DeepMind co-founder, Demis Hassabis, has refuted criticisms made by Meta's chief AI scientist, Yann LeCun, who accused him and other industry leaders of exaggerating the potential dangers of AI technology. Hassabis argues for a responsible and foresighted approach to AI development, highlighting the need for international agreements on long-term risks.
DeepMind's Hassabis disputes Meta's AI risk claims
Demis Hassabis, co-founder of DeepMind, has fought back against criticisms made by Meta's leading AI scientist. The Meta representative accused Hassabis and others in the AI industry of overstating the existential threats posed by AI, a claim that Hassabis brushed off as 'preposterous'. The war of words between these two tech titans underscores the ongoing debate about the potential risks of AI technology, and how much emphasis should be placed on them.
Hassabis's response was prompted by a post from Meta’s Yann LeCun earlier in the week. LeCun criticized some industry leaders for unnecessarily hyping up doomsday scenarios linked to AI. According to LeCun, this kind of rhetoric only provides fuel to those who want to ban open-source AI research and development. He warns that such a move could be catastrophic, potentially leaving the future of AI in the hands of a small group of companies.
AI control concentration: A potential disaster
LeCun's argument hinges on the belief that overly dramatizing AI risks could lead to restrictive measures on AI research and development. This, he fears, would result in the concentration of control over future AI systems in a small number of companies - a scenario he labels as a potential 'catastrophe'. His viewpoint raises important questions about the balance between open-source innovation and the control of AI technology.
The UK's AI Safety Summit has served as a platform for leaders from the US, Asia, and Europe to unite and discuss protection measures against risks from the most advanced 'frontier' AI systems. This collective effort showcases a global commitment to addressing AI safety and maintaining a proactive stance on technology's frontier.
Hassabis calls for foresight on long-term AI risks
Hassabis emphasizes that addressing near-term issues, like watermarking generated images and audio to prevent deepfakes, is just the tip of the iceberg. He calls for an international agreement on the longer-term risks of AI technology. While these may still be a decade or more away, Hassabis stresses the importance of being well-prepared long before these potential risks materialize.