With the recent white supremacists issues in Charlottesville, many were claiming that this type of behavior is learnt, it is not inherent when we are born. The same may be true of AI systems.

Whilst AI can be programmed to remove gender bias, as well as re-enforce it, we may have a bigger problem with machine learning. If the information a system is fed is inherently biased the system will accept those biases. therefore although we make AI to be as pure as a baby, we might be creating monsters.