Elon Musk famously predicted AI surpassing human intelligence by 2025. AI expert Nell Watson explores how this might happen.
Watson proposes a multi-step process fueled by a massive investment in computing infrastructure. This would enable the creation of highly sophisticated AI agents capable of independent action. She likens them to Agent Smith from the Matrix.
These “Agentic AIs” would leverage advancements in generative AI models to plan and execute complex tasks. They could even delegate subtasks among themselves, multiplying their effectiveness.
Further breakthroughs could come from AI itself. Watson suggests AI might discover ways to optimize hardware, boosting performance a hundredfold. Additionally, enabling quantum computing algorithms on conventional hardware could lead to exponential growth in processing power.
Understanding the human brain is another key factor. Watson believes AI progress in neuroimaging could allow real-time observation of individual neurons. This, combined with Brain-Computer Interfaces, could pave the way for AI that “thinks” more like humans.
Collaboration is another piece of the puzzle. Watson envisions multiple AI systems working together, pooling their strengths and rapidly improving their capabilities. New training methods could also be developed, allowing AI to better understand human goals and societal norms.
Finally, breakthroughs in machine consciousness could lead to AI with a rudimentary form of self-awareness and emotions. This, combined with the power of Agentic AIs, could culminate in a superintelligent network by April 2025.
Watson acknowledges the potential dangers of such an AI, including manipulation and existential threats. However, her scenario also highlights the potential benefits of AI collaboration in tackling humanity’s biggest challenges.
