@hrisskar I think the person in the video misses the point about AI and bias. Algorithms can be biased because what they do depends on what the implementer thinks is important, for example if something need to be real time it may use different algorithms even if on average it will perform slower (a good example is garbage collection). Training neural networks will also often favor the normal quality case rather than the logical quality case due to use of association rather than reason.

@hrisskar That doesn't mean all hope is lost though, by being smarter in our data sets and chosen techniques more can be done to ensure fairness. That being said, please don't depend on AI for anything capable of determining the course of society.