Microsoft AI chief warns AI superintelligence could be too powerful to control, says it should be ‘anti-goal’

While many of the top AI leaders have been racing against time to reach AI superintelligence, Microsoft AI CEO Mustafa Suleyman has warned that it might not be a positive thing and even claimed in a recent podcast that it should be an “anti goal”.

In an episode of the “Silicon Valley Girl Podcast”, Suleyman defined superintelligence as an AI system that can self improve, set its own goals and act independently of humans, and said that it does not “feel like a positive vision of the future”.

“It would be very hard to contain something like that or align it to our values. And so that should be the anti goal,” he added.

Suleyman also clarified the differentiation between AGI and superintelligence, saying, “You can think of AGI as a step before maybe super intelligence, but roughly speaking, they’re used fairly interchangeably.”

Suleyman noted that Microsoft is building a humanist version of superintelligence which he said would be “aligned to our interests, on our team, in our corner backing us up”.

Suleyman agrees with Demis Hassabis’ prediction on AGI

Suleyman also agreed with Google DeepMind CEO and his ex co founder Demis Hassabis’ prediction on AGI being reached in the next five years. The Microsoft AI CEO noted that AI would be able to achieve human level performance in most tasks in that time.

He argued that the current AI models are already capable of achieving many tasks better than humans like summarisation, translation, transcription, research, document writing and poetry. Suleyman further said that the AI models are “taking steps towards being as good as a human at being a project manager or being a marketing person or an HR person”.

He also predicted that this change will “fundamentally change work in the most profound way”.

“It’s going to change the type of work that we do,” he added.

Suleyman also argued that the AI models are leading to “democratising access to intelligence” and that would lead to an “unbelievable amount of competition because the distance between an idea and the realisation of that idea is going to collapse”.

“People are just going to be thinking new companies into existence, new products, new pieces of poetry,” he noted.

Suleyman also noted that there will be a need to figure out regulations and guardrails to make sure that these autonomous agents work with humans and not on their own.