Meta AI chief doubts AI super intelligence is coming “any time soon”

TL;DR: As the AI arms race accelerates, Meta’s chief AI scientist, Yann LeCun, has taken a grounded approach to the future of AI. He challenges the popular opinion that AI will surpass human intelligence in the near future and isn’t making any big bets on quantum computing either.

While the tech world is brimming with optimism about AI’s potential, Yann LeCun, Meta’s chief AI scientist, injects a dose of reality. Speaking to CNBC on the 10th anniversary of Meta’s AI lab, LeCun expressed skepticism about AGI (Artificial General Intelligence) going big and questioned the current capabilities of quantum computing.

He later clarified via his LinkedIn: “By ‘not any time soon’, I mean ‘clearly not in the next 5 years’, contrary to a number of folks in the AI industry. Yes, I’m skeptical of quantum computing, particularly when it comes to its application to AI.”

LeCun’s opinion sharply contrasts with Nvidia CEO’s prediction that AI will match human intelligence within five years. A pioneer in deep learning, LeCun believes that current AI systems are decades away from achieving a level of sentience or common sense. He argues that the industry is more likely to develop “cat-level” or “dog-level” AI before reaching human-level intelligence.

LeCun thinks ChatGPT is “not particularly innovative”… ouch

The fact that current AI training methods mostly rely on text data makes it quite limiting from his perspective. LeCun points out that text is a “very poor source of information” for training AI, as it lacks the richness and complexity of real-world interactions. This is why Meta is exploring multimodal AI systems that integrate audio, image, and video data, moving beyond language models like ChatGPT.

Microsoft’s president also recently stated that the creation of super-intelligent AI is not on the immediate horizon and is “going to take years if not many decades.” While tech giants like Microsoft, IBM, and Google invest heavily in this area, LeCun questions how practical and feasible it would be to create useful quantum computers, as classical computers can solve most problems often attributed to quantum computers.

LeCun’s views also contrast with Elon Musk, who has frequently expressed concerns about AI posing an existential threat to civilization. In a podcast, LeCun dismissed Musk’s fears as “completely ridiculous,” arguing that just because AI might become smarter than humans, it doesn’t necessarily mean it will seek to control or harm humanity. Meta AI, under LeCun’s guidance, is focusing on AI that hopes to make a “real difference” with practical applications.

With projects like Audiobox and Seamless, Meta wants to push AI into new territories like audio generation and language translation, creating tools that can enhance our daily lives.

Unlike Google’s DeepMind, which keeps its AI developments close to the chest, LeCun means to democratize AI development with projects like the LLaMA model. So, the sci-fi fans among us can rest assured that Meta isn’t building Skynet any time soon (hopefully).