Microsoft’s AI chief, Mustafa Suleyman, believes artificial general intelligence (AGI) is farther away than many industry experts think. He stated that current hardware is not capable of supporting AGI, a view that contradicts the perspective of Sam Altman, who believes it can be achieved with existing technology.
Please follow us on Facebook and Twitter.
In an interview with The Verge, Suleyman emphasized that today’s graphics cards lack the necessary capability to support AGI. The DeepMind co-founder and head of Microsoft’s AI division explained that while AI advancements could progress within five generations, the current hardware limitations make achieving AGI difficult.
“I don’t think it’s possible on GB200s. I believe it could become plausible within the next two to five generations,” Suleyman said, referencing the GPUs currently used by NVIDIA in its data centers. These GPUs are up to 30 times faster than the H100s when performing inference tasks on large language models (LLMs).
According to Mustafa Suleyman, artificial general intelligence could emerge in five to seven years, depending on NVIDIA’s development pace. He noted that each GPU generation takes about 18 to 24 months, meaning AGI might realistically arrive within the next decade.
What Is Artificial General Intelligence and When Will It Be Available?
Artificial General Intelligence (AGI) is a theoretical form of AI capable of performing any intellectual task that a human can do. This means AGI would have the ability to reason, learn, and adapt to new situations just like a person. Unlike today’s AI, which excels in specific tasks, AGI aspires to replicate the entire range of human cognitive abilities.
While the fundamental definition of AGI has remained consistent throughout AI’s history, some recent claims have slightly shifted the narrative. NVIDIA’s CEO has suggested that “super intelligence”—a step beyond AGI—could surpass human intelligence within five years. Meanwhile, Microsoft researchers believe GPT-4 has shown glimpses of human-like intelligence, suggesting it might be an early-stage AGI prototype.
Mustafa Suleyman described artificial general intelligence (AGI) as “a general-purpose learning system that can perform well in all human-level training environments.” According to the Microsoft AI CEO, the concept of AGI has been over-dramatized, which distracts from its realistic potential. Suleyman envisions AI giving rise to companions capable of assisting humans in both physical and cognitive tasks but does not foresee this leading to the feared singularity.
Although definitions of AGI vary, many experts agree it could become a reality within the next decade.