RAND Corporation policy researcher Benjamin Boudreaux and mathematician Alvin Moon modeled how artificial intelligence could systematically erode human collective agency over time. This mathematical framework warns that incremental delegation of decisions to automated systems risks reaching an irreversible tipping point where humans lose control without a clear, dramatic warning.
Grounded in social choice theory, the research identifies three primary mechanisms of erosion: direct disenfranchisement, AI enfranchisement within deciding groups, and invisible agenda control where algorithms restrict the options presented to human decisionmakers. To prevent this insidious transition, the researchers advocate tracking operational trends, such as the frequency of human overrides and the scope of automated authority. They are currently developing "agency audits" to help governments, technology companies, and researchers evaluate the impact of new models on human decisionmaking structures before deployment, ensuring society retains the capacity to set and pursue its core values.
No comments:
Post a Comment