Maybe the behaviour of LLMs can shed some light into this discussion. Does a LLM have free will? Obviously not because it is just a software running on a computer. But even when you ask an LLM exactly the same questions you will get totally different results. Sometimes it makes up things and sometimes you can push it to state that 1+1=3. It depends on the surroundings but also on some "magical free" will of the LLM to answer differently each time. Of course it is because of the random seeds that are needed to start the inference. Even though the human brain is not a LLM, we learn that there are systems that behave like having "free will".

just my 2 ct