According to Business Insider,"AI Godmother" Li Feifei said that the current discussion on AI is too extreme.A lecture by Li Feifei at Stanford University was made public on Thursday. She said in her speech: "I would like to say that I am the most boring speaker in the field of AI today, because the holders of the AI threat theory and AI omnipotence theory are all exaggerating, and this is what makes her disappointed."
Li Feifei
“All we hear are arguments about the complete extinction of humanity and the end of the world, saying that AI will destroy humanity and machines will dominate the world,” she said. “On the other hand, there are also people who hold a ‘completely utopian’ (overly idealistic) view, saying that AI will bring about a ‘post-scarcity era’ (extremely abundant resources) and ‘infinite productivity.’”
Fei-Fei Li is a long-time computer science professor at Stanford University who is best known for creating the ImageNet dataset. Last year, she co-founded World Labs, a company dedicated to developing AI models that can sense, generate and interact with three-dimensional environments.
She said in a lecture at Stanford University,This “extreme rhetoric” is flooding technology discussions and misleading susceptible people.
“People all over the world, especially outside of Silicon Valley, need to hear the facts and know what this technology is,” she said. “Yet this type of discussion, this type of communication, this type of public education, has not yet achieved the effect that I would like it to have.”
In addition to Li Feifei, other top computer scientists are also calling for a more balanced promotion of AI and its social impact.
In July this year, Google Brain founder Andrew Ng said that he believed general artificial intelligence (AGI) was overvalued. AGI refers to the fact that an AI system has human-level cognitive abilities and can learn and apply knowledge like humans. Executives at leading AI companies are often asked when they think AGI will arrive and what it will mean for human workers.
"AGI has been overhyped," Ng said in a Y Combinator speech. "There will still be many things that humans can do that AI can't do for a long time."
Yann LeCun, Meta’s former chief AI scientist, once said that although large language models are “amazing”, they have limitations.
"They're not a path to what's called AGI, which I hate that term," he said in an interview last year. "They do work, there's no question about it, but they're not a path to human-level intelligence."
Last month, Yang Likun announced on LinkedIn that he would end his 12-year career at Meta and start an AI startup company.