Maybe you are even aware of the fact that AI performs best in complex calculations and monotonous analysis. How can we trust that an AI is acting in our best interest? These questions can be answered with the same set of criteria derived from research and government initiatives: safety, privacy, security, robustness, transparency, fairness, ethics and accountability. I would argue that if any of these is not fulfilled, then an AI is not worthy of trust. And for the lack of universal defenses, AI engineers tailor measures of security to each new one.

Source: Forbes December 28, 2020 12:11 UTC