AI & Algorithmic Registries

Governments are increasingly turning to algorithmic systems, often using AI technologies, for decision-making support, but residents often have little understanding or even awareness that these technologies are being used. Registries are one form of transparency that can also help improve transparency and accountability by requiring agencies to publish information about AI and algorithmic systems in consideration or use. 

The Pittsburgh Task Force on Public Algorithms recommended that government agencies be required to publish information about algorithmic systems on a public website. This work provides an overview of considerations for public algorithmic registries as well as prototype sketches.  

Considerations for Public Registries

This paper discusses considerations for public sector algorithmic registries, identifying the kinds of information that can be captured by government agencies to be shared with the public.

Algorithm Registry User Interface

This prototype of a user interface for a public algorithm registry provides an example of how governments can operationalize the considerations discussed in the accompanying paper.

Participating Researchers: 

  • Carl DiSalvo, PhD, Georgia Institute of Technology, College of Computing
  • Rosta Farzan, PhD, University of Pittsburgh, School of Computing and Information 
  • Robert Gradeck, University of Pittsburgh, Center for Social and Urban Research