Elon Musk predicts that his company xAI could achieve artificial general intelligence (AGI) within the next couple of years, and maybe as soon as 2026, according to a new report from Business Insider. If it feels like you’ve heard that one before, it’s probably because you have.
Musk predicted the same thing in 2024, claiming AGI would be achieved by 2025. Take a look at any calendar, and you’ll see that we’re just a few weeks away from the end of 2025.
“How long until AGI?” asked Logan Kilpatrick, the head of product at Google AI Studio, in May 2024.
“Next year,” Musk replied, to which Kipatrick responded, “Big if true.”
It wasn’t true, of course. But Musk has a long history of, shall we say, optimistic predictions about his own company’s future accomplishments. And his predictions often have ulterior motives.
Remember when Musk was making the most noise about the dangers of AI and worries that it could destroy the world? The billionaire signed on to a letter in March 2023 calling for a six month pause in all AI development. It was revealed less than a month later that Musk was secretly building his own AI project at Twitter. By July 2023, Musk had officially announced the creation of xAI, the company that makes his Grok AI chatbot.
The CEO wasn’t earnestly worried about the risks posed by AI. He was just frustrated that OpenAI was way ahead at the time.
Musk’s treatment of AGI, or any new technology, largely depends on how he can hype his companies at any given point in time. And the perpetually prospect of achieving AGI, whether you think it would be good or bad for the world, helps drive investment in AI technology, the thing that seems to be propping up the entire U.S. economy at the moment.
The new report from Business Insider also says that Musk told xAI staff that investment in the private company was going well, with “around $20 billion to $30 billion in funding per year.” An email to xAI with questions about the report was met with an auto-response that simply said “Legacy Media Lies.” Musk has great contempt for the news media and previously had an auto-responder at Twitter that sent a poop emoji.
Part of the problem in discussing AGI is that there’s no single agreed upon a definition. As IBM describes it, we’ll have achieved AGI when artificial intelligence can “match or exceed the cognitive abilities of human beings across any task.” But obviously defining terms like “cognitive abilities” and “any task” is extremely complicated.
Other folks like to define AGI as a kind of self-awareness that would make artificial intelligence more like humans. Instead of just regurgitating words from its training data, the AI would understand itself as a kind of consciousness. People in that camp are excited and/or concerned about that theoretical tipping point because they assume it would be the start of the robot revolution and AI’s attempt to destroy humanity. Musk has hyped those fears tremendously, though he’s backed off recently.
Absent large robotic armies, achieving AGI in the present day with a system that loathes humanity would probably look more like the 1970 sci-fi movie Colossus: The Forbin Project, where non-humanoid systems engage nuclear weapons systems to threaten the world. We don’t really have the advanced humanoid robots for a Terminator 2 situation just yet.
But Musk is working on that too. He predicts Tesla will produce 1 million humanoid Optimus robots per year within the next five years, and they’ll even be babysitting your kids. He just needs to figure out how to get Optimus working without teleoperation before all of that can happen.
Who knows? AGI could magically be achieved in the next few weeks, and maybe Musk’s old prediction will come true. But the billionaire also has another prediction deadline just over the horizon. Back in October, Musk told Joe Rogan he’d demonstrate a flying car by the end of this year.
Read the full article here
