Part 2/9:
Hassabis discussed the current state of AI research, emphasizing that while tremendous strides have been made in recent years, true AGI—defined as a system exhibiting all human cognitive capabilities—likely remains a few years away. He estimates a timeframe of three to five years before we see substantial advancements that could be classified as AGI.
Presently, AI systems excel in specific tasks, particularly in language processing and certain mathematical problem-solving scenarios. However, they still lack crucial attributes like reasoning, long-term planning, and the ability to invent original hypotheses. Hassabis notes that true AGI must demonstrate consistent and robust behavior across a range of cognitive tasks, not just excel in narrow domains.