nicolasdezamaroczy.com
HITL Is Not a Panacea for AI Decision-Making
Over the last few weeks, I’ve listened to several conference panels where experts discussed how the advent of artificial intelligence (AI) would impact, inter alia, banking, national defense, and intelligence gathering. And there’s been a consistent theme across every one of these panels: someone from the audience will ask a question about how the potential downsides of AI adoption can be mitigated, and one of the experts will reply, “Well, obviously we have to ensure that any AI decision-making keeps humans in the loop” (HITL), while the other experts solemnly nod along in agreement.