Recent AI’s development did not follow the neat paths imagined by much science fiction: transformers unexpectedly enabled a fast, sprawling rollout of chatbots to billions of users. The more useful question, the writer argues, is not whether fiction forecast the future precisely, but what it reveals about living with a given technology. To compare stories with actual AI, the writer scores them across how AI is created, what it can do, and how it is deployed, including learning methods, human feedback, uneven abilities, interfaces, economics, and social effects.
Ted Chiang’s The Lifecycle of Software Objects ranks first overall because it captures the prolonged work of teaching and maintaining AI, along with the business and platform problems that shape its use. Other works anticipate particular aspects: Galatea 2.2 treats language as something a neural network can learn through experiments; The Moon Is a Harsh Mistress depicts a capable but inconsistent system that learns from human judgments; and “50 Things Every AI Working with Humans Should Know” explores AI-generated writing and disputes over authorship. Her and “The Perfect Match” offer strong visions of consumer assistants, while Avogadro Corp imagines software acting through company systems. Across the works examined, none clearly predicts transformer attention or the rapid progression from chatbots to agents making novel mathematical discoveries.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.