The current rhetoric around AI is overly dramatic, says the Godmother of AI.
“I like to say I’m the most boring speaker in AI these days, precisely because my disappointment is the exaggeration of both sides,” Fei-Fei Li said in a talk at Stanford University published Thursday.
“All this talk about us being on the verge of total extinction or doomsday and that AI is going to destroy humanity, that it’s the machine overlords,” she said. On the other hand, she said, there is also a “perfect utopia” scenario, where people use terms like “post-scarcity” and “unlimited productivity.”
Li is a longtime computer science professor at Stanford University, famous for inventing ImageNet. Last year she Co-founded World Labis a company that builds AI models to perceive, generate, and manipulate 3D environments.
Speaking at Stanford, she added that this “extreme rhetoric” is filling discussions about technology and misinforming vulnerable populations.
“People around the world, especially people who are not in Silicon Valley, need to hear the facts, and they need to hear this is what’s real,” she said. “But that kind of discourse, that kind of communication, that kind of public education is not as good as I would expect.”
Lee is one of the top computer scientists advocating for more balanced messaging about AI and its impact on society.
July, Founder of Google Brain Andrew Ng He said he thinks artificial general intelligence is overrated.
AGI refers to the stage at which AI systems have human-level cognitive abilities and can learn and apply knowledge in the same way as humans. Executives at leading AI labs are often asked when they think AGI will arrive and what it means for human workers.
“AGI is overhyped,” Ng said in a talk at Y Combinator. “For a long time, there will be many things that humans can do that AI cannot do.”
Former Chief AI Scientist at Meta Inc. Yan LeCunthe large language model is “Amazing,” but limited.
“They are not the path to what people call AGI,” he said in an interview last year. “I hate these words. There’s no doubt they’re useful, but they’re not the path to human-level intelligence.”
Last month, LeCun announced on LinkedIn: leave meta I launched my first AI startup in 12 years.
