Microsoft AI CEO calls artificial superintelligence an 'anti-goal'

In a thought-provoking discussion, Microsoft’s AI leader, Mustafa Suleyman, is urging caution in the race for artificial superintelligence. Featured on the 'Silicon Valley Girl Podcast,' Suleyman proclaimed that the pursuit of this advanced AI should be deemed an 'anti-goal.' He expressed concerns that superintelligence—AI with reasoning capabilities far exceeding human intellect—does not paint a hopeful picture for the future. Suleyman highlighted the challenges of controlling or aligning such advanced AI with human values, stating, 'It would be very hard to contain something like that or align it to our values.' As the co-founder of DeepMind, Suleyman has transitioned to Microsoft with a vision of developing a 'humanist superintelligence' that prioritizes human interests. He firmly believes that attributing consciousness or moral standing to AI systems is misguided, asserting, 'These things don't suffer. They don't feel pain. They're just simulating high-quality conversation.' His cautionary remarks come amidst a backdrop of enthusiasm from other tech leaders regarding the potential of superintelligence. OpenAI’s CEO, Sam Altman, has frequently referred to artificial general intelligence (AGI)—AI that mimics human reasoning—as a central objective for the company, suggesting that superintelligence could be on the horizon. Earlier this year, Altman indicated that superintelligent tools could revolutionize scientific discovery and enhance global prosperity. Altman, in a September interview, expressed confidence that superintelligence could emerge by 2030. Similarly, Demis Hassabis, co-founder of Google DeepMind, echoed this sentiment, projecting that AGI could be realized within the next five to ten years. He envisions a future where AI systems possess a deep understanding of their surroundings, deeply integrated into daily life. Conversely, some industry experts urge a more measured approach. Meta's chief AI scientist, Yann LeCun, suggested that achieving AGI may still be decades away, emphasizing that complex problems do not always scale effectively. 'You cannot just assume that more data and more compute means smarter AI,' he noted during a talk at the National University of Singapore. This ongoing debate reflects a critical juncture in AI development, balancing ambition with ethical considerations.