringe on individual freedoms and privacy. But that's just the first step, and in the U.S., a divided Congress means no meaningful action is likely.
Musk today believes that "AI may be the most disruptive technology ever" and quips that he wants to live to witness "some AI apocalypse... But hope is not the cause of it.
Earlier this year, Musk urged slowing advances in artificial intelligence. He joined Steve Wozniak and other prominent politicians and writers in signing a letter calling for a six-month moratorium on the development of AI such as ChatGPT to understand how they work and identify their potential risks. A month later, however, Musk said he was building his own ChatGPT competitor, called TruthGPT, designed to counter the "political correctness" of other chatbot programs.
"I don't think anybody would agree to a suspension," Musk said today. "But for the record, I think we should pause.
Last month, OpenAI CEO Sam Altman appeared before Congress and called for regulation of AI. One idea he reportedly supports is to create an agency that licenses organizations that create AI models and have them pass certain tests before releasing new models to the public.
Brad Smith, Microsoft's president, supports a similar idea. Last month, the company released 41 pages. The report concerns controlling AI, which requires the United States and other countries to establish their own government agencies to specifically regulate AI.
At the same time, Musk also pointed out that current technical standards are not enough to limit the influence of artificial intelligence systems. Specifically, he said, CAPTCHA puzzles designed to prevent bots from accessing human-facing web pages are useless in the age of artificial intelligence.
"It's clear that with today's AI, computers can pass every 'Are you human' test, like identifying traffic lights," Musk said. "Open-source AI stuff can pass all human testing, so you need something better than this for identity verification."