핵폭탄 만들었을 때처럼...'인간 통제력 상실' 초지능 AI 위기 [지금이뉴스] / YTN

Rapid advances in Artificial Intelligence technology have led to claims of approaching Superintelligence beyond human levels, sparking fundamental concerns over losing human control. According to the IT industry, Meta CEO Mark Zuckerberg recently declared in his annual letter and on Facebook that Superintelligence is within sight. Zuckerberg stated that over the past few months, he has observed AI systems improving themselves. He further noted that after examining numerous AI algorithms across platforms like Facebook and Instagram, they have started to exhibit initial signs of self-improvement. He added that superintelligence will bring forth new safety concerns of a different dimension, thus requiring strict measures for risk mitigation. He also emphasized that while the benefits of superintelligence will be shared globally, its release as open-source should be a very cautious decision. This signifies a major shift for Meta, moving away from the open-source policy it has maintained with the Llama series towards a closed-source strategy.