On Thursday (January 23), the CEO of mankind may accelerate the progress of artificial intelligence (AI), especially in biology, and to double the life of humans in 5 to 10 years. He said. 。 “
between A panel In the Davos’ World Economic Forum, Amday called this a “magnificent vision.” If he believes that today’s AI can reduce the first century work in biology to 5 to 10 years and doubles the average length of humans, “doubling the life of humans. If everything is crazy, you can accelerate that you can get it in 5 to 10 years. “
Amodei is also working on a “virtual collaborator”, and is working on AI agents that can run high -level tasks at work, such as opening a Google document, using Slack messaging channels, and dialogue with workers. I said. Managers only check in “occasionally” in this AI agent, as well as management as a human employee.
Amodei said in the past three to six months, “I am more confident.” Really Almost all tasks are heading to AI systems that are better than almost all humans. He has seen some AI models are starting to be as good as the doctorate. Mathematics students, programming and Biology.
Amodei believes that by 2026 or 2027, “there is a wide and excellent AI system. largely Almost all humans. Humanity was established by a former Openai engineer and was mainly supported by Amazon and Google。 Humanity is differentiated in focusing with his chief rival, open rival. Constitution AI I’m trying to make it AI Kind and harmless.
Red tape, a physical world, hinders AI
But when all these smart and competent AI agents came, Amodei asked, “Do we immediately solve the world’s problems?” The tacit answer is no. He said that two major hurdles of AI proceeded rapidly. teeth Physical world and bureaucrats from Human organization.
Another example given by Amday was in the area of pharmaceutical development. AI can think of drug solutions, but you need to take clinical trials and get the approval of regulatory authorities.
Albert Burla of Pfizer CEO, in The same panel stated that AI has reduced the drug discovery timeline from many years to months to several months. but Several steps are required before these progress directly brings profits to ordinary people. For example, AI helps to identify promising targets, but there is a subsequent process, such as the need to combine antibodies with specific proteins commonly found in cancer cells, and are armed with “warheads” and cancer. Destroy only cells. A healthy thing.
Uber CEO’s DARA KHOSROWSHAHI acknowledged it In the case of an autonomous car The physical world remains a problem that needs to be solved. For example, dirty stop signs can confuse autonomous driving cars. However, KHOSROWSHAHI remains bullish for autonomy. He regards AI -driven vehicles as the future of transportation. He believes that all vehicles on the road will adjust each other someday and reduce congestion on the street.
KHOSROWSHAHI also pointed out another challenge to autonomy, that is, lack of trust. He said that it would be easier for people to accept mistakes from other humans, but AI needs to be 10 times better than humans to be trusted. Therefore, he believes that automatic driving cars are safer than human -led cars, but even a handful of accidents Erosion Public Trust.
Regarding the bureaucratic deficit, Mark Latte, Secretary -General of NATO, acknowledged that his institution could be “too late”. In many cases, NATO wants to work almost completely before the technology solution develops. But Latte said that was a mistake: “Speed is essential.”
Last CEO with all human labor
Salesforce CEO’s Marc BenioFF, which was on the same panel as Amodei, could have the last of the Davos CEO harvest to manage all human workers because the AI agent came to work. Was observed. Immediately, managers cooperate with humans and digital workers. Salesforce itself is provided Agent forceThe client can build and customize the AI agent.
Google’s president and Cio Ruth Porat, Who was Another panelist said that Google is already ahead of quantum computing. Google last December announcement Willow, quantum chips that can perform standard benchmark calculations in less than 5 minutes that He said that it took for years to resolve one of today’s fastest supercomputers, or to solve 24 zero after 10 years. “We are already investing in the next location,” Polat said.
Looking ahead, Amodei said he was worried about the impact of AI’s political system. Will it strengthen more democracy or more dictators? “This is one of my deepest worries,” confessed. “Is AI stable for dictatorship and is unstable for democracy?”
He wondered what would happen to China Development 10 million Super Smart AI agent. “What can they do in the monitoring state? The power of the dictatorship has been restricted by the need for a person who performs the dictator’s will, how bad the dictatorship is. It is restricted.
“There is a terrible possibility that AI can remove some of these restrictions and enable darkness on the” 1984 “(novel about dystopia society) or international stage,” Ambity warns. 。 “This is very serious and I don’t understand It will go Go to the right. “
Google’s Porat said that was the reason It is important to stay on the west side Beyond AI development. This requires a promotional environment for AI. Currently, the United States is ahead of the AI model, Chip Progress of about one year. But this can change. “We need to be bold and responsible, but we need to be sure I have the ability to do Maintain the lead. “
PORAT also stated that US companies must be involved in foreign countries in AI development. She said that when talking to the head of the state, they all expressed their interest in enabling AI in their country. “But they are also very clear that if we are not there, we will be a partner elsewhere.”
