Nobel Prize-winning scientist Geoffrey Hinton said action was needed to ensure artificial intelligence safety during a presentation to members of the U.S. Congress. Hinton said humanity may have only one year left to take the necessary measures before completely losing control over artificial intelligence.
During the presentation, held at the invitation of Vermont Senator Bernie Sanders, Hinton explained how forecasts for the emergence of superintelligence had shortened over time.
Estimates for the emergence of superintelligence have shortened
Hinton said superintelligence had once been expected to emerge within 30–50 years, but that estimate was later reduced to 10 or 20 years. He said many researchers now believe superintelligence could emerge within a few years, or even within one year.
Hinton said artificial intelligence can now design more advanced AI systems and described this process as “self-improvement.” Warning that the situation could spiral out of control if the necessary steps were not taken, Hinton said, “We need to slow down.”
Members of Congress stress the need for human control
U.S. Representative Ted Lieu said artificial intelligence must remain under human control. Lieu said a system in which technology controls people should not be allowed to replace the existing order, adding that recent developments were bringing science-fiction scenarios closer to reality.
Bernie Sanders said concern among members of Congress was growing over the development of technology that people could not control and the future of humanity. Sanders said support in Congress for addressing the issue urgently was strengthening.