Cultural Responsiveness
Attention to the cultural and linguistic contexts in which learners and educators experience assessment.
MASTER’S THESIS · RESPONSIBLE AI
The CARE Framework.
From stakeholder evidence to a framework for culturally responsive AI-mediated assessment.
THE QUESTION
What does responsible AI adoption look like across educational cultures?
AI-mediated assessment raises questions about fairness, language, and institutional readiness. My thesis asks how a framework can help institutions consider these conditions together, rather than judge an AI tool apart from the people who use it.
The unit of inquiry is bigger than the AI tool. It includes the educational context around it.
METHOD RATIONALE
Understand the context before defining the criteria.
Interview educators and administrators
I interviewed eight people: one educator and one administrator in each of India, the USA, the UK, and France. Recruitment responses shaped this country selection. Role-specific, semi-structured guides let me examine classroom practice alongside institutional decisions without treating either perspective as the whole picture.
Code, compare, and synthesize
I analyzed interviews using four predefined themes: pedagogical, institutional, ethical, and cultural. I developed subthemes within them and compared accounts across roles and contexts. Written coding justifications connected the evidence to my interpretations and framework criteria.
Gather student feedback
I designed and analyzed a student survey using a plain-language framework summary and lifecycle diagram. Of 31 respondents, 28 completed the framework-evaluation questions. Respondents reported institutions in seven named countries: India, the USA, the UK, France, Canada, Ireland, and Spain; two did not specify a country. I also compared CARE with existing frameworks to clarify its contribution.
RESEARCH EVIDENCE
Different concerns pointed to connected requirements.
Interview accounts linked AI assessment to linguistic diversity, unequal access, cultural assumptions, institutional guidance, and transparency. I treated these as contextual accounts, not representative descriptions of countries.
Student feedback added questions about who owns each action and how assessment risk should change the requirements. That feedback informed the framework’s treatment of accountability and risk. It evaluated the framework as presented, not its effectiveness in institutional use.
EVIDENCE TO FRAMEWORK
How I moved from an account to a criterion.
These three examples paraphrase the analysis. The criteria are my design responses to the evidence, not participant quotations or instructions.
Language is part of assessment access
Evidence: Interview accounts raised differences in students’ language backgrounds and the language of instruction.
Interpretation: A tool’s suitability depends on the learners it must serve.
Framework response: I translated this into Cultural Responsiveness criteria for understanding the student population and checking multilingual capability where needed.
Bias requires scrutiny before adoption
Evidence: Accounts raised concerns about training-data bias and assessment norms that favor dominant groups.
Interpretation: Technical performance alone cannot establish that an assessment tool fits its context.
Framework response: I connected Cultural Responsiveness and Ethics through criteria for auditing cultural bias and training-data diversity.
Access depends on more than the tool
Evidence: Accounts described socioeconomic and digital divides alongside differences in whose knowledge is recognized.
Interpretation: Selecting a tool without understanding the student population can overlook barriers to participation.
Framework response: I made understanding students’ backgrounds and access conditions part of the pre-adoption assessment.
THE FRAMEWORK
Four dimensions.
One connected assessment.
CARE brings four dimensions into a shared model for considering institutional readiness and cultural inclusivity in AI-mediated assessment.
Assessment Alignment
Alignment between the use of AI, pedagogical intentions, and the learning that assessment is meant to capture.
Readiness and Governance
Institutional preparedness, guidance, and accountability for adopting AI in assessment.
Ethics
Fairness, transparency, and equitable treatment throughout the assessment experience.
CARE organizes criteria and reflective questions across pre-adoption, transition, and post-adoption, with review feeding back into decisions. Its contribution is a way to connect context, assessment goals, readiness, and ethics throughout adoption, not a single readiness score.
Published concept · Subsequent thesis research
The INTED 2026 paper presents the conceptual rationale and a proposed three-phase mixed-methods design. It is a published research-in-progress contribution, not a report of all the findings shown here. The subsequent thesis interviews, analysis, student feedback, and framework comparison are complete. I am writing the final thesis draft.
LIMITATIONS
Feedback on a framework is not proof of impact.
The interview sample was small and shaped by recruitment responses. Participants were familiar with AI; their accounts do not represent whole countries or institutions. The student survey was weighted toward master’s and STEM students, with the USA the largest group.
Students evaluated a summary and diagram, rather than using CARE in an institutional decision. The study supports interpretation and refinement of the framework. It does not establish that CARE improves assessment outcomes. Institutional implementation has not been tested.
NEXT STEPS
Finish the thesis. Keep the boundary clear.
My remaining task is the final thesis draft. Whether CARE’s criteria work in institutional practice remains an open question, not a completed validation result.
RELATED PUBLICATION
CARE: A Framework for Cultural Assessment of Responsive AI through an Ethical Lens
Mahajan, R. & Beaton, C. · INTED 2026 Proceedings
Oral presentation · Valencia, Spain