According to Demis Hassabis (@demishassabis) on Twitter, Yann LeCun is conflating general intelligence with universal intelligence, emphasizing that both human brains and AI foundation models function as approximate Turing machines capable of learning any computable task given sufficient data, time, and memory (source: https://twitter.com/demishassabis/status/2003097405026193809). Hassabis argues that, while the no free lunch theorem means practical systems require some specialization, the underlying architecture allows for broad generalization. This perspective suggests significant business opportunities in developing AI models with adaptable architectures, enabling them to tackle a wide range of computable problems across industries. For AI companies, investing in scalable, generalist models could lead to breakthroughs in fields demanding flexible intelligence, such as autonomous systems, scientific discovery, and complex decision-making.