Unless you have been living under a rock, you’ll know by now that algorithms – a sequence of steps to process some information and return a result – are responsible for a lot of things in our lives. Your bank tells you that you are pre-approved for a loan? Algorithm. Website shows you different things to your friends? Algorithm. Freaky matches on your dating app? Algorithm.
But how do we know what those algorithms are doing? How do we know we can trust them? How do we know they are not biased against us, because of our age, sex, ethnicity or sexual orientation? How do we know that, even if these particular data points are hidden from the algorithm, that it is not inferring them from other information?
There is a responsibility on the designers of algorithms to consider all of these things. And there is a responsibility on those that employ and commission them to validate their work against such standards. But that assumes a green field site. A situation where we are building an algorithm using known and trusted data, and without reliance on existing steps and processes.
Layers of complexity
Algorithms have been around for a long time. They rarely operate in isolation in such a virgin environment. They are compounded and integrated, fed skewed data sets. Their outputs get twisted from the intention. They end up as one element in a complex chain of processes that is near incomprehensible, even to those who created them.
Who do you call when this happens? The Algorithm Archaeologists. Data professionals who can spelunk down through the chained processes to understand the complete sequence, and audit and identify any bias or breakdown.
These people exist today. They are consultants from places like Accenture, whose innovation centre, The Accenture Dock, I visited again on a research trip yesterday. Algorithm Archaeologists is my term for them and it feels right.
As we build up more and more layers of technology, one on top of the other in opaque strata of complexity, we will need more people with these skills to help us maintain our faith in the systems that support us, and ensure that they are not exacerbating existing inequalities, or creating new ones.