In July 2022, the Center for Advanced Study in the Behavioral Science’s “Toward a Theory of AI Practice” project convened a conference on Responsible Hybrid Intelligence at the Rockefeller Foundation’s Bellagio Center. The participants coauthored a white paper, completed in October 2022, that advances a new paradigm broadening the focus beyond algorithms viewed in isolation to processes of human-algorithm collaboration in service of the public good.
We propose an alternate approach to mainstream AI practice that broadens the focus beyond algorithms viewed in isolation to processes of human-algorithm collaboration. The envisioned practice would harness human and machine complementarities to develop systems of human-machine hybrid intelligence. Such systems integrate the best capabilities of both machine intelligence and human users while mitigating the deficits of each. In this approach, outcomes can be improved not only by improving the underlying technologies but also by improving the human-machine collaboration processes.
Given its focus on human-machine collaboration, the hybrid intelligence paradigm presents a new set of scientific questions and design requirements that flow from the simultaneous consideration of machine capabilities and human psychology, behaviors, needs, and values. The envisioned practical field will require a conceptual foundation that extends beyond computational and statistical sciences to also integrate concepts and methods from the behavioral and decision sciences, human-computer interaction (HCI), human-centered design, and applied ethics. Responsibly scaling up the envisioned practical field will require providing development teams with tools that enable them to draw from multiple disciplines and perspectives, and engage in inclusive and participatory approaches to design. The proposed approach complements recent calls to foster responsible computing research and “operationalize ethics” in industry.