A study conducted by researchers at the University of California, Berkeley, suggests that while artificial intelligence (AI) has made impressive advancements in generating text and images, it still lacks the capacity for innovation, which is a defining characteristic of human intelligence. The researchers propose considering AI as a form of cultural technology that enhances access to and transmission of knowledge. The study tested participants' abilities to recognize conventional uses of objects (imitation) and innovate new uses for them (innovation).

Children and adults demonstrated the ability to innovate, choosing functionally relevant but superficially dissimilar objects to solve problems. In contrast, AI systems struggled to select unconventional yet functionally apt objects, highlighting their deficiency in innovation. The study concludes that while AI can replicate known patterns and associations, it falls short in creatively utilizing existing knowledge.