* Humans are not deterministic.
* Humans that make mistakes are still considered to be reasoning.
* Deterministic algorithms have limitations, like Goedel incompleteness, which humans seem able to overcome, so presumably, we expect reasoning to also be able to overcome such challenges.