The way to make an artificial general intelligence (AGI) may affect "catastrophic effects", specified as cyber attacks on energy or water infrastructure, warned Demis Hassabis, CEO of Google DeepMind. He suggested that AGI could appear within the next decade.
Speaking at the Axios AI+ summit in San Francisco last week, Hassabis described AGI as a model that shows ‘all cognitive abilities’ man, including invention and creativity.
He argued that the current large language models stay "frayed intelligence’ with gaps in reasoning, long-term planning and continuous learning.
However, he suggested that AGI could shortly become a reality through continuous scaling and "one or 2 more major breakthroughs".
At the same time, Hassabis acknowledged that the pre-AGI period would likely entail crucial risks and ‘catastrophic effects’, such as cyber attacks on energy or water infrastructure.
"It's most likely already happening... possibly not yet in the case of highly advanced artificial intelligence” "The most apparent vector of vulnerability," he said.
He added that attack actors, autonomous agents and systems that "detach" from their intended targets require serious mitigation measures
. “This is not a zero probability” – said about the anticipation that advanced systems can "to jump the protective barrier".
Hassabis' concerns reflect wider warnings in the technology industry.
An open letter published in October, signed by leading technologists and public figures, stated that ‘superintelligent’ systems can endanger human freedom and even survival, calling for a global ban on the improvement of artificial intelligence until it is secure.
Among the signatories of the letter were the co-founder of Apple Steve Wozniak, pioneers of artificial intelligence Geoffrey Hinton and Yoshua Bengio, founder of Virgin Group Richard Branson, and prominent political and cultural personalities.
Others have taken a more optimistic approach. Elon Musk said last period that advancement in artificial intelligence and robotics could make work become ‘optional’ within 10 to 20 years, and has predicted that currency may lose its importance in an economy based on artificial intelligence, while pointing out that considerable technological advancement is inactive needed before specified a future can come true.
Translated by Google Translator
source:https://www.rt.com/news/629166-advanced-ai-catastrophic-outcomes/
