The Laboratory for Human-Centered AI

The existing institutional landscape tends to silo relevant areas of expertise. There are few organizations set up to provide opportunities for integrated projects that include broad philosophical inquiry and the production of code. Existing work that has passed through all three stages is often not fully recognized as having done so, since the starting points and end points of the “pipeline” fall into different disciplinary silos. The recent history of transformer-based large language models has demonstrated that concepts relevant to AI can be impossible to evaluate without computational experiments. Increasingly, we must not only know to build, but build to know.

HAI Lab seeks to amplify a perspective that sees AI as a powerful toolbox for the enhancement of human intelligence and human flourishing rather than its replacement. We take the view that this toolbox must be rooted in an understanding of human intelligence. It is in this sense that HAI Lab is the laboratory for human-centered AI.

Technologists are increasingly motivated to think more deeply about the meaning of their products and visions. We want to help cultivate a new kind of philosopher-technologist: One who asks how to build things that people might want and that contribute to human flourishing. If you are a young technologist at the beginning of your career, you do not merely want to be a cog in a machine. If you are on top of your field, you want to be aiming your hard-won skills at a conception of the good that stands up to scrutiny. Philosophy can help us get clearer on notions of human flourishing, and as a practice, it is itself a form of human flourishing. Just like the practice of engineering, you have to learn by doing. Having the right kind of dialogue is central to developing your practice of philosophy. As AI increasingly connects with all aspects of our lives, this practice will be key for deciding what we want to create.

The locus classicus is the story of how the philosopher Bertrand Russell’s work on the nature of mathematical knowledge ultimately led to Turing’s definition of a universal computer. Another classical case is the philosopher Saul Kripke’s work on the meaning of necessity, which has had significant impact in computer science. Philosophy has even influenced the development of AI powered space craft: The philosopher Michael Bratman’s work on the nature of practical reason, which has had applications in the design of intelligent agents through the work of computer scientist Mike Wooldridge and others. More recent cases are philosopher Helen Nissenbaum’s work on privacy and philosopher Ruth Chang’s work on choices. More examples can be found if we look at the branches of cognitive science that are continuous with philosophy, with classics including work by David Marr, Phil Johnson-Laird, and Joshua Tenenbaum.