Meta’s head AI guy says that’s not that smart
AI is currently too stupid to regulate, says Meta’s head of the artificial intelligence department in an interview with Financial Times. Yann LeCun, Meta’s chief AI scientist, sought to allay the most exaggerated fears about AI while taking a libertarian stance toward any real attempt to regulate the rapidly growing field of artificial intelligence.
LeCun is a Turing Award winner and a leader in computer vision and neural networks. His work in this field is cited alongside other major AI researchers as a reason for the current boom in artificial intelligence. The renowned researcher compared AI regulation today to what would have happened if governments had hindered the early internet. He says putting a stop to AI now would be like regulating jet airlines before they were even invented.
His comments are about how any regulation today would focus on “existential risks” for systems “that can even compete with cat in terms of learning ability, but which we currently don’t have.” He called it “counterproductive” and said , AI skeptics “want the takeover of regulation under the guise of AI safety.”
LeCun’s opinion attacks people like Geoffrey Hinton, another “godfather of AI.” who recently spoke out in favor of AI regulation. Hinton suggested that artificial intelligence based on our current large language models would be so advanced that it could harm humanity, whether through bad human actors or through some kind of self-awareness.
LeCun completely dismissed Hinton’s fears of this supposedly inevitable singularityHe said most people were too influenced by films like Terminator Imagine a time when thinking machines become smarter than normal people. Instead, he said that AI models “simply don’t understand how the world works.” They are incapable of planning. They are incapable of really reasoning.”
It refers to the concept of artificial general intelligence, or AGI for short. Companies like OpenAI claimed that the transition to true machine intelligence was near, but LeCun called this “over-optimistic” as it would require many “conceptual breakthroughs” to reach this stage of AI. Simply put, he says that scientists have no real idea how to get from ChatGPT to Skynet, at least not yet.
It’s a nice change to hear a prominent AI researcher try to dampen the hype around AI, but then LeCun makes a straw man out of the complaints of most AI ethicists. He doesn’t mention the terrible bias of AI art generators, Language modelsor autonomous vehicles. He also doesn’t mention how AI helps create one new wave of online cheating and disinformation.
LeCun was responsible for Meta’s AI research bot Galactica. The assistant was intended to help researchers speed up their work, but Meta took the bot offline last November after scientists discovered it misquoted text and even invented magazines. LeCun whined about the bots’ removal, saying: “It is no longer possible to have fun by casually abusing them.” Happy?”
Ultimately, AI will become smarter, even smarter than humans, LeCun admitted. But he claimed that he envisioned AI helping us or tackling larger societal challenges like curing cancer or tackling climate change, rather than causing harm. You know, the usual tech evangelist stuff.
“Intelligence has nothing to do with the desire to dominate. That doesn’t even apply to people,” he told FT. “If it were true that the smartest people wanted to dominate others, then Albert Einstein and other scientists would have been both rich and powerful, and they were neither.”
This ignores the worldwide celebrity status that Einstein had during his lifetime, but of course intelligence is not a necessary indicator of being a bastard. It’s similar what he has said in the past where he claimed that society had the ability to “prevent evil from attaining infinite power.” But this optimism will not help them facing job loss because your company thinks that way can save money from Replace personnel with AI.
