Heterogeneous Engineering and Responsible AI are closely linked, especially when it comes to ensuring fairness and ethical design. Fairness cannot be treated as a purely technical standard, as human contexts such as relationships, organizational structures, and social norms must be part of the design and engineering process from the beginning. By considering these factors, we can avoid misleading framings of issues and oversimplified solutions often found in AI development.

Heterogeneous engineering emphasizes the need to think about both the technical components of AI systems and the human actors who interact with them. This doesn’t mean we can engineer human behavior, but it does suggest that we broaden the scope of engineering to include the social systems that influence AI outcomes, such as local incentives, regulatory environments, and institutional cultures. By integrating these sociotechnical factors, we can create AI systems that account for fairness as a property of the entire system.

This holistic view, where human and technical elements are analyzed together can fosters a deeper understanding of how AI systems affect society. As noted in From Abstraction Traps to Sociotechnical Risks,

“a heterogeneous approach helps identify social factors that influence technical outcomes. While this approach isn’t a magic wand since it can’t eliminate all ripple effects, it promotes responsible and human-centered design that serves the public good”.

Source

AI Development for the Public Interest: From Abstraction Traps to Sociotechnical Risks by McKane Andrus, Sarah Dean, Thomas Krendl Gilbert, Nathan Lambertand Tom Zick

A. D. Selbst, D. Boyd, S. A. Friedler, S. Venkatasubramanian, and J. Vertesi, “Fairness and abstraction in sociotechnical systems,” in Proceedings of the Conference on Fairness, Accountability, and Transparency, 2019, pp. 59–68.