Microsoft AI CEO Mustafa Suleyman recently challenged OpenAI CEO Sam Altman’s assertion during a Reddit AMA that artificial general intelligence (AGI) could be achieved with today’s hardware. In an interview with The Verge’s Nilay Patel on the Decoder podcast, Suleyman expressed skepticism, suggesting AGI might take up to 10 years to materialize, even with advancements in computing technology.
Suleyman clarified his stance when Patel defined current hardware as being within one or two generations of existing systems. Suleyman stated, “I don’t think it can be done on [Nvidia] GB200s.” Instead, he believes AGI may become plausible within two to five hardware generations, emphasizing that each generation typically spans 18 to 24 months. Consequently, achieving AGI could be a five-to-ten-year journey, depending on the pace of technological development. He added, “The uncertainty around this is so high that any categorical declarations just feel sort of ungrounded to me and over the top.”
Suleyman also drew a distinction between AGI and the “singularity.” He described AGI as a general-purpose learning system capable of performing well across a broad spectrum of human-level training environments, including knowledge work and physical labor. “To me, AGI is a general-purpose learning system that can perform well across all human-level training environments,” he explained, contrasting this with the singularity, which he defined as “an exponentially recursive self-improving system” that rapidly surpasses human intelligence. Suleyman expressed doubts about the timeline for AGI due to the complexity of integrating robotics into such systems but acknowledged the possibility of AI systems mastering diverse tasks with minimal human intervention within the next five to ten years.
However, he cautioned against conflating AGI with theoretical superintelligence, noting that these are fundamentally different concepts. “I think the reason why I shy away from the language around singularity or artificial superintelligence is because they’re very different things,” he said. His primary focus is on creating AI companions that are practical, accountable, and aligned with human needs rather than pursuing a speculative quest for superintelligence.
Suleyman also criticized the dramatization of AGI in public discourse, arguing that it detracts from understanding the specific capabilities of emerging AI systems. His goal is to develop AI tools that function as reliable assistants for humans, enhancing productivity and working collaboratively with users. “That’s what I care about with respect to building AI companions,” he said. “Getting them to be useful to you as a human, work for you as a human, be on your side, in your corner, and on your team.”
These remarks come as Altman recently adjusted his expectations for AGI. During The New York Times DealBook Summit, Altman downplayed AGI’s immediate impact, claiming it will arrive “sooner than most people in the world think and it will matter much less.” He also suggested that the transition from AGI to superintelligence would be gradual, stating, “AGI can get built, the world mostly goes on in mostly the same way, things grow faster, but then there is a long continuation from what we call AGI to what we call superintelligence.”
This exchange highlights an apparent divergence in perspectives between Suleyman and Altman, just a year after Microsoft played a pivotal role in reinstating Altman as OpenAI’s CEO. Despite the differences, Suleyman downplayed any friction between Microsoft and OpenAI, acknowledging the inherent tensions in partnerships. “Every partnership has tension. It’s healthy and natural,” he remarked. He also noted that OpenAI operates independently, and partnerships must adapt over time. “We’ll see how that changes over the next few years,” he added.
Suleyman further confirmed that Microsoft is developing its own frontier AI model capable of competing at the scale of OpenAI’s GPT-4. This announcement reflects Microsoft’s strategic commitment to advancing its AI initiatives and underscores the competitive dynamics in the AI industry. As these developments unfold, the evolving relationship between Microsoft and OpenAI will likely play a significant role in shaping the trajectory of AI research and innovation.
In summary, while Suleyman and Altman share an overarching vision of AGI’s potential, their differing views on timelines, capabilities, and priorities underscore the complexities of this rapidly evolving field. Suleyman’s cautious optimism and focus on practical AI applications contrast with Altman’s more ambitious projections, illustrating the diverse approaches within the industry. As AI continues to progress, these discussions will remain central to shaping public understanding and guiding the development of responsible and effective AI technologies.