Mind the Gap: The Bidirectional Relationship between Diversity, Equity, and Inclusion (DEI) and Artificial Intelligence (AI)



Shima Salehi

Stanford University


Rod Roscoe

Arizona State University

The education research community aspires to support the learning process of a diverse body of students and ensure that students’ educational opportunities are unconstrained by their background or identity. This panel discussion will explore how these goals may be served by the bidirectional relationship between (a) artificial intelligence (AI) methods and (b) diversity, equity, and inclusion (DEI) approaches in education.

Panelists will discuss how AI methods can promote DEI in learning environments to support all students regardless of their background. From this “AI for DEI” perspective, AI methods can empower educators and researchers to more accurately monitor and identify learners’ needs and progress. In turn, these insights might inform more equitable learning. Panelists will also discuss how DEI approaches can improve AI analysis and interpretation, such that outcomes are not tuned to a narrow population of learners while biased toward others. This “DEI for AI” mindset argues that applying a DEI lens to conceptualizations, methods, and applications of AI may reduce bias while preventing the creation or exacerbation of discriminatory outcomes.

This panel will present scholarly works that contribute to the bidirectional links between AI and DEI, and explore the challenges, next steps, and future plans the research community should consider to strengthen this connection.

You can submit your question(s)/comment(s) for the discussion section of the panel here.


Nia Dowell

University of California, Irvine

profile picture

Rose Luckin

London Knowledge Lab

Chris Piech

Stanford University.

Marcelo Worsley

Northwestern University

Research-based Digital-first Assessments and the Future of Education


profile-pic (4)

Alina A. von Davier


AI, learning engineering, computational Psychometrics and big data coupled with numerous technology breakthroughs propose a new paradigm for education. From adaptive learning systems to digital-first -testing with automated content generation and automatic scoring – the possibilities for efficiency, scalability and access are promising. The unprecedented disruption of COVID-19 leaves little doubt that advances in learning sciences and technology can augment the in-classroom educational experience.

Digital-first assessments, sometimes called intelligent assessments are a new generation of tests where the technological advances and AI affordances are used to (re)create comprehensive assessments that are adaptive, efficient, rigorous, valid, and, most distinctively, attuned to perfect the user’s experience. Digital-first assessments may be integrated into other systems (school systems, LMS, etc) being part of the new Internet of Education (IoE), where through integrative frameworks and standards one can optimize the support for each student while protecting their privacy. Stealth assessments through the use of process data from interactive tasks and multimodal data sources are moving from research labs into practice.

The panelists will share their research, provide evidence of how these new methodologies work, and engage the audience in a thought-provoking discussion on the impact of the new tests on education in general.


Saad Khan

FineTune Learning

Michelle Barrett


Jill Burstein


Valerie Shute

Florida State University