By
Godfrey Benjamin
April 26th, 2024
Meta’s AI chief, Yann LeCun said it will take decades to achieve the adoption of AGI.
Meta’s AI chief, Yann LeCun, has expressed skepticism about current AI systems surpassing human intelligence anytime soon. Mr. LeCun shared this sentiment while speaking at a Meta media event to celebrate the 10th anniversary of its Fundamental AI Research team.
LeCun: AGI Systems Not Possible for Decades
Last week, Reuters reported that OpenAI cofounder Sam Altman was removed as CEO when a team of OpenAI researchers discovered a project called Q*. Q* focused on Artificial General Intelligence (AGI) defined as autonomous systems that surpass humans in most economically valuable tasks. The project was said to be at an advanced stage and nearing breakthrough levels without proper safety precautions.
Since the report, there have been varying opinions on the subject of AGI and the safety of advanced AI systems for humans. While Altman confirmed the existence of such a project, he merely said that he expects progress on the technology to be rapid. However, Mr LeCun has said it will take decades to achieve AGI. “Society is more likely to get cat-level or dog-level AI years before human-level AI,” LeCun said.
Like Mr LeCun, Microsoft president Brad Smith shares similar sentiments. According to Smith, AGI systems will not be available in the next 12 months, cautioning it could take decades.
Interestingly, LeCun and Smith’s comment stands in contrast to that of Nvidia CEO Jensen Huang. Huang recently asserted that AI will beat humans at multiple mentally intensive tasks in less than five years. To this, Mr. LeCun stated that Nvidia has much to gain from such an AI craze so the prediction is not surprising.
Skepticism Also Trails Quantum Computing Pursuits
Meanwhile, LeCun is also skeptical about quantum computing. While many researchers believe quantum computers could pave the way for innovation in data-intensive fields like drug discovery, LeCun is not convinced.
Consequently, unlike Microsoft, IBM, and Google, Meta has refused to commit resources to quantum computing. According to LeCun, classical computers can also solve the problems one will need a quantum computer for efficiently.
While acknowledging it as an interesting topic, LeCun noted the subject is unclear about the “practical relevance and the possibility of fabricating quantum computers that are actually useful.”
Former tech chief at Meta, Mike Schroepfer concurs with LeCun on this. Schroepfer noted he periodically evaluates quantum technology and believes that useful quantum machines will eventually be developed. However, he is unsure how soon that will be.
“It’s got such a long time horizon that it’s irrelevant to what we’re doing,” he concluded.