Meta Scientist Warns of Overly Optimistic Expectations for AI
Meta's chief AI scientist warned against overoptimism about AI's role in the immediate future. During a presentation Wednesday at CES in Las Vegas, Yann LeCun said making new technology work “is always harder than we think.” Telecom carriers have invested heavily in AI, which they see as a network management tool (see 2411190057).
Sign up for a free preview to unlock the rest of this article
Communications Daily is required reading for senior executives at top telecom corporations, law firms, lobbying organizations, associations and government agencies (including the FCC). Join them today!
AI won’t be able to do everything, LeCun said, citing plumbing as an example. We have the technology to build a robotic plumber, “it’s just that we can’t get them to be smart enough to really handle the physical world,” he said. Cats and dogs have a better understanding of the physical world than any AI system does today, he added. AI won’t have “human-level intelligence” for years, said LeCun, also a professor at New York University.
LeCun said AI regulation should be at the application, not the model, layer. “Regulating R&D, how does that make any sense?” If you have a Wi-Fi router running on a Linux operating system that gets hacked, you don’t sue Linux; you sue the company that built the router, he said.
AI is “everywhere” at CES, Bill Briggs, chief technology officer at Deloitte Consulting, said during a CES panel discussion. “AI is embedded in almost everything that we’ll see at CES,” he said, adding that the physical and digital worlds are coming together in a way not previously seen.
For instance, Briggs said, hardware and not just software matters again in the AI era. Hardware “was not a part of the conversation that I had with clients a lot in the last decade, at least not compared to all the discussion about digital and data and information.” He added, “It matters a lot, beginning now.” CEOs increasingly are focused on hardware investments “to fuel the tech trajectory that we know is coming in the next decade.”
The pace of change in technology has never been faster, and the volume has never been greater, Briggs said. “The real action” is “how do we transfer that potential into the investments we should be making, to not just do the things we’ve always done a little differently?”
For example, “How do we think about partnering differently, how do we think about our investment portfolio … and not just being saddled with all the things we already are doing today?” Cyber trust, security, privacy, regulatory compliance and ethics “can’t be things we think about at the end,” he said: They have to be “baked” into investments.
Q2K
Briggs envisions much more focus on quantum technology and how it can be used to solve problems. But quantum also poses a threat to network security (see 2402200064) because the technology will make current encryption “irrelevant,” he said.
The world just commemorated the 25th anniversary of Y2K and the massive focus on preparing systems for 2000, he noted. Some are calling the current challenge "Q2K," but unlike Y2K there’s not a fixed deadline to prepare for, he said. There was a “flurry of investment” for Y2K, Briggs said: “It’s coming again without the hard deadline. It’s real, and we need to start now.”
Using AI for interacting with customers is, “we think, 100% a massive change” for companies, said Amit Ahuja, Adobe senior vice president-experience cloud products. Most people have already tried generative AI through ChatGPT, he said. “Everyone has experienced this new technology in different ways already.” Executives are asking, “What does that mean for my brand interaction?”
During a keynote address Tuesday night, Delta CEO Ed Bastian discussed how the airline is using AI through Delta Concierge to make travel experiences more enjoyable. “New marvels like AI, the digital revolution and sustainable technology are giving us incredible tools to transform the travel experience,” Bastian said.