Does this possibly demonstrate the limits of usefulness of using LLMs to explore new ideas or predict things?
In this case - the concept of using automated computing devices to manipulate numbers that represent ideas at arbitrary levels of abstraction was, by 1930, nearly an entire century old. Talkie's myopic viewpoint does not represent the most farsighted viewpoint, merely the average. So if, in 1930, you had read the writings of Ada Lovelace, gotten very excited, and wanted to figure out how to pitch it to investors - Talkie might have been very useful.