CRASA

Community Responsive Algorithms for Social Accountability (CRASA)

CRASA is a multidisciplinary, community-based participatory research program to develop an algorithm accountability benchmark to meet societal and legal needs and guide best practices.

image
image
image
News

CRASA Project Leads Received Faculty Excellence Awards

Assoc. Prof. Ryan Kennedy and Distinguished University Prof. Ioannis A. Kakadiaris received the Awards for Excellence in Research, Scholarship and Creative Activity for the 2021-2022 academic year.

An illustration containing electronical devices that are connected by arm-like structures. By Anton Grabolle / Better Images of AI / Human-AI collaboration / CC-BY 4.0

Algorithms play an important role in our lives

Data-driven automated decision systems play an increasingly substantial role in making or assisting many critical decisions around the world.

Research Program

CRASA is a mixed-method and multidisciplinary research program for addressing the issue of algorithm accountability. The overall research strategy involves five objectives.

image

Participatory Research

In-depth interviews with stakeholders and establishing a community advisory board to guide and evaluate the research.
image

Legal Review

Review of existing and proposed legislations and regulations for algorithm accountability, transparency, and fairness.
image

Accountability Benchmark

An algorithm accountability benchmark (AAB) to audit automated decision systems for accountability standards.
image

Behavioral Experiments

Conducting behavioral experiments to assess the elements of the AAB and building a body of behavioral evidence.
image

Scoring Tools

Developing scoring tools and processes for automated decision systems along the dimensions of the AAB.

AAB is the foundation for aligning AI to human values

The algorithm accountability benchmark will help achieve a trustworthy and safe AI that is welcome in society and can help tackle humanity's most significant challenges with the goal of improved social good.

Two digitally illustrated green playing cards on a white background, with the letters A and I in capitals and lowercase calligraphy over modified photographs of human mouths in profile. By Alina Constantin / Better Images of AI / Handmade A.I / CC-BY 4.0