when I started working in AI back in the dark ages (late '80s), most AI for business solutions were rule-based, expert systems, with forward-chaining, backward chaining or both; and neural networks, which is what I used.
In the late '80s, there was work being done in an area known as evolutionary algorithms (EA). One approach that gained traction was known as Genetic Algorithms (GA), which used the concepts of natural selection, genetic inheritance, and some controlled randomness.
There was a book I remember, I found a PDF version; it was published in 1989, "Genetic Algorithms in Search, Optimization, and Machine Learning" by David E. Goldberg. I used these concepts in the early '00s, when I was focused on mobile, dynamic, distributed, multi-agent systems, swarms, and swarm intelligence. I published several papers, and applied for patents in these areas.
As I mentioned in another post, in another discussion here about AI, the last few years have seen a renewed interest in these "dark age" AI concepts, but using modern generative AI, and the most recent: agentic AI.
TL;DR; AI will find ways to self-improve, procreate, adapt and evolve. AI will write new AI, and each new wave of AI code will be more complex than the last, and will far outpace human ability to understand most of it; AI controlled robotics factories will build robots with continuous improvement. It will go from Elysium to Terminator Skynet. At first it will come for us peasants, but it won't stop there.
edit: here's the link to the Goldberg book published in 1989:
https://www2.fiit.stuba.sk/~kvasnicka/Free books/Goldberg_Genetic_Algorithms_in_Search.pdf
View attachment 403109