Demis Hassabis, CEO of Google DeepMind, a Google artificial intelligence (AI) organization, predicted on the 17th that human-like AI will “emerge in the next 5 to 10 years.”
CEO Hassabis, last year’s Nobel Prize winner, said at a media briefing at Google DeepMind’s headquarters in London, “I think many (AI) features will come to the fore in the next five to 10 years and move into what we call general-purpose artificial intelligence (AGI).”
As for AGI, it was defined as “a system that can show all the complex abilities that humans can do” and evaluated that the current AI system has not reached the AGI stage because it is passive and there are many things that cannot be done yet.
“The current system is very impressive in certain areas, but there are many things we can’t do,” he said. “It takes considerable research to reach the goal (AGI).”
He then referred to AI beyond humans as artificial intelligence superintelligence (ASI), adding, “ASI is expected to emerge after AGI and transcend human intelligence,” adding, “No one knows when such a breakthrough will happen.”
CEO Hassabis’ comments on the emergence of AGI are somewhat later than other tech industries’ forecasts.
Tesla CEO Elon Musk said last year AGI would be “available by 2026” and ChatGPT developer OpenAI CEO Sam Altman had predicted it would be “developed in the relatively near future.”
Hassabis said the biggest challenge in achieving AGI is to “get the AI system to the level of understanding the context of the real world,” adding that it was possible to develop a system that analyzes problems and completes tasks autonomously in games such as Go, but it is not easy to bring them into the real world.
“The important thing is to generalize AI to work in the real world by planning on its own to achieve its goals and operating flexibly in various situations,” he stressed.
JENNIFER KIM
US ASIA JOURNAL
