I am currently reading The Power of Habit: Why we do what we do and how to change by Charles Duhigg. One of the things the book describes is the habit loop: a cue, a routine, and a reward. The cue triggers a part of the brain to take over, carry out the routine and then there is a reward. The routine can be quite complex, but it is very similar each time.
This is why we can get in the car and not remember anything between sitting down in the seat (cue) and getting to work (reward). The routine bit of driving gets done by the other part of the brain. Of course this only works if it is routine. If things are out of the ordinary, like major (new) roadworks, or an accident happens, then the routine is disrupted and your conscious brain takes over.
It strikes me that the current state of machine learning (ML) and artificial intelligence (AI) is that they can replicate the habit loop. They can perform routine tasks that people can do as a habit. But as soon as the routine is broken, the AI is unable to function correctly.
Rereading articles in this light, that computers can emulate our habits but not the conscious part of our brains. So AI is fine with many tasks, but as long as they don't have to deal with something unexpected.Go Top