A wider view of human-centered AI.
We look beyond how an AI behaves toward a person. We ask what conditions around a group of people make desirable human relationships and behaviors more possible.
This means studying information flows, roles, norms, and relationships alongside computational systems. An intelligent tool is one part of that environment.
Start with the situation.
Use theoretical and qualitative inquiry to understand what matters to members of a group and how their environment shapes participation.
Make assumptions visible.
Build limited-purpose computational models and simulations that make relationships and alternative arrangements easier to examine.
Test with care.
Develop bounded experiments and instruments. Proposed field research should establish an appropriate protocol and consent before participation.
Follow the effects.
Evaluate how changes affect agency, relationships, and collective outcomes. Treat uncertainty and unwelcome effects as part of the inquiry.
Individual boundaries matter.
People's ability to understand and control their digital representations is central to this agenda. Research into collective intelligence must also account for individual agency, access, and the boundaries between people.