More people should study basic AI.
At its core AI consists of many nodes, the job of which is to sum numbers (typically a number between 0 and 1) from nodes connected to it, perform some function on them and pass on the result (0 to 1) to more nodes.
The network is given inputs (say 64 0-1 values representing pixel brightness of an 8x8 image).
The nodes each petform their task, many layers feeding into other layers, until the final layer (the outputs) produce an answer (say one output, is this image a letter A? 1 for yes 0 for no)
The network is shown an image, there's a flurry of activity, and an answer emerges. If the answer is incorrect, the correct answet is imposed on the output and inputs and each node (using a clever derivative based algorithm) has its function slightly adjusted in the right direction.
A letter B is shown, we expect a Zero (we want an A) so we force back that answer.
This training continues, with the letter A rotated, or made serif etc. Thousands and thousands of times, until, when shown an A and only when shown an A, the network outputs a 1.
We train up 27 such networks, all connected to the same inputs. Each trained to spot a different letter. We show the inputs any letter, One of the outputs is happy.
We can put any inputs into the network. A megapixel image. An audio sample.. And we can chain the results through other networks (deep learning).
The bigger the network, the longer it will.take to 'settle' on an answer.
The brain works exactly the same way. And it is noticed that up to a second before a person 'chooses' between two objects placed in front of them, the brain fires like crazy.
Determinists like Dennett say that this is proof that free will is an illusion. That consciousness is just an executive window on the settling networks of biological AI.
Decisions are actually the result of a seconds worth of node settling, not the transient ephemeral split second thing we percieve them to be.
This is almost certainly.true of lower animals.
But there is a difference (currently) between AI and human intelligence.
The human 'backprop perceptron network' never completely settles to no activity even in complete sensory deprevation. The human brain has a tickover rate like a car at traffic lights.
Humans can resurrect stimulii from long ago. Run them through the network and suppress the outputs. Or we can override processing of immediate stimulii and choose to respond later. Consciousness is infinitely extended cognition. And somewhere in there is 'free will'.
Some suggest that this lies in the point where the cerebellum swaps left into right in the cerebrum and vice versa. in the pre frontal cortex. Who knows.
The point is that free will has something to do with humans a brain that never stops processing, and can use 'memories' as stimulii.