How does it come that a tiny, technical detail like the choice of the Bluetooth version, which at first glance looks as the incarnation of technological neutrality, raises mayor ethical concern? Why is it that the actors individually were not able to discern the discrimination of an AI app, but only in the collaborative act, facilitated by a reflection tool? In my contribution, I would like to show that this is precisely where the so-called principles-to-practice gap closes. It is often inconspicuous technical issues where values are concretised in technical detail and become effective in practice. The re-framing of the problem was made possible by an arranged meeting between ethicists and developers– and a specific way of collaborative reasoning and thinking, facilitated by some reflection tools. They introduced a guided reflection journey with which the focal points and consequences of technology ethics could be analysed. Answering how abstract values can be "translated" for a specific technical application: this challenge is currently discussed in AI ethics as the so-called principles-to-practice gap (e.g. Bleher & Braun, 2023; Corrêa et al., 2024; Morley et al., 2020, 2021). By drawing on the above case of an ethics intervention, I will argue that the gap is currently misplaced which makes it so difficult to find a satisfatory way of bridging it. What is needed instead is an epistemology and ethics of integration which allows for entry points to bring together disciplinary thinking styles and their corresponding epistemic practices. I will show that in a successful integration, the latter are brought together in design moves which allow to explore the preconditions and impacts of the future artefact in a systematic way. This makes it possible to scrutinise the design from an ethical point of view and to anchor ethical considerations directly in the processes of generating knowledge about the potential product. As I would like to show, this process requires a number of prerequisites. It requires a re-conceptualisation of the concept of technology, which must start with a contextualisation in order to be able to understand artefacts from a processual and praxeological approach. I will conclude by comparing the approach for an early ethics intervention to existing ones like embedded ethics, STIR, real-time TA, disclosive ethics, and VSD.

References

Bleher, H., & Braun, M. (2023). Reflections on Putting AI Ethics into Practice: How Three AI Ethics Approaches Conceptualize
Theory and Practice. Science and Engineering Ethics, 29(3), 21. htps://doi.org/10.1007/s11948-023-00443-3
Corrêa, N. K., Santos, J. W., Galvão, C., Pasetti, M., Schiavon, D., Naqvi, F., Hossain, R., & Oliveira, N. D. (2024). Crossing the
principle–practice gap in AI ethics with ethical problem-solving. AI and Ethics. htps://doi.org/10.1007/s43681-024-00469-8
Morley, J., Elhalal, A., Garcia, F., Kinsey, L., Mökander, J., & Floridi, L. (2021). Ethics as a Service: A Pragmatic Operationalisation of
AI Ethics. Minds and Machines, 31(2), 239–256. htps://doi.org/10.1007/s11023-021-09563-w
Morley, J., Floridi, L., Kinsey, L., & Elhalal, A. (2020). From What to How: An Initial Review of Publicly Available AI Ethics Tools,
Methods and Research to Translate Principles into Practices. Science and Engineering Ethics, 26(4), 2141–2168.
htps://doi.org/10.1007/s11948-019-00165-5