Changemakers: Natalie Garrett on Ethical Debt and Ethics Education in Tech

Natalie Garrett is a PhD student researching technology ethics and privacy at the University of Colorado Boulder. Her pioneering research at The Internet Rules Lab and Pervasive Data Ethics for Computational Research (PERVADE) examines why tech companies fall short on ethics and what can be done to change the tide. Natalie spoke with Andrew from ATIH about the concept of ethical debt and the importance of ethics training for technical workers.

Andrew: How does your upbringing affect how you understand your relationship with technology today?

Natalie: As a 38 year old who was born into a home with a rotary phone, I’m not sure where to start with this question. I guess my relationship with technology starts with my upbringing without technology. There are some people who believe the right technology can solve any problem, and I tend to disagree. While I love tech and the ways it has enhanced my own life, I believe that there should be a consideration of various approaches before jumping to a technical solution. I guess an upbringing without technology has made me skeptical of technological solutionism. 

How and why did you first become passionate about the need to think critically about the ethical dimension of technology?

I started wondering about the ethical dimensions of technology when working at and with startups and connecting with the tech community in Boulder in 2014. There was and still is an ‘anything goes’ approach which is great for some things (like, let’s try to revolutionize space launch and make it cheaper), but not so great for other things (like, an app that listens to your doctor/patient conversations and has a founder who hasn’t considered privacy issues). There’s a rush to solutions in tech and the number of tech controversies we see in the news daily is evidence that the rushed approach needs to shift. 

You and your colleague, Casey Fiesler, are researching a concept called ethical debt. What is the inspiration for this research?

The inspiration for this research is my work with product teams at startups and my interactions with the tech community here in Boulder. At startups that are under-resourced or experiencing intense business pressure to validate a product in the market, products and features are shortcut. When teams don’t have the time, expertise, or resources to build the best product possible, they often refer to the shortcuts or identified areas that will need fixing (aka refactoring) later as ‘technical debt’. 

When we transition this concept to ethical debt, the results are similar. Many of the reasons a team may accumulate technical debt (lack of technical expertise, lack of time, lack of consideration for all potential uses of a technology, and/or strategic business decisions about risk and cost) are similar to those that lead to the accumulation of ethical debt. The mentality of “we’ll fix that later” does not work so well for ethical debt. While it is relatively easy to refactor code, it is harder for companies to fix harm and recover from public outrage over a controversial feature or product. It takes a long time to build public trust and not long to lose it. 

What makes ethical debt a uniquely useful framework for understanding ethical failures in tech?

Ethical debt is a useful term that names the knowledge gaps on tech teams and it borrows from the familiar concept of technical debt. Additionally, and like technical debt, ethical debt identifies systemic gaps in expertise in the tech industry that need to be addressed. Gaps in expertise are unsurprising since our research in ethics education in undergraduate computer science departments indicates that ethics, when it IS taught, is often the last course in a curriculum or topic in technical courses. While universities are definitely offering more and more tech ethics courses, the ideal is to teach students to consider social ramifications of technologies as part of technical learning. 

You mention in a working paper you and Casey wrote that technologists can be prone to thinking of ethics as a specialized topic for philosophers to worry about. While better ethical training for technologists is important, do dedicated expert ethicists have a role to play in creating ethical technologies? If so, what should that look like? 

Absolutely, expert ethicists definitely have a role to play! While universities are making great strides in teaching ethics to computer science majors, it is still taught as a specialization or (at its worst) as an afterthought at the very end of a student’s undergraduate career. We want computer scientists to graduate with an understanding that ethics is a part of being a computer scientist. When it is a part of technical practice, it will be much easier for those dedicated experts to do their work within tech companies. 

What we want to avoid is using expert ethicists as public relations add-ons or scapegoats. For ethics to be considered as a business function, the entire company needs to buy into considering the social ramifications of products. Ideally, the expert ethicists lead the charge and everyone from the C-level to the entry-level programmer considers ethics as part of their daily work. Ideally an expert ethicist would spend zero time convincing their organization that considering ethics is important work and avoid an internal communication/persuasion struggle. 

What are some of the key things that schools, bootcamps, and tech companies should keep in mind when designing and implementing ethics training programs for technical workers? 

The answer here is simple yet takes some effort and intention-- ethics needs to be integrated into technical curriculum so it is seen by students and professionals as part of technical practice. The era of ‘move fast and break things’ is over.

Here’s an example of incorporating ethics in computing from our research here at CU Boulder: Thanks to the support of Mozilla’s Responsible Computer Science Challenge, we’ve been working with professors who teach introductory programming to retrofit existing assignments to include ethics. The assignments still cover basic coding concepts, but we also introduce the ethics considerations prior to the assignment in a short discussion. The two assignments we’ve created so far are about contextual and behavioral advertising and college admission algorithms. I’ve been really delighted at the response from the students. Anecdotally, the expressions on their faces show that many of them are learning about these concepts for the first time. 

You can connect with Natalie on LinkedIn and Twitter and can keep up with her latest work on Medium.