Comment by poyu

This exactly scenario is why our company is so afraid to put AI into production without the results being completely clear that it could be wrong. But if there’s even a chance that it could be wrong, why are we offering it to the user? How much due diligence does the user need to do? Does the benefits outweigh the cons?