If not, how do we teach values to an autonomous intelligence? Can we codify them or simply enter it somewhere in the system? Is it more of an iterative process where we will correct parameters on the fly as systems learn on their own and potentially behave unexpectedly?

It does not seem practical, ideal or even risk-free to teach values to an AI to preserve ourselves and avoid unwanted situations. Situations comes to mind where an AI’s behavior was observed but its actions were unpredictable and there were not any options to correct the course. As we are facing those new complexities, behaviors and potential uses, it makes sense to reflect on and explore what rules are needed. It is a pressing concern as we expand in this new field, especially as AI usage grows and takes over critical applications.

#technology #startup #innovation #software-development #artificial-intelligence