Let the Machines In

The author, who runs an AI company, reflects on the paradoxical nature of using statistical prediction machines (LLMs) to shape our future when prediction itself is notoriously difficult. She expresses surprise that large-scale neural networks trained on data have proven capable of capturing much of what makes human intelligence and creativity work, challenging long-held assumptions about what artificial intelligence would require. This unexpected breakthrough has sparked both market enthusiasm and fundamental philosophical questions about the nature of intelligence itself.