Beyond the Algorithm: How Humanists Became Indispensable to AI Governance

When the White House Office of Science and Technology Policy released its AI Bill of Rights blueprint in October 2022, the document's framing was striking. It spoke not of optimization functions and training parameters but of civil rights, discrimination, due process, and human dignity. The language was unmistakably humanistic—and its presence in a federal technology policy document was not accidental. It reflected a quiet but consequential repositioning of humanities scholars as central actors in one of the most consequential technological conversations of our time.

For much of the past decade, discussions about AI governance took place in a register dominated by computer scientists, economists, and policy analysts trained in quantitative methods. The implicit assumption was that the problems posed by artificial intelligence were fundamentally technical—problems of accuracy, efficiency, and scale that technical expertise was best positioned to address. That assumption is now being systematically challenged, and the scholars challenging it most effectively are those whose disciplines have spent centuries developing the tools to analyze power, interrogate assumptions, and ask what it means for a system to be just.

The Disciplinary Realignment

The reorientation did not happen overnight. Its roots lie in a growing body of critical scholarship—emerging from philosophy, science and technology studies, history, and legal theory—that identified the limitations of purely technical approaches to AI ethics in the early 2010s. Researchers like Safiya Umoja Noble, whose 2018 book Algorithms of Oppression documented racial and gender bias in search engine results, and Virginia Eubanks, whose Automating Inequality examined how automated decision systems harm poor Americans, demonstrated that the most consequential questions about AI were not about technical performance. They were about whose interests systems were designed to serve and whose were systematically discounted.

This scholarship found its way into policy conversations through several channels. Congressional testimony, which had previously featured almost exclusively technical witnesses when AI was on the agenda, began to include philosophers and sociologists. The National Science Foundation expanded its AI research funding to explicitly include social and humanistic inquiry. And a growing number of technology companies—responding partly to public pressure and partly to genuine internal concern—began hiring ethicists, anthropologists, and policy scholars into roles with actual organizational authority.

Philosophers in the Boardroom

The figure of the corporate ethicist has attracted considerable skepticism, some of it deserved. Critics have noted that ethics roles at technology companies are often structured to produce the appearance of moral deliberation without constraining the decisions that actually matter. The rapid dismissal of Google's Ethical AI team leadership in 2020—following the publication of research that reflected poorly on the company's flagship language model—provided a high-profile illustration of what happens when corporate ethics functions conflict with commercial imperatives.

Yet the picture is more complicated than this cautionary example suggests. At a number of institutions, humanists have secured positions with genuine influence over technology development and deployment decisions. The Partnership on AI, a multi-stakeholder organization whose members include major technology companies, academic institutions, and civil society organizations, has integrated humanistic perspectives into its working groups on fairness, transparency, and accountability. The AI Now Institute at New York University—co-founded by Kate Crawford, a scholar whose work draws on critical theory, history, and media studies—has produced policy-relevant research that has directly informed federal regulatory proposals.

Several federal agencies have also moved to institutionalize humanistic expertise in their AI oversight functions. The National Institute of Standards and Technology's AI Risk Management Framework, finalized in 2023, incorporates concepts drawn from philosophy of technology, organizational sociology, and legal theory alongside the technical benchmarks one might expect from a standards body. The framework's emphasis on trustworthiness as a multidimensional concept—encompassing fairness, explainability, and accountability as distinct properties—reflects sustained engagement with humanistic scholarship on what it means for a system to be worthy of trust.

The Case for Disciplinary Diversity in Tech Oversight

The argument for including humanists in AI governance is not simply that diverse perspectives produce better outcomes in some abstract sense—though the evidence for that proposition is substantial. It is that the specific intellectual tools developed by humanities disciplines address problems that technical methods are structurally ill-equipped to handle.

Consider the problem of value alignment—the challenge of ensuring that AI systems behave in ways consistent with human values and social goods. This is, on its face, a philosophical problem. It requires clarity about what values are at stake, how they conflict, whose values should be privileged in cases of disagreement, and how values that are difficult to quantify can be represented in systems that operate through quantification. These are precisely the questions that moral philosophy, political theory, and the study of ethics have been developing frameworks to address for millennia. Delegating them exclusively to engineers—however talented—is not a division of labor that any serious institution would accept in other high-stakes domains.

Historians bring a different but equally essential contribution. The tendency to treat AI as an unprecedented rupture with the past—a technology so novel that historical analogies are inapplicable—has repeatedly produced governance failures by obscuring the ways in which current systems reproduce and amplify historical patterns of discrimination and exclusion. Scholars with training in the history of technology, the history of surveillance, and the history of labor have consistently provided analytical frameworks that allow policymakers to anticipate second-order effects that purely prospective analysis misses.

Institutional Models Worth Watching

Several American universities have moved to institutionalize the integration of humanistic expertise into AI research and governance in ways that go beyond individual faculty appointments. The University of Michigan's Institute for Data Science houses a dedicated program in the social and ethical dimensions of data science that funds collaborative projects between humanists and computer scientists. Princeton's Center for Information Technology Policy similarly bridges technical and humanistic inquiry, producing research that informs both engineering practice and policy deliberation.

At Stanford, the Human-Centered AI Institute has made the inclusion of social scientists, philosophers, and legal scholars in its research agenda a founding commitment, explicitly rejecting the model of ethics as an add-on to technical research. These institutional structures matter because they create conditions under which humanistic perspectives are integrated into research from the outset rather than consulted retrospectively when problems have already been baked into deployed systems.

The Work Ahead

The repositioning of humanities scholars as essential participants in AI governance represents genuine progress, but it also carries risks that the academy should name clearly. The demand for humanistic expertise in technology contexts creates incentive structures that may distort scholarly agendas, drawing researchers toward questions that are legible to technology companies and government agencies while undervaluing work that is critical, speculative, or difficult to translate into policy recommendations. The scholar who can speak fluently to a Senate subcommittee or a corporate ethics board is not always the same scholar producing the most rigorous or challenging intellectual work.

These tensions are navigable, but they require deliberate institutional attention. Universities that are serious about the contribution of humanities scholarship to AI governance must protect the conditions under which critical, independent inquiry can flourish—including the freedom to reach conclusions that powerful actors find inconvenient. The value of humanistic expertise in these conversations derives precisely from its independence. Instrumentalizing it in ways that compromise that independence would be a poor trade, however attractive the seat at the table.