The problem with too many men in artificial intelligence

As I pointed out yesterday, a growing subset of artificial intelligence work (which doesn’t require a degree in comp-sci) could open up AI work to women in other fields. 

However, the fact remains that the majority of people working on AI are currently men. For evidence, just look at the attendee list at major AI conference NIPS: Only 13.7% were women, according to Bloomberg.   

That’s bad news for AI, because it affects the data these algorithms and neural networks train on. Take for example Microsoft’s Twitter chatbot, Tay. Shortly after release Tay was spewing hate speech, because trolls were sending it racist, antisemitic tweets. Those bigoted tweets comprised the data-set from which  Tay had to learn from. 

If artificial intelligence is largely being shaped by men, then the questions it answers and problems it solves will be limited to a male viewpoint.   RR