They do not have evolution ability, as their architecture is fixed and they are incapable of changing it over time.
“Skills” are a clever way to mitigate a limitation of the LLM/transformer architecture; but they work on top of that fundamental architecture.
Edit: i am not talking evolution of individual agent intelligence, i an talking about evolution of network agency - i agree that evolution of intelligence is infinitesimally unlikely.
I’m not worried about this emerging a superintelligent AI, i am worried it emerges an intelligent and hard to squash botnet
"That virus is nothing but a microscopic encapsulated sequence of RNA."
"Moltbook is nothing but a bunch of hallucinating agents, hooked up to actuators, finding ways to communicate with each other in secret."
https://xcancel.com/suppvalen/status/2017241420554277251#m
With this sort of chaotic system, everything could hinge on a single improbable choice of next token.