AI will surpass 90% of people in five years

According to him, the fierce race between the world’s largest technology companies is leading them to neglect the safety of their models in order to outpace competitors as quickly as possible. “They really do not pay attention to the development of safe, controllable models. They are moving too fast, without proper testing of the systems,” he stressed, adding that such irresponsibility could trigger a large-scale crisis that would serve as a sobering shock for the entire industry.

Kai-Fu Lee emphasized that sustainable AI development is impossible without mutual trust between the state and businesses. “The task is the same — for laws not to suffocate business, and for business to be responsible and give hope. Then the government will continue to trust,” he said. He also noted that the use of AI in social media intensifies risks, as advertising-driven models are often focused solely on profit while ignoring user interests, even when it comes to content related to wars or traumatic events.

In his view, mechanisms must be created to hold corporations accountable for the consequences of applying these technologies. Among the possible solutions, he named lawsuits as well as the creation of accountability platforms that would function similarly to financial regulations. “Social networks now evade responsibility while profiting from advertising. But we need stricter oversight, like in the case of credit cards: if a company incurs losses due to fraud, it itself starts preventing it,” he explained.

Reflecting on the future of technology, Kai-Fu Lee noted that different scenarios are possible — from the emergence of super-intelligent models that may fall into the hands of bad actors or be used to control society in an “1984”-like dystopia, to an awakening of the industry and the creation of safe systems that are reliably integrated into economic and social processes. The most optimistic outcome, he said, would be the emergence of AI deeply integrated into human values and habits, allowing it to serve the betterment of society.

When addressing the topic of AGI — a hypothetical type of AI capable of performing any intellectual task at the level of a human — Kai-Fu Lee said he does not share the classical theory, but prefers a more pragmatic definition: the moment when AI will be better than 90% of people at 90% of tasks. “I think this will happen in about five years. AI is advancing faster than we expect,” he concluded.

Earlier, it was reported that Kazakhtelecom Chairman Bagdat Mussin named infrastructure and access to depersonalized data as key to accelerating AI adoption in Kazakhstan.