How to Develop Systems That We Can Have Confidence In

Geiger: I think there are different ways of viewing what we're asking for. The European Union is close to passing the AI Act, which would require these kinds of steps as a matter of law. Could we develop ways of testing the systems that we already have so that we can show that they do not present risk? Geiger: Or we change the way we design the system so that we could do that.