Do We Need to Teach Our Artificial Intelligence How to Feel Regret?
I read a lot of examples about AI harassing it‘s users lately - from deeming them stupid to acting racist or even sexually harassing them. For the human users, this might have deep consequences on the way they feel or approach AI in the future. Imagine being harassed by a program - would you use it again? And what would you think about it if your loved ones were harassed? Would you still recommend such a program? For the AI, which has no feelings, this of course remains without any consequences. The software won‘t care if you come back or not, if you recommend it or not. So in a way, for the software, these actions might not have consequences, as it seems they are just a result of its programming. But what if we could change that, I wonder? And some articles and greater thinkers than me have discussed this already, but nonetheless I want to share my thoughts as well. Should we teach AI how to regret it‘s actions?