Whenever I give a talk, I always share how much I love Q&A. It’s a real joy to hear what people are curious about and provide resources or share insightful lived experiences as a consultant in the data ethics space. In this line of work, it’s usually not about having tidy, easy answers or the “right” answer – it’s much more nuanced and complex.
This is why I thought it might be fun and useful to write an advice-style column about data ethics for the DATAVERSITY community. There is so much data that we already have, much more data that we could collect, and so many things we could do with that data – but should we? Beyond privacy regulations and our cybersecurity commitments to safeguard data, we have a vast array of choices to make when it comes to data. This is the realm of data ethics. Increasingly these questions are also entangled with AI and other digital technologies.
I’ve come to this work on data ethics from artificial intelligence. I started my research into AI ethics in 2018 and it led me to appreciate the role of data and data governance. Things are now coming full circle for me as many data professionals are now tasked with understanding what AI means for their work. Through this column I hope to cover not just data ethics but also AI ethics too.
For our first column, I’m tackling a common question that I’m regularly asked:
Why Does Data Ethics Matter?
Data is powerful. We know that with great power comes great responsibility. Spiderman fans will attest to this. This is especially important when it comes to data that involves people either directly or indirectly.
We can think about the power of data to determine important things in our lives. For example, consider the tale of Doug Lee. He was denied boarding a flight to take his dream vacation in Ireland because his ticket said “Doug Lee” but his passport said “Douglas Lee.” Three missing letters cost him $8,000 in non-refundable fees. Doug(las) isn’t alone – there are far too many stories like this.
You might think this is just a data quality issue and on one level, that is true. His name didn’t match on the two systems. But there is a deeper level to this story. The documentation, the data, was more “real” and carried more authority than the actual physical person – Doug(las) Lee – who was standing at the gate trying to board the flight. Doug(las) Lee’s own declaration that he was indeed “Douglas Lee” – that he was the same person represented by both pieces of documentation – did not matter. The government-issued passport data was given the final decision and carried more power than the physical person. In his book, “How We Became Our Data,” philosopher Colin Koopman outlines the history of both how this happened and why. It has to do with running complex bureaucracies, and it’s a fascinating tale about the last hundred years of information processing that have led to this outcome.
AI Increases Our Need for Ethics
We are in a time where AI and machine learning are enlarging the scope of data’s power. Data can be processed in complex, sometimes inexplicable ways, and then used to make inferences or predictions that can determine high-stakes decisions. Data can decide who gets a job, a loan, access to education, or a social services benefit, or even who is deemed to be a criminal. Most of us see these decisions made by data as objective and authoritative, even if they might be based on inaccurate, incomplete, or biased inputs.
Data ethics calls us to consider the impacts of data on people. Data ethicists help organizations to critically examine these impacts from multiple stakeholder perspectives and to make decisions throughout the data lifecycle that will align with and uphold agreed-upon values and commitments.
Send Me Your Questions!
I would love to hear about your data dilemmas or AI ethics questions and quandaries. You can send me a note at hello@ethicallyalignedai.com or connect with me on LinkedIn. I will keep all inquiries confidential and remove any potentially sensitive information – so please feel free to keep things high level and anonymous as well.