The optimal balance between trusting AI technology and incorporating human expertise is still being explored, but the goal is to have humans trust AI when it's working correctly and identify when it's not.