Well.. aren't we actually already past this? at least by one foot? AI based just on LLMs is said to be already hitting its limits of processing power/model size. While evangelists claimed it will get to AGI/etc state, it seems now obvious that we won't have single universal tool like that any time soon, and instead what we can do is make dedicated tools trained/integrated for certain purposes. Or as the 'glue' between more material functions/features/math/etc, the 'glue' which was traditionally written in code by the programmer, or in high-level flow charts by the executives, and now can be just spec in natural language. So, yeah, from some angles it can look like a paradigm change, but to me, it looks more like a change in the default/entrylevel programming language. More people and/or companies will be able to get their business/algorithms/integrations/etc assemble and customized easier and without need of 'real programmers', and 'everything will change' but at the same time 'nothing will change' yet again.