Abstract:
"The real question is: If you think you should slow down, don't run at high speed, and don't wait until external forces force you to stop." This is what Microsoft Vice Chairman and President Brad Smith said to Alan Murray during the United Nations General Assembly. The topic he discussed was: Should cutting-edge large model manufacturers step on the deceleration pedal to ensure that the technology they develop will not bring catastrophe to mankind.
On the same day that he made these remarks, Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman spoke at the United Nations, calling on countries to establish AI safety protection mechanisms to prevent the development of AI beyond human control capabilities.
Smith, who has many years of industry experience, said: "If you, as the CEO of a company, judge that your product has safety hazards, should you continue to market it? I think the answer is obvious - no. Companies have a responsibility for public safety."
Even so, he added: "The introduction of certain regulatory measures will eventually be inevitable." Historically, many products relied on regulation to achieve safe implementation. For example, there used to be problems with milk contamination, "so we set up an agricultural inspection department and introduced relevant laws. Now when you go to the supermarket, all types of milk are safe and drinkable."
When Smith proposed that there should be the ability to "slow down or shut down the AI system directly," the host Allen confirmed that he was talking about the AI emergency shutdown switch (kill switch), and then asked: "Do you think this is a good idea?"
Smith replied: "Of course. ... The model developer should have the authority to shut it down; cloud infrastructure providers like Microsoft should have that authority; and the agency responsible for controlling the AI agent should also have the ability to shut it down. Even if this mechanism is mandated through legislation, that would not be a bad thing."
Alan asked him how business managers should handle AI security issues within their companies. "My point is, don't stop using AI. AI is a fundamental innovation and an important tool for improving productivity. But it's important to remember that technology requires governance. ... The governance of this technology is as important as the technology itself."
He also reminded: "I don't wake up every day and feel that there is a 10% probability of human extinction. But I would rather we worry too much than do too little. After all, what we are discussing is about human survival. For God's sake, this matter is more important than the latest technology, and we must take it seriously."
Comments