Abstract:
On September 23, according to Bloomberg, AI pioneer Li Feifei called for more supervision of AI by independent agencies and the public sector. She believes that the security assessment of increasingly powerful AI systems should not be entirely left to the companies developing these systems.
Li Feifei
As a professor at Stanford University and co-founder of World Labs, Li Feifei said on Tuesday that although internal researchers at AI companies can evaluate the capabilities and safety of individual models, this cannot replace broader standards jointly developed by government, industry and academic institutions.
“I still believe that it is very important for benchmarks to be set by independent institutions as well as public sectors such as academia. At the same time, industry and government should also share the responsibility.” Li Feifei said in an interview with Bloomberg TV.
As the capabilities of AI systems continue to increase, so do concerns about their potential risks. Some leaders in Silicon Valley have warned that the technology is developing too fast and may eventually exceed human control. In recent weeks, Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman have both called for slowing down the development of cutting-edge models. The two will discuss AI safety at the United Nations Security Council on Wednesday.
Li Feifei said that the existential risks that AI may bring ultimately depend on how humans choose to develop, govern and use this technology. She said that this principle also applies to technologies other than AI, from electricity and steam engines to modern transportation. Human society also needs to find a balance between the benefits that technology may bring and personal safety.
Humanity, she noted, "has the power of technology at its disposal." "We must take control. AI is not about AI, but about humans. It is about our common and individual dignity and autonomy, and it is also about our responsibilities." Li Feifei said.
Li Feifei’s comments add a new voice to the increasingly heated debate among policymakers and AI industry leaders. The debate centers on how seriously warnings about the technology's most extreme risks should be taken. Earlier on Tuesday, U.S. President Donald Trump dismissed concerns about the existential threat of AI, calling it "a hoax" and stressing the need to continue developing the technology at full speed during a speech at the United Nations General Assembly in New York.
Li Feifei said that the discussion about the behavior of AI systems has gone beyond the so-called cutting-edge models driving the current AI craze. The company she founded, World Labs, is developing so-called "world models" that can generate and reconstruct three-dimensional environments. She said the technology could be used to train and evaluate robots and enable applications in design, medical and other industries.
As these systems move towards real-world applications, she said, the industry faces the same fundamental question: how to evaluate increasingly powerful AI and determine whether it is operating as expected.
Comments