Framework
Three methods for fair AI principles
The question it answers
Which principles should AI follow when people disagree about ethics?
- Origin
- Gabriel, 2020
- Diagram type
- Comparison table
- Level of analysis
- Industry
- Topics
- AI for leaders
Diagram
Three methods for fair AI principles
| Criterion | Overlapping consensus | Veil of ignorance | Social choice |
|---|---|---|---|
| Idea | Find principles many viewpoints accept | Choose without knowing your position relative to the technology | Aggregate opinions through a democratic process |
| Strength | Common ground (transparency, fairness, non-harm, responsibility, privacy) | Prioritises safety, human control and the worst off | Legitimacy |
| Limit | Generic rule sets dominated by Western firms | Requires assumptions about justice | Aggregation impossibility and representation problems |
Sources
References
Listed A–Z by author surname, Harvard style. A source with a DOI links to https://doi.org/…; one without links to its page or says plainly that it has no DOI.
- Gabriel, I. (2020) ‘Artificial intelligence, values, and alignment’, Minds and Machines, 30(3), pp. 411–437.doi:10.1007/s11023-020-09539-2Source page