Demis Hassabis, the 48-year-old British scientist and CEO of Google DeepMind, has been at the forefront of artificial intelligence research. With a background in computer science and neuroscience, Hassabis has been driven by a passion to understand the world and advance human knowledge. His journey into AI began with a fascination for games, becoming a world-champion chess player at 12, and later co-founding DeepMind, which was acquired by Google.
Advancements in AI
Hassabis believes that AI is on an exponential curve of improvement, with significant advancements in recent years. The development of Project Astra, an AI model that can interpret the world through vision and hearing, has shown remarkable capabilities. Astra can identify artworks, describe scenes, and even create stories around them. For instance, when shown a painting by Edward Hopper, Astra described the subject as “pensive and contemplative, her expression suggesting a sense of solitude.”
Towards Artificial General Intelligence
DeepMind is now training its AI model, Gemini, to not only understand the world but also act in it, such as booking tickets and shopping online. This is a step towards achieving artificial general intelligence (AGI), which Hassabis believes will be possible within the next five to ten years. AGI promises to revolutionize various aspects of life, from healthcare to robotics.
Challenges and Risks
While the potential of AI is vast, Hassabis also acknowledges the risks associated with its development. He worries about bad actors misusing AI and the potential for AI systems to become uncontrollable if they are not aligned with human values. Ensuring that AI systems are developed with safety and responsibility in mind is crucial.
The Future of AI Development
Hassabis envisions a future where AI could lead to “radical abundance,” eliminating scarcity and potentially curing diseases. However, he emphasizes the need for international coordination and the establishment of guardrails to ensure that AI development is aligned with human values. The question of whether AI can be taught morality is also being explored, with the understanding that AI learns through demonstration and teaching.
As the world races towards AGI, the implications are profound. Will AI become so advanced that it will be indistinguishable from human intelligence? And what does the future hold for human innovation and existence in a world where machines are increasingly intelligent? Hassabis believes that new philosophers will be needed to understand the implications of AGI and its impact on society.
