Microsoft's AI Chief Warns of the Difficulty of Containing Superintelligence
Mustafa Suleyman, Microsoft's chief artificial intelligence officer, issued a stark warning amid the pursuit of superintelligence. Suleyman believes that once AI reaches the stage of artificial general intelligence (AGI), it may become difficult for humans to control. His comments come as companies like OpenAI, Google DeepMind, and xAI work toward achieving superintelligence.
In an episode of the "Silicon Valley Girl" podcast, Suleyman shared his thoughts on a future where AI might be smarter than humans, stating that this scenario "doesn't look very positive."
Mustafa Suleyman emphasized that humans could lose control of AI entirely at this stage, with no guarantees in its development that the technology, even after AGI, will remain solely focused on human interests. The Microsoft AI chief added, "It will be very difficult to contain something like that or reconcile it with our values." These comments come as some industry leaders predict the emergence of super-intelligent AI within a few years, fueling both excitement and anxiety in the sector.
Mustafa Suleyman, who co-founded DeepMind before joining Microsoft, described the idea of building super-intelligence as "a pipe dream," something to be avoided rather than pursued. He explained, "These things don't suffer, they don't feel pain; they're just simulations of high-quality conversations."
According to Suleyman, Microsoft is taking a different approach by "trying to build a human-like superintelligence." This, he argues, would ensure that AI remains focused on supporting human interests even as it becomes more intelligent. Suleyman has made similar comments about AI before, and Microsoft's head of AI recently stated that if anyone isn't afraid of AI, they simply don't understand it.