Neil Gershenfeld offers a perspective on the discussion surrounding artificial intelligence (AI). He highlights the extremes in arguments about AI, which he sees as a sign of the evolving nature of the debate. Gershenfeld explains that disruptive technologies often start as exponentials, with gradual growth initially, but eventually reaching a point where the exponential growth becomes evident. This transition from exponential to sigmoid growth is currently happening with AI.
He notes that many contributions to AI progress come from individuals who are often overlooked in the debate. Gershenfeld mentions advances in areas like random matrix theory, convex relaxations, and kernel methods, which are changing our understanding of intelligence.
Gershenfeld suggests that the evaluation of AI has been evolving, with shifting criteria for success. He questions the notion of consciousness as a defining attribute of human uniqueness and points to empirical evidence that self-awareness exists in other species. He emphasizes that we have become symbiotic with thinking machines, relying on them for various cognitive tasks.
He concludes by stating that the question of whether AI is dangerous is prudent, but simplistic yes/no arguments do not capture the complexity of the issue. Gershenfeld highlights the history of technology and the need to focus on the nuanced details rather than making binary judgments.