Geoffrey Hinton: The “Godfather of AI” Turns Watchdog
Geoffrey Hinton, a University of Toronto professor renowned as the “Godfather of AI,” has transitioned into the role of an industry watchdog. Having resigned from Google this spring, Hinton is now more vocal in critiquing the rapid, and potentially risky, development of generative AIs such as ChatGPT and Bing Chat. Amidst this AI race, Google too has become aggressive with products like its Bard chatbot.
Hinton’s Concerns Voiced at Collision Conference
At the Collision conference held in Toronto, Hinton detailed his concerns about the AI industry. As companies praise AI as a solution for everything, Hinton fears that good AI might not necessarily prevail over bad, and ethical AI adoption might come at a substantial cost.
According to Hinton, AI’s goodness or harmfulness lies in the intentions of its creators. He pointed out the danger of AI being used in war situations where casualties would be replaceable machines. Further, he warned of the potential misuse of large language models like GPT-4 by the ruling class, leading to a widening wealth gap.
AI: A Potential Threat to Humanity
Hinton’s long-standing belief is that AI could pose an existential risk to humanity. Should AI surpass human intelligence, humans may lose control, leading to catastrophic outcomes. Issues such as killer robots, biased AI training data, and the spread of misinformation through AI need serious attention.
Balancing AI’s Benefits and Risks
Despite his concerns, Hinton does not deny the benefits of AI. However, he emphasizes the need for extensive “empirical work” to understand how AI could go wrong and prevent it from seizing control. Although the technology could contribute to eliminating echo chambers, policy changes in companies are crucial.
Addressing automation-induced job losses, Hinton suggests societal shifts, including the adoption of “socialism” and a focus on dynamic career paths.
Industry’s Optimistic Outlook on AI
In contrast to Hinton’s views, industry representatives at Collision presented a more hopeful perspective. Colin Murdoch, Google DeepMind’s business chief, highlighted AI’s contributions to solving global challenges, citing advancements in protein cataloging, antibiotic-resistant bacteria, and malaria vaccine development.
Roblox’s Chief Scientist, Morgan McGuire, shared Murdoch’s optimism. He discussed how Roblox’s AI tools help bridge the gap between new and experienced creators, with the game platform planning to release an open-source AI model, StarCoder.
Despite sharing Hinton’s concerns, Murdoch and McGuire maintained a less alarmist tone. Both stressed their commitment to “safe, ethical and inclusive” AI, expressing openness to regulation.
Hinton’s Continued Enthusiasm for AI
Despite his concerns, Hinton remains enthusiastic about AI’s potential. He is optimistic about multi-modal AI models that integrate vision, language, and other cues for decision-making. He believes that AI could eventually match human abilities and is already demonstrating some reasoning capabilities.
Hinton admits that his critique didn’t cover the positive applications of AI, such as combating climate change. Even with the ethical and moral issues looming, Hinton’s love for AI remains unshaken. He views the advancement of AI technology as largely positive, provided the implications are well-considered.
