Question 2/27 v4 lecture 5

AI algorithms (for instance, for determining health benefits) are so resilient to errors they do not need supervision nor a systemic way of identifying and fixing issues. True or false, and why?


The deployment of AI based software requires a good recourse and appeals process. Lack of transparency and oversight on life affecting decisions can bring forth a lot of suffering. A good way of monitoring the performance of an AI based system along with effective processes for making corrections are a must.

Relevant part of lecture