There is a difference between a problem being complex and you try to find a solution to it (hard), vs a program being complex. The latter is trivial to execute, but that is entirely different from analysing it.
LLMs can (more often) successfully find solutions for far more complex problems than animals can. So where does that leave us?