AICE team member Madeleine Hayenhjelm has delivered Deliverable 6.2, an original ethics research paper on AI ethics that has been submitted to a peer-reviewed international ethics journal.
Titled Getting AI Ethics Right: The Principles We Have vs. The Principles We Need, the paper builds on earlier AICE work and sets out a clear set of standards to judge ethical principles, lists of principles, and wider frameworks. It acts as a practical guide that does two things: it helps design stronger, more consistent ethical guidance, and it provides a clear way to see where current approaches fall short.
Alongside issues already discussed in the literature, the paper highlights four further weaknesses that can undermine ethical advice for action and for making the right moral decisions:
- Moral overloading, where one principle points in several directions
- Mixed categories, where different kinds of ethical claims sit on the same list
- Moral dilution, where clear obligations are softened in practice
- Weak moral foundations, which make conflicts and complex cases hard to resolve
The message is forward-looking. We do not need to abandon principles. We need standards that ensure clarity, moral accuracy, and genuine action guidance across principles, lists, and frameworks.
Coming Soon. Once accepted, the full paper will appear in the Outputs section of the AICE website alongside our other publications. Keep an eye out, as it marks a significant step forward in shaping how we think about getting AI ethics right.
