How far can we actually see down the technological tree created by a scientific breakthough?
Take AI, for example. By the 1940s and 1950s people like Turing and von Neumann were already thinking seriously about machine intelligence, learning machines, self reproducing automata and computers doing things that previously required human thought.
Obviously they could not predict the actual path. They did not know about transformers, modern GPUs or the internet. But something about the destination was already visible surprisingly far ahead.
Compare that with something like quantum mechanics in the 1920s and 30s. Even if you completely understood the new physics, how much could you realistically have predicted about the world it eventually helped create? Transistors, semiconductor fabrication, lasers, MRI and modern computing seem to require a huge number of intermediate discoveries and complementary technologies that did not yet exist.
So I wonder if different scientific fields have fundamentally different forecast horizons.
One for example, AI forcaster were maybe such good at forcasting was that they already had a pre-destined path- our brains. Flight also had an obvious target because birds already demonstrated that heavier than air flight was possible.
Other discoveries, on the other hand, might open a huge branching search space where the important applications depend on discoveries that cannot themselves be predicted in advance.
Can we say anything systematic about when the consequences of a scientific breakthrough are forecastable?
For example, forecastability seems to be good when there is visible end state, but bad at how many complementary discoveries are required, how deep the dependency tree is, or whether progress has a measurable direction?