Why would AI systems exhibit racism?

Originally from my blog: Full article on: Diverse Customer

"AI facial recognition works better for white skin - because it's being trained that way - an article in  World Economic Forum, in collaboration with MIT News makes an interesting observation."

I am not an AI specialist but a reader made a valid point:  

Light reflects more off of lighter skin than darker skin. Cameras work by seeing light. It's really that simple. Lots can be done to improve camera response, but you can't change the laws of physics by calling them racist.

However, there is indeed an opportunity to encourage our young minorities to enter these fields because, as we see, representation matters. We build off of what we know, frame of reference. A.I. is a technology enabler and in diversity marketing it not only harnesses the power of data through predictive analytics, but also help companies to be more proactive than corrective – being ready to face pitfalls traversing through fluctuating consumer behaviour and also help them create products that match up to common sense definitions of goodness of a diverse cohort who connect with the product deeply and emotionally.

To the consumer at large, A.I. can help in distinguishing between products that actually offer value as they claim to provide vs the products that are fraudulent and driven by greed. But A.I. like many technologies that are enabling humanity to reach its greatest potential has a limitation that is impossible to override – and that lies in its Creation.

It’s like machine learning, it is in a constant process of acquiring learning from humans that can be applied to drive radical changes in our world. It can gain insights on human emotional intelligence, become empathetic at some stage of its evolution.

Like anything that’s created, A.I. too has a solid imprint of its creator and in the hands that wield that power-. It’s beginning to look, think and act like them! Whether it will evolve and become all-pervading and unassuming remains to be seen in our ever growing complex operating environment.

For now, I trust every word spoken and so quoting Melinda Gates:

“You often hear the big challenge is teaching machines to think and behave more like humans but I can’t help but think we should be more specific,” she said. “We don’t want to teach computers racism or sexism. Or selfishness or greed. We want to teach computers the very best of what humanity has to offer. And for that to happen. We need the people building A.I. to represent not just one small slice of the human experience. But all genders. All ethnicities.”

To me personally, A.I. is bringing an idea synthesis - where innovation happens at the points of intersection of humanity and technology, leading to a rapid adaptability.We can either use AI to amplify our abilities, or our limitations - which points to a rigid dichotomy between technology and humanity!

-Amit Anand