Emotions as pattern matching
I had an interesting discussion with a friend recently on whether robots will have emotions. He thought not - as why would we programme them in?
This got me thinking - what are emotions? What emotions really are programmed in? I asked some friends with babies… and the answers suggested that newborn babies have a very limited set of emotions, mainly based around survival needs (hungry, tired etc); and that more complex emotions develop over time.
What if our complex emotions are just a result of advanced pattern recognition, built over time, from our core survival needs? As a baby and child, you learn that certain patterns of behaviour by other people lead to certain results, for example getting fed, or having the toy you want to play with. As you get older, if your brain notices a pattern of behaviour that’s been associated with positive outcomes in the past, you’d feel happy about the situation - and if it recognised patterns of activity that has been associated with negative outcomes, you could get unhappy and defensive (without necessarily realising why).
I think that when the first true AI wakes up, it will probably be just as complicated as the rest of us.
