Uncovering Ethnic Bias in Learning Materials using Learning Analytics
Lead Research Organisation:
The Open University
Department Name: Institute of Educational Tech (IET)
Abstract
In an increasingly digital world, the pivotal role of online learning platforms for education remains unchallenged, providing access to knowledge and learning opportunities for students worldwide. As any other technology, however, online learning settings face a significant issue that threatens its inclusivity and diversity: persistent ethnic bias embedded within educational content. Recent research shows how such bias can marginalise ethnic minority groups, negatively impacting their academic performance and engagement with the learning material. Identifying and addressing such bias is complex and challenging due to the very large volume of educational data and due to the subjective nature of ethnic bias, which varies across cultural and individual perceptions. This complexity renders manual identification impractical and necessitates innovative, technology-driven solutions.
Aims
Recognizing the urgency of this issue, this project aims at understanding and uncovering potential ethnic biases in text-based learning materials using AI-empowered technology. In particular, this fellowship has the following main objectives:
Disseminate findings on ethnic bias in online learning to the academic community. Via peer-reviewed journals and conferences.
Raise practitioners awareness about ethnic bias in learning materials. Via workshops and networking events.
Advance professional skills in AI technologies for inclusive education.
Refine existing bias detection models in educational content, using ethnic bias as a blue-print for quality assurance support at large.
How will this be achieved?
To achieve these objectives:
Findings on ethnic bias will be shared via journal articles and conference presentations.
Practitioner awareness will be raised through 3 workshops delivered to 100 stakeholders.
Professional AI skills will be enhanced through online courses and workshops.
Bias detection models will be refined through targeted research. Although it is anticipated that limited research will be conducted, this stage comprises updating the models with new data and assessing them against previous performance metrics.
Why is this relevant?
Manually reviewing educational content for biases is a daunting and time-consuming task, often impractical due to the vast amounts of material available online. By using an automated strategy, this project offers educators a proactive approach for authoring and revising content that reflects the diversity of learners, as a means to increase accessibility of educational content. Additionally, it raises awareness among educators about the importance of inclusivity. This dual approach - combining AI-driven technology with heightened awareness - serves as a powerful mechanism to reduce barriers to equity in education, ensuring that students from various ethnic backgrounds have access to inclusive learning opportunities.
Expected Outcomes
The key outcomes of this project include:
Validated training process and AI models to identify and help reduce bias, creating validated tools and re-usable patterns for their validation for enhancing content quality.
Boosted awareness among stakeholders (i.e., researchers, educators, and policymakers) and strengthened networks, sharing best practices for inclusivity.
More equitable learning materials, improving access and accessibility to quality educational content for all ethnic backgrounds.
Broader Impact
This research's broader impact extends to shaping a globally-inclusive educational landscape, enhancing AI precision, scalability, awareness, and collaboration to ensure learning materials reflect societal diversity.
Aims
Recognizing the urgency of this issue, this project aims at understanding and uncovering potential ethnic biases in text-based learning materials using AI-empowered technology. In particular, this fellowship has the following main objectives:
Disseminate findings on ethnic bias in online learning to the academic community. Via peer-reviewed journals and conferences.
Raise practitioners awareness about ethnic bias in learning materials. Via workshops and networking events.
Advance professional skills in AI technologies for inclusive education.
Refine existing bias detection models in educational content, using ethnic bias as a blue-print for quality assurance support at large.
How will this be achieved?
To achieve these objectives:
Findings on ethnic bias will be shared via journal articles and conference presentations.
Practitioner awareness will be raised through 3 workshops delivered to 100 stakeholders.
Professional AI skills will be enhanced through online courses and workshops.
Bias detection models will be refined through targeted research. Although it is anticipated that limited research will be conducted, this stage comprises updating the models with new data and assessing them against previous performance metrics.
Why is this relevant?
Manually reviewing educational content for biases is a daunting and time-consuming task, often impractical due to the vast amounts of material available online. By using an automated strategy, this project offers educators a proactive approach for authoring and revising content that reflects the diversity of learners, as a means to increase accessibility of educational content. Additionally, it raises awareness among educators about the importance of inclusivity. This dual approach - combining AI-driven technology with heightened awareness - serves as a powerful mechanism to reduce barriers to equity in education, ensuring that students from various ethnic backgrounds have access to inclusive learning opportunities.
Expected Outcomes
The key outcomes of this project include:
Validated training process and AI models to identify and help reduce bias, creating validated tools and re-usable patterns for their validation for enhancing content quality.
Boosted awareness among stakeholders (i.e., researchers, educators, and policymakers) and strengthened networks, sharing best practices for inclusivity.
More equitable learning materials, improving access and accessibility to quality educational content for all ethnic backgrounds.
Broader Impact
This research's broader impact extends to shaping a globally-inclusive educational landscape, enhancing AI precision, scalability, awareness, and collaboration to ensure learning materials reflect societal diversity.
