A Demis Hassabis promise has tech watchers on high alert: DeepMind’s AI will soon master the elusive “AGI”.
AGI, or artificial general intelligence, refers to an AI system that can perform any intellectual task that a human can, according to Demis Hassabis, DeepMind’s CEO. At Google’s I/O developer conference, Hassabis asserted that this milestone is near, but his team won’t declare victory until their AI can surpass human performance in all areas of cognition.
The Competition Heats Up
Hassabis described the AI landscape as “the most ferocious competition that’s ever been in tech history.” With the likes of Microsoft, Meta, Amazon, and other tech giants vying for dominance, the stakes are high. However, DeepMind has a leg up, he believes, thanks to its “broader research bench.” This refers to the vast resources and expertise available to DeepMind, which can draw on Google’s vast research and development capabilities.
The AGI Test
So, what exactly does DeepMind’s AI need to do to earn its AGI stripes? Hassabis mentioned that passing a test similar to the Turing Test would be a key benchmark. This test, devised by Alan Turing in 1950, assesses a machine’s ability to exhibit intelligent behavior equivalent to, or indistinguishable from, that of a human.
However, Hassabis also emphasized that this test is just the beginning. AGI, he said, requires a system that can excel across a wide range of cognitive tasks, from reasoning and problem-solving to creativity and learning. In short, DeepMind’s AI needs to demonstrate a level of intelligence that is not only on par with humans but also capable of improving upon existing human capabilities.
What this means
While the prospect of AGI may seem like science fiction, its development holds significant implications for the future of work, society, and our daily lives. If DeepMind succeeds in building an AGI that surpasses human capabilities, it could lead to breakthroughs in fields like healthcare, education, and environmental sustainability. On the other hand, concerns about job displacement, bias, and accountability will only intensify.
As Hassabis and his team hurtle towards the finish line, one thing is clear: the stakes have never been higher. Will DeepMind’s AI be the first to crack the AGI code? Only time will tell, but one thing is certain: the journey will be fascinating to watch.
