Rogue Algorithm
  • 19th March, 2019

Rogue Algorithm

By Ekow Duker

3 min read

In this the second in our series, “Don’t want to embrace AI? What’s the worst that can happen?”, we look at why it pays to have some understanding of how these new technologies work.

When talking about the rapid democratisation of technology, the analogy is often made that a chef doesn’t need to know how a microwave functions. That’s certainly true for cooking. It’s not necessarily true for data science or machine learning.

At a recent seminar in Cape Town, I heard a representative from a global technology company extolling the virtues of advanced machine learning tools that anyone can simply access from the cloud. “They’re really easy in practice,” he gushed. “And the beauty of it is, you don’t have to be a data scientist to use them.”

Technological advances tend to be simplified over time to the point where what was once only accessible to a few cognoscenti, is used on a daily basis by many. I can remember many years ago in London, having to call the operator to ask to place an international call. That would be laughable nowadays when we can phone friends and family in other countries without a second thought. The technology just works and we don’t need to know the intricacies of satellite switching to take advantage of international calling.

My neighbour in Johannesburg, a remarkable Jewish woman, once told me of her grandparents’ arduous journey from Lithuania to Cape Town in the early part of the last century. Upon arriving in Cape Town and seeing black people for the first time, the new arrivals innocently misclassified them as bears. For within their frame of reference, any dark skinned creature, standing upright on two legs could only be a bear.

Joy Buolamwini, the founder of the Algorithmic Justice League, has written authoritatively about the shortcomings of commercial facial analyses systems and the dangers they pose to civil liberties. Consider a police technician using a commercial facial analyses system for mass surveillance. Let’s also suppose that the technician’s upbringing and personal belief system leads him to expect certain demographics to have criminal tendencies. Just like the Lithuanian couple who landed in Cape Town some one hundred years ago and could not imagine that a black person could be anything other than a bear, is the police technician likely to pause and reflect when a flawed algorithm routinely confirms his world view? Maybe. Maybe not.

So yes, commercial AI systems can be startlingly easy to use. However given the furious pace of development of the technology and the lack of oversight, it pays to be skeptical about them. And to know a little about how they work under the hood.

Photo by Corey Motta on Unsplash


AI News

In this article, Cassie Kozyrkov, Chief Decision Intelligence Engineer at Google takes an unsentimental look at what it really means to be a data scientist in an organisation.

Share This Artcle :

About Author

Ekow was previously the Chief Analytics Officer for the Retail and Business Bank, Barclays Africa where he was responsible for harmonising and repurposing the bank’s Analytics and Data Science functions. An ex-oilfield engineer, Ekow has C-suite experience in strategy consulting and private equity investing and brings a deep commercial understanding of several industries. He is passionate about data and getting things done. Ekow is also a published author of four novels - Dying in New York, White Wahala, The God Who Made Mistakes and Yellowbone.