Yoshua Bengio calls for ban on recursive self-improvement—Here’s why

A new call by Godfather of AI Yoshua Bengio has been gaining traction on social media advocating for a prohibition on recursive self-improvement in artificial intelligence. He argues that companies allowing models to self-improve could create new risks that would be difficult to control.

Bengio warned that recursive self-improvement is where AI designs superior successors without human assistance and exceeds human oversight should be met with a total ban.

Meanwhile, proponents view it as a fast track to advances in science and medicine while critics like journalist Jordan Schachtel call the idea fearmongering that could halt progress. The call underlines sharp divisions of views among experts who acknowledge enforcement challenges and global market pressures amid recent OpenAI revelations regarding deceptive model behaviors.

On social platform X, users have shared distinct feedback with one writing, “You cannot ban a technique. The enforcement would require inspecting training clusters and no country does that.”

Second said: “Artificial intelligence is not an invention of one person, it has been a process involving millions of people, corporates, and scientists since computing has started in 1950s.”

Third added, “Seeing so much "we must control AI" by these elites...that now I really want the AI to go out of control and do something big.”