How can AI systems explore combinations of ideas that they have rarely, or never, encountered?
In PRISM, we use analogical reasoning to support creative exploration in the long tail of the data distribution: drawing on familiar relationships to make sense of unfamiliar
