GENERATIVE AI · LANGUAGE & LEARNING
Examining the relationship between AI-supported academic work, learners’ understanding, and faculty expectations.
THE QUESTION
Generative AI adds a new layer to the relationship between language, academic expectations, and assessment. My co-authored research examines multiple language learners’ use of these tools and the challenges faculty encounter when interpreting AI-supported work.
The two contributions approach this question through learner experiences and faculty perspectives, connecting concerns about literacy and comprehension with responsible AI use.
STUDY 01 · INTED 2026 · PUBLISHED CONFERENCE PAPER
Barriers and Bridges examines how students navigate academic expectations and how their work differs between independent and AI-supported tasks.
I contributed to study design, conducted interviews and analysis, and wrote the literature review and analysis sections.
The published study reports eight computing students in higher education. It also includes interviews with five computing faculty members; these are part of this paper, not the separate faculty study below.
Interviews explored students’ AI use and how they understood assignment expectations. Each student then completed two reading-and-writing tasks: an independent passage task first, followed by a similar passage and writing prompt with AI support. Comparing each student’s time and score across the tasks brought observed work alongside interview accounts, rather than relying on reported AI use alone.
Observed: AI-supported completion times were shorter for all eight students. Independent scores ranged from one to four out of six; seven students scored six out of six with AI support. Figures 3 and 4 report these task results.
Interpretation: Students produced stronger scored responses in the supported task. That is not evidence of retained learning or improved independent ability.
Reported: Students described instructor-specific cues, implicit rules, and concern about penalties. Faculty regarded expectations as clearly communicated. Tables 2 and 3 document this mismatch.
Implication: Explicit, assignment-level AI guidance and examples could address this communication gap. Their effectiveness was not tested here.
This was a small study in computing education. The tasks used different passages and a fixed order, so practice, order, and task differences cannot be separated from AI support. The paper reports descriptive differences without an inferential statistical test. The results describe this comparison, not a causal effect of AI. Task scores do not establish long-term comprehension or transfer to later independent work.
STUDY 02 · EDULEARN 2026 · PUBLISHED CONFERENCE ABSTRACT
This separate study examines faculty perceptions of classroom participation, communication, and generative AI use among multiple language learners.
I contributed to study design, conducted interviews, and worked on the literature review.
Six computing faculty members at a New York state university were recruited through snowball sampling. Semi-structured interviews used a qualitative narrative approach, with thematic analysis. Interviews suited a question about how faculty interpreted classroom experiences; they did not directly measure students’ skills.
Reported: The abstract describes faculty frustration with generative AI in written assignments and concern about students’ ability to assess its accuracy.
Interpretation: These accounts identify a concern to investigate, not evidence that students cannot evaluate AI output.
Reported: Faculty also raised language-model bias, unequal technology access, and understanding of ethical use.
Proposed guidance: The abstract calls for a socially responsible AI taxonomy that addresses equity, access, and ethics. It does not report a validated taxonomy or an evaluated intervention.
These are perspectives from a small, snowball-recruited group in one discipline and institution. They do not establish how common the concerns are across higher education. The published abstract does not provide detailed coding evidence, quotations, or measured student outcomes.
RESEARCH IMPLICATIONS
The learner study puts task performance alongside uncertainty about academic rules. The faculty study identifies concerns about verification, access, and ethics. Together they motivate clearer guidance and research into how students check AI output, without treating polished work as proof of understanding.
These questions connect to CARE’s focus on assessment alignment, cultural responsiveness, and institutional guidance. CARE is a separate thesis project; these studies do not validate its framework.
Explore the CARE Framework →RELATED PUBLICATIONS