- Westin’s privacy preferences
- Forsyth’s moral taxonomy
- Winner’s technology as politics
- AI challenges historical frameworks
- References
The OpenAI board crisis shows that AI ethics debates are contentious. The ethics debates in AI cover issues like privacy, regulation, and technology driven social inequality, which are debates that go back to before modern AI.
Westin’s privacy preferences
Westin studied American perspectives and found three different categories of views, privacy fundamentalists, privacy pragmatists, and privacy unconcerned and tracked changes over time. AI assessment will need to balance data collection needs with widely varying privacy expectations.
Forsyth’s moral taxonomy
Forsyth found 4 types of views about the importance of rules, absolutists want strict rules, exceptionalists favour rules with edge exceptions, relativists want rules but alter for contexts, subjectivists adapt as required. Different groups are likely to react quite differently to AI assessment policies.
Winner’s technology as politics
Winner said that technology is political. AI might require a power structure, be consistent with a political orientation, or be used to decide social issues. In its strongest form, AI is a power structure. AI assessment systems are going to concentrate power in the hands of systems designers and administrators.
AI challenges historical frameworks
Many issues go across categories and different groups have different starting points. AI ethical solutions need to be partial rather than universal, interim rather than lasting, and have wide rather than complete agreement (i.e., Lindblom’s incrementalism from the 1950s).
AI also challenges these historical frameworks with its extreme capabilities, but the roots of the issues aren’t modern. Newer work e.g, Emily Bender, Margaret Mitchell, Timnit Gebru, Meredith Whittaker, Cathy O’Neill, Safya Noble) focuses specifically on AI (e.g. whose ethics get implemented).
References
Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021, March). On the dangers of stochastic parrots: Can language models be too big? In Proceedings of the 2021 ACM conference on fairness, accountability, and transparency(pp. 610-623).
Forsyth, D. R. (1980). A taxonomy of ethical ideologies. Journal of Personality and Social psychology, 39(1), 175.
Lindblom, C. (2018). The science of “muddling through”. In Classic readings in urban planning (pp. 31-40). Routledge.
Noble, S. U. (2018). Algorithms of oppression: How search engines reinforce racism. In Algorithms of oppression. New York university press.
O'Neil, C. (2017). Weapons of math destruction: How big data increases inequality and threatens democracy. Crown.
Westin, A. F. (1968). Privacy and freedom. Washington and Lee Law Review, 25(1), 166.
Whittaker, M., Crawford, K., Dobbe, R., Fried, G., Kaziunas, E., Mathur, V., West, S.M., Richardson, R., Schultz, J. and Schwartz, O., 2018. AI now report 2018 (pp. 1-62). New York: AI Now Institute at New York University.
Winner, L. (2017). Do artifacts have politics? In Computer ethics (pp. 177-192). Routledge.
Next section
Peep into the future: fine-tuned, multimodal, neuro-symbolic
Last section
LLMs and psychometric bias
Return home
Psychometrics.ai
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0).
