Three Kinds of AI Ethics

AI ethics is not one unified field but three distinct approaches: The paper identifies "ethics and AI" (using AI to study ethics), "ethics in AI" (embedding ethical considerations into AI development), and "ethics of AI" (examining the ethical implications of AI systems). Understanding these differences helps clarify what specific problems different researchers are actually trying to solve.

Different kinds of AI ethics require different expertise: Each approach demands different skill sets and disciplinary backgrounds. For example, "ethics in AI" may require computer scientists and engineers, while "ethics of AI" may require philosophers and social scientists. Recognizing these distinctions can help organizations assemble appropriate teams for their ethical challenges.

Many criticisms of AI ethics miss their target: The paper argues that critiques of AI ethics often fail because they evaluate one type of AI ethics using the standards of another. This suggests that debates about whether AI ethics is "working" may be confused about what goals different researchers are actually pursuing.

The rapid, chaotic growth of AI ethics creates confusion: The field's sudden expansion across multiple disciplines makes it difficult to track debates and compare findings. This fragmentation means stakeholders may struggle to understand what AI ethicists actually do and what they can realistically accomplish.

Clearer categorization enables better discussions: By systematizing the three kinds of AI ethics and their distinct goals, the paper provides a framework for more productive conversations about the scope, methods, and training needed for responsible AI development.