The CEO of Google Deep Mind, Demis Hassabis, believes that the creation of general artificial intelligence (AGI) capable of performing tasks at the human level or surpassing it could become a reality within 5-10 years. About this informs CNBC.
“Today’s systems are quite passive, and there’s still a lot they can’t do. But over the next decade, many of these capabilities will begin to emerge, and we’ll gradually move toward general artificial intelligence.” Hassabis explained.
The main challenge in AGI development, according to Hassabis, is teaching the AI to understand real-world context. Although modern systems can successfully analyze tasks and perform them in game environments, translating these capabilities to real life remains a challenge.
He emphasized that an important stage will be the adaptation of planning algorithms, agent behavior and decision-making to real conditions, using the so-called “models of the world” — systems that understand and interact with the environment.
Hassabis is not the only one who believes that AGI is still a distant prospect. For example, the CEO of China’s Baidu, Robin Li, said last year that AGI would not be available until 10 years from now, skeptical of overly optimistic predictions.
