(Yicai) July 28 -- In the rapidly advancing artificial intelligence era, where clear regulations are still lacking, it is essential to establish technology boundaries through international cooperation, according to a former chief executive of US tech giant Google.

The parties involved need to reach a clear consensus on some key issues via discussion, Eric Schmidt, who is now executive chairman and CEO of US three-dimensional-printed rocket maker Relativity Space, said while speaking with Microsoft's former executive vice president Harry Shum at the three-day 2025 World AI Conference in Shanghai, which ends today.

Whether AI can control weapons, replicate itself, engage in autonomous learning, and other such matters must be clarified, Schmidt said, noting that there is currently no method to ensure that AI "plays by the rules."

A study from the United States indicated that AI models can still lie and cheat with just a bit of guidance, Schmidt pointed out, and proposed an ideal scenario, where AI is designed in such a way that it "never learns to do bad things."

Schmidt said that the competition with tech giants like Microsoft and Apple was intense during his time at Google, but this ultimately drove overall progress in the industry.

Using the views of former US Secretary of State Henry Kissinger to illustrate the feasibility of AI technology cooperation between China and the US, Schmidt said that "As long as both sides have a common goal, an agreement can be reached."

The breakthrough in establishing diplomatic relations between China and the US started from scratch, and as long as there is a consensus, mutual trust can be built, Schmidt said, adding that an AI technology alliance between the pair can also be similarly promoted.

Editor: Martin Kadiev