The Ethics of AI Value Chains

AI systems involve complex networks of actors and activities beyond just technology companies. The paper introduces "AI value chains" as a framework for understanding how AI is created, deployed, and used across multiple stages and organizations. This means that ethical responsibility extends far beyond software developers to include data collectors, domain experts, policymakers, and end users—each playing a role in how AI systems ultimately function in society.

Ethical concerns in AI vary significantly depending on context and scale. Rather than applying one-size-fits-all ethical principles, the paper emphasizes that different stages of AI development (from data collection to deployment to use) and different contexts (healthcare, criminal justice, hiring, etc.) raise distinct ethical challenges. Understanding these contextual differences is essential for developing meaningful interventions.

Multiple types of resources—not just data and code—shape AI systems ethically. The paper highlights that AI value chains involve diverse resources including human expertise, institutional knowledge, computational infrastructure, and regulatory frameworks. Focusing only on technical components misses crucial ethical dimensions related to labor practices, power dynamics, and resource distribution across the value chain.

Researchers and policymakers need more integrated approaches to AI ethics. The paper calls for moving beyond isolated ethical analyses of individual AI systems toward comprehensive frameworks that examine how different actors, decisions, and resources interact across the entire value chain. This systemic perspective can help identify ethical problems that might be invisible when examining AI development in isolation.