A Love Letter to Data Ethics#
What’s this?
This is a standalone blogpost written by Vanessa Hanschke in February 2024. All opinions expressed are strictly her own. Vanessa Hanschke is a PhD researcher in Responsible AI. You can check out her latest project at https://www.gooddeed.ai. Amy Joint, Huw Day and Nina Di Cara helped with the final edit.
“Dear Data Ethics, We’ve been together for what 6, 7 or 8 years now? We were acquaintances throughout university, but I will never forget the first time I felt the sparks. It was at that Kate Crawford talk, where her slides didn’t work, but she was a complete boss at improvising her talk about you on the spot. That’s when I knew that I wanted to spend my every day with you. That I wanted to risk leaving my corporate job and doing a PhD just to be closer to you. Now that I can see the finish line of the PhD, I realise we’re at a crossroads. But I wanted to take some time this Valentine’s day just to tell you all the things I appreciate about you…”
Probably similarly to marriage, when you start a PhD and commit 4 years of your life to studying a topic, you have to be sure that you will enjoy the ride during the good and the bad times. One way to help you during the bad times (like a thesis writer’s block) is to remind yourself of why you love them. Today I want to do that with data ethics. So here are my top 5 reasons for why I love data ethics.
Data Ethics makes us question what it means to be human and do human things
The words “artificial intelligence” get thrown around these days, as if there was some prize money to be made for inserting them into conversations. Well, some people are definitely making money off proposing it as the proverbial hammer that makes everything look like a nail, but that is not the point I want to make. Unpacking the words behind AI, we see the core idea behind it, which is to create machine versions of human intelligence. However, to define what it means to be intelligent is no mean feat. When does a machine sufficiently imitate a human to be called intelligent? Should we do this? Are humans intelligent at all? Is our worth as humans defined by intelligence? And what should our relationship look like with things that possess human intelligence? We have found some things more easy to automate than others and so discussing data ethics opens up questions about the way we have categorised human skills. If some manual labour such as care jobs, truck driving and preparing a meal are so difficult to automate, why are programmers living a lavish life doing a profession that chat-GPT will soon conquer? So what kind of work is intelligent work? While I disagree with many of the answers that tech companies are offering us to this question, I love this age-old question that is so fundamental to human existence.
Data Ethics makes us question how we value our cultural heritage and the arts
Instead of focusing on dangerous or highly unpleasant tasks, nowadays developers of the most advanced AI systems have been fixating on domains, which are generally considered to be gratifying passions such as writing literature and making art. The backlash against this from scriptwriters and artists has provoked questions about the value of art and labour. Does art created by a machine have the same meaning as when it is created by a human? Who owns art, its techniques and the library of human knowledge? What is public domain and is crawling your data and ingesting it into a black box fair use?
Data Ethics makes us question old systems
Several of the most ethically contentious algorithmic systems are ones that replace established institutions by training on historical data, such as in policing, public services, health or insurance. In doing so, they often lay bare entrenched problematic notions that these institutions have been built on. For example, employment agencies that automatically sift through CVs and produce bias against women, made me question if hiring a person according to the minute differences in hobbies, really is the most effective way to build a workforce. Could effective training on the job be a more sustainable way? Algorithms which increase police presence in an area according to historical number of arrests made me think about the effects of prioritising funding the police (as opposed to restorative community work) and how urban geographies are shaped by it. Thinking about data ethics deeply means closely observing the things in our world that we are reproducing and reflecting on if we like these system, irrespectively of them being “datafied” or not. And so the one good thing about data and AI encroaching into every aspect of our lives, is that it creates an opportunity for data ethicists to question absolutely EVERYTHING, even things we took for granted.
Women and other minoritised people have been able to dominate the space
It makes sense for people who belong to groups that have been historically discriminated to immediately recognise the destructive technical manifestations of capitalism, colonialism and classism. As a mixed-race Muslim woman who grew up in Germany at the turn of the century and has moved around continents, I’ve experienced a fair bit of racism. When I got into Computer Science, I collected several anecdotes of sexism and machoism that maybe deserve a hate letter in a different blog post. So it comes as no surprise that the arguments made by tech ethics scholars - many of them women of colour such as Birhane, Buolamwini and Gebru - resonated deeply with me. I sometimes wonder if the reason that this community has been able to form a group and pierce through to public discourse is because we have achieved critical mass in representation in the industry. Obviously the state of today’s diversity metrics are not sufficient or satisfactory in any way, but there are enough of us in data work, AI and tech in general to make noise, so that we can’t be ignored.
Many people who do Data Ethics are really nice, caring and thoughtful
The data ethics club has been a such a great safe space for us to bond over our frustrations, worries and ideas of how technology could be better. So if you are not a part of it already, come join the data ethics club.
So that was it, my love letter to data ethics. Today I chose to look at it through rose-tinted glasses, but I am aware that some of these endearing aspects, can also be seen as flaws. For example, the existential questions around AI have been instrumentalised to generate hype and to distract from the more mundane impacts of technology such as contributing to the climate breakdown and exacerbating inequalities. Another example is that having your identity associated with advocacy for minoritised groups can be a burden that many people don’t want to add to their daily struggle. But again, this is all material for a different letter.