Ethics in the Wild: Why we need to keep asking questions
By Sam Osys, GAABS Director and GAABS Editorial Board Member
One thing that stood out at this year's Design Research Society (DRS) conference in Edinburgh was the absence of a dedicated design ethics theme. Two years ago, at the DRS conference in Boston, ethics had a visible home of its own. This year, there was no standalone ethics track. Instead, there were sessions on Design for and with Values, Designing for Social Change, and Design with Care: Theories, Practices, Methods, and Politics.
But it didn't feel like ethics had disappeared. It felt like the opposite. Questions about values, responsibility and the consequences of design choices appeared across discussions about AI, care, sustainability, social justice and more-than-human design. Ethics seemed to have moved from being a topic in its own right to becoming part of how researchers approach almost every topic. This feels significant, as it suggests ethics no longer needs a dedicated track because it is becoming part of the wider conversation.
But there is another question worth asking: Where does ethics happen when we leave the conference and return to the work?
From universal principles to ethics in context
The arrival of ChatGPT 3.5 and the subsequent expansion of generative AI have brought questions about technology's consequences to the forefront of many conversations. This has led to overall greater attention to ethics, alongside interest in methods, tools and frameworks that might help us address unintended consequences.
One response has been to look for universal principles. What rules should everyone follow? What values should guide the development of AI? How can we make sure that technology serves people rather than harms them? These questions matter. Principles can help organisations establish expectations and give people a basis for discussion. But principles alone cannot tell us everything about how ethics works in practice.
A principle such as fairness may sound straightforward until we have to decide what fairness means for a particular group of people, in a particular setting, using a particular intervention. What counts as fair may depend on who is affected, what information is available, what decisions the organisation can make, and what consequences follow.
As such, research in ethics has increasingly shifted to a more situated and contextual understanding of ethics (Chivukula et al., 2024). Rather than treating ethics as a set of rules that can be applied in the same way everywhere, this work asks how ethical questions emerge through particular situations, relationships and practices. This shift matters for behavioural science, because we work with people, organisations, and systems, and the context is not something we can separate from the intervention. It shapes what the intervention does and how people experience it.
My own research has taken me further into this question.
At this year's DRS conference, I presented a paper in the Design in the wild track: Beyond situated ethics: Towards a practice-based understanding of how designers do ethics (Osys, 2026). The empirical study that I mention in this paper comes from my PhD research, which explores how design practitioners do ethics in organisational settings.
While reviewing the literature around design ethics, I noticed that much of the existing work focuses on two areas: normative questions about what designers ought to do, and empirical research with students.
Both provide insight. But they do not tell us everything about how ethics gets negotiated, supported or sidelined in professional practice. To understand that, we need to look at the people doing the work.
I interviewed design practitioners working in areas such as UX, Service Design, business strategy, and transformation, across different corporate organisations and sectors. I wanted to understand what happens when ethical concerns enter a project, how practitioners respond, how other stakeholders respond, and what happens when other priorities take over. The practitioners I spoke to worked on both internal-facing projects that impacted colleagues and external-facing products and services that impacted customers.
One thing that stands out is that ethics rarely enters as a conversation labelled “ethics”. Instead, practitioners use current company priorities to talk about ethics and unintended consequences. Ethics may appear as a question about accessibility, a concern about data, or a discussion about a feature. This is why studying ethics in practice matters. It allows us to look beyond what organisations say they value and examine how people work through ethical questions in the settings where decisions happen.
Reflection as a way of doing ethics
One recurring theme in my research is the role of reflection. Frameworks and methods can provide structure. But because project work often comes with deadlines, competing priorities and limited resources, there may not always be time to run a dedicated ethics workshop or complete a lengthy assessment. Reflection offers another way to create space for ethical questions (Buwert & Sinclair, 2024). The practitioners that I spoke to often talked about the process of setting aside time to reflect, potentially with other team members of with wider project teams. They spoke of using stand-ups or community of practice sessions to reflect and gather feedback and insights from others. This reflection usually took the form of asking questions. These questions could be broad or could focus on a very specific feature or intervention.
The questions do not need to be complicated:
What could go wrong?
Who might be affected?
What assumptions are we making?
What happens if people use this in a way we did not intend?
What happens if the organisation changes how it uses the data?
What might this intervention make possible in the future?
One of my favourite thought experiments is called the Black Mirror test, which I first heard about in a podcast mentioned by guest speaker Roisi Proven. Inspired by the dystopian television series Black Mirror, the exercise asks you to imagine that the intervention or feature you are working on becomes the subject of an episode and ask yourself: What is the worst possible outcome?
Imagine you are designing a behavioural intervention that collects information about people's habits. The intervention might help people change their behaviour. But what happens if another organisation later gains access to that data? What happens if someone uses it to make decisions about employment, insurance or access to services? The point is not to predict the future. It is to make potential consequences visible while there is still time to question the design.
The Black Mirror test is one example. Other thought experiments, including the trolley dilemma (Thomson, 1984), have been used to explore ethical questions in areas such as autonomous vehicles. But we do not need a famous thought experiment to start reflecting. Asking a question about the people affected by our work can be enough.
What does this mean for behavioural science?
For behavioural scientists, there are two main areas where all of this matters.
Firstly, ethics in research. We need to understand the people we are designing for. That means engaging with real people, listening to their experiences and questioning the assumptions we bring to the research. This becomes particularly important as AI enters more of our research practices. AI can help us organise information, explore ideas, check writing and identify patterns. But it can also produce interpretations that sound plausible without data support.
I recently tested this while analysing qualitative data. I used a secure AI model to triangulate my findings and explore whether patterns emerged in how practitioners talked about behavioural science in relation to the demographic information they disclosed. The model identified what it described as a gender difference in language. It suggested that women used more experience-related words, while men used more concrete examples and specific frameworks. My data did not support this finding. I reviewed the material again and found no significant difference in emotional language between genders. Women were also just as likely to mention specific frameworks as men.
The exercise was useful because it showed me how easily an AI-generated interpretation can sound convincing. It also showed the value of checking what the model produces against the research material. AI can act as a collaborator. But it cannot replace the work of engaging with people and examining the evidence. It also has inherent biases, as seen in the example.
This is also why the distinction between assistive AI and generative AI matters. Using AI to check spelling or improve the flow of writing is different from asking it to generate an analysis, interpretation or section of research. Research guidance is beginning to distinguish between these uses, including when researchers need to disclose AI involvement.
The important question is not simply whether we use AI. It is how we use it, what we ask it to do, and how we check its output.
The second area where I think ethical reflection matters is in designing interventions.
Behavioural science interventions change situations. They may influence what people do, what information they see, what choices they make, or how organisations respond. That means we need to keep questioning the intervention throughout its development. What happens if people respond differently from what we expected? What happens if an organisation uses the intervention for a different purpose? What happens if a feature designed to help one group creates a problem for another?
We cannot anticipate every consequence. But reflection can help us identify some of the questions before the intervention reaches people. And if we identify a risk, we can ask ourselves and other on our team what changes might reduce it.
Ethics in the wild
The absence of a dedicated ethics track at conferences may signal that ethics is becoming a fundamental part of research rather than something separate from it. But ethics does not become embedded in practice simply because we talk about it at conferences or publish principles and frameworks. It happens through the questions people ask, the decisions they make, the concerns they raise, and the things they choose to investigate further. For all practitioners, this means keeping ethics close to the research and the intervention.
Talk to people. Question the data. Challenge the assumptions. Reflect on what might happen next. And when AI becomes part of the process, ask the same questions of the technology.
References
Buwert, P., & Sinclair, M. (2024, June 23). Thought Experiments In The Ethics Of Designing For Future People. DRS2024: Boston. https://doi.org/10.21606/drs.2024.518
Chivukula, S. S., https://orcid.org/0000-0003-2294-0119, View Profile, Gray, C., https://orcid.org/0000-0002-7307-1550, View Profile, Li, Z., https://orcid.org/0009-0009-2461-1715, View Profile, Pivonka, A. C., https://orcid.org/0009-0006-9421-2339, View Profile, Chen, J., https://orcid.org/0009-0005-1812-0670, & View Profile. (2024). Surveying a Landscape of Ethics-Focused Design Methods. ACM Journal on Responsible Computing, 1(3), 1–32. https://doi.org/10.1145/3678988
Osys, S. (2026). Beyond situated ethics: Towards a practice-based understanding of how designers do ethics. DRS 2026: Edinburgh. https://doi.org/10.21606/drs.2026.2077
Thomson, J. J. (1984). The Trolley Problem Comment. Yale Law Journal, 94(6), 1395–1415. https://heinonline.org/HOL/P?h=hein.journals/ylr94&i=1415
A Next Step Worth Taking
Members of GAABS receive early notification and offers to register for live events. To receive your early notifications, JOIN US.