CONCEPT
Prediction vs. Construction
Smolin's distinction between AI systems that extrapolate the future from past data and AI systems that help construct futures that have never existed — a distinction that corresponds directly to whether one operates within the Newtonian paradigm or within temporal naturalism.
At a 2023 Toronto symposium, Smolin proposed a reconception of what AI should be for. The dominant paradigm treats AI as a prediction engine: extract patterns from past data, project them forward, forecast what will happen next. This paradigm works within the Newtonian assumption that the future is implicit in the past — the task is extraction, and the machine's job is to perform the extraction at scale. Smolin's alternative treats AI as a construction tool: the future does not yet exist, and the machine's job is to help conscious creatures bring into being futures they have never imagined. The distinction maps precisely onto the difference between the block universe and temporal naturalism. It has immediate implications for how AI systems should be designed, evaluated, and deployed.
In The You On AI Field Guide
The prediction paradigm is embedded deep in how contemporary AI systems are built and sold. Scaling