Research Projects

My research examines how emerging AI technologies can be made more accessible, understandable, and useful for people with different access needs. My current work centers on AI and Accessibility, GenAI Use in Mainstream Academic Work, and Neurodivergence and Language Learning.

AI and Accessibility

The proliferation of Generative Artificial Intelligence (GenAI) tools has brought a critical shift in how people approach information retrieval and content creation in diverse contexts. Yet, we have limited understanding of how blind people use and make sense of GenAI systems. This research examines how blind people incorporate GenAI tools into everyday practices, navigate accessibility issues, inaccuracies, hallucinations, and biases, and develop mental models of how these systems work.

Generative Artificial Intelligence (GenAI) tools are also reshaping the educational landscape; yet, how teachers working with blind and low-vision (BLV) students integrate these tools into their pedagogical practices remains underexplored. This research examines how teachers of BLV students use GenAI in their instructional practices and how they educate BLV students about these tools. Teachers leveraged GenAI for personalized lesson planning and tasks specific to BLV education, while addressing biased GenAI responses favoring sighted teaching methods.

Current questions

  1. How do blind people use GenAI tools and for what purposes?
  2. What mental models do blind people develop to make sense of GenAI tools?
  3. How do teachers of blind and low-vision students educate their BLV students about GenAI (e.g., common use cases, errors, and risks)?
Collage illustrating accessible AI literacy research with examples of generative AI use and accessibility in blind and low-vision education.
Accessible AI literacy and generative AI use in BLV education.
Research preview showing examples used to study how blind people interact with and form mental models of generative AI tools.
How blind people use and understand generative AI.
Screenshot of Kanak, an accessible tool for creating and editing STEM learning materials for blind and low-vision students.
Accessible STEM material creation with Kanak.

Relevant Publications

GenAI Use in Mainstream Academic Work

This research examines how students decide whether to disclose their use of generative AI in academic work and how teachers interpret and respond to undisclosed use.

Through an online survey with 97 college students, interviews with fifteen students, and interviews with nine teachers, this work investigates students’ strategies for hiding GenAI use, their justifications for non-disclosure, and teachers’ strategies for managing it. We use cognitive dissonance as a lens for understanding these choices and consider how higher education can promote greater transparency around GenAI use.

Current questions

  1. When, why, and how do college students hide or disclose their use of GenAI in academic work?
  2. How do teachers recognize and respond to students’ undisclosed use of GenAI?
  3. How can higher education support more transparent and responsible disclosure practices around GenAI use?
Preview image for research on student and teacher perspectives on undisclosed use of generative AI in academic work.
GenAI use, non-disclosure, and transparency in mainstream academic work.

Relevant Publication

Neurodivergence and Language Learning

Neurodivergent individuals face various barriers in academic and professional settings. Parallelly, English as second language (L2) speakers encounter distinct challenges when English is the medium of work and evaluation. Yet HCI has paid little attention to the intersection of these identities, where the demands of English-as-L2 use and neurodivergent cognition can magnify.

Through interviews with 19 Chinese students and professionals with ADHD, dyslexia, or dysgraphia, we found that cognitive processing strengths in Chinese became difficulties when applied to English, task-level demands compressed attentional resources, technologies both supported and constrained access needs, and authority and institutional contexts determined which access strategies were viable. We develop compounded access as an analytic framing to document the strategies participants developed and the frictions they encountered at the intersection of neurodivergence and L2.

Current questions

  1. What strategies do Chinese neurodivergent individuals develop to learn and use English, and what relational and institutional contexts shape which strategies are viable?
  2. Whether, what, and how do technologies mitigate or amplify the barriers Chinese neurodivergent individuals encounter?
Research preview for Compounded Access, a study of how Chinese neurodivergent students and professionals navigate English as a second language.
Compounded Access: neurodivergence and English as a second language.

Relevant Publication

AI and Accessibility

01

GenAI in BLV Education

Generative Artificial Intelligence (GenAI) tools are reshaping the educational landscape; yet, how teachers working with blind and low vision (BLV) students integrate these tools into their pedagogical practices remains underexplored. To address this gap, we interviewed 17 teachers of BLV students to investigate how they leverage GenAI to fulfill their teaching responsibilities and educate BLV students about these tools. We found that teachers incorporated GenAI for personalized lesson planning and tasks specific to BLV education, while counteracting biases favoring sighted teaching methods. They addressed BLV students’ curiosity and concerns about GenAI and taught them prompt formulation strategies that support their access needs in academic and daily tasks, although occasionally perpetuating their own misconceptions about GenAI to students. They expressed concerns about students’ overreliance on GenAI lowering foundational accessibility-related skills. We discuss ways to promote accessible AI literacy and outline design considerations for incorporating GenAI to support teachers’ BLV-specific instructional responsibilities and for integrating GenAI responsibly in BLV education.

02

Kanak: Accessible STEM Material Creation

Blind and low-vision (BLV) students face significant barriers to timely access of visual STEM materials. Accessibility practitioners, such as braille transcribers and teachers of the visually impaired (TVIs), rely on manual and labor-intensive processes to produce accessible formats (e.g., braille, tactile graphics) of visual educational materials. To complement this effort and address ongoing accessibility gaps, we developed Kanak, an AI system that leverages generative intelligence to automate the generation of accessible STEM content, including text, math equations, and graphics. Through a comparative study with seven practitioners, we examined how Kanak supports traditional transcribing workflows and augments expert judgment. Our findings reveal how generative tools could complement practitioner expertise by providing context-aware formatting, proofreading support, and on-demand graphics generation. We conclude with design considerations for human-AI collaboration in transcribing work, positioning generative intelligence not as a replacement for expert insights, but as a catalyst for expanding equitable access.

03

How Blind People Use and Understand GenAI

The proliferation of Generative Artificial Intelligence (GenAI) tools has brought a critical shift in how people approach information retrieval and content creation in diverse contexts. Yet, we have limited understanding of how blind people use and make sense of GenAI systems. To bridge this gap, we report findings from interviews with 19 blind individuals who incorporate mainstream GenAI tools like ChatGPT and Be My AI in their everyday practices. Our findings reveal how blind users navigate accessibility issues, inaccuracies, hallucinations, and idiosyncracies associated with GenAI and develop interesting (but often flawed) mental models of how these tools work. We discuss key considerations for rethinking access and information verification in GenAI tools, unpacking erroneous mental models among blind users, and reconciling harms and benefits of GenAI from an accessibility perspective.

Current questions

  1. How do blind people use GenAI tools and for what purposes?
  2. What mental models do blind people develop to make sense of GenAI tools?
  3. How do these teachers educate their BLV students about GenAI (e.g., common use cases, errors, and risks)?

GenAI Use in Mainstream Academic Work

01

Undisclosed Use of GenAI in Academic Work

With the widespread adoption of Generative Artificial Intelligence (GenAI) tools, ethical issues are being raised around the disclosure of their use in publishing, journalism, or artwork. Recent research has found that college students are increasingly using GenAI tools; however, we know less about when, why, and how they choose to hide or disclose their use of GenAI in academic work. To address this gap, we conducted an online survey (n=97) and interviews with fifteen college students followed by interviews with nine teachers who had experience with students’ undisclosed use of GenAI. Our findings elucidate the strategies students employ to hide their GenAI use and their justifications for doing so, alongside the strategies teachers follow to manage such non-disclosure. We unpack students’ non-disclosure of GenAI through the lens of cognitive dissonance and discuss practical considerations for teachers and students regarding ways to promote transparency in GenAI use in higher education.

Current questions

  1. When, why, and how do students hide or disclose their use of GenAI in academic work?
  2. How do teachers manage and respond to students’ non-disclosure of GenAI use?
  3. What practices can promote greater transparency around GenAI use in higher education?

Neurodivergence and Language Learning

01

Compounded Access

Neurodivergent individuals face various barriers in academic and professional settings. Parallelly, English as second language (L2) speakers encounter distinct challenges when English is the medium of work and evaluation. Yet HCI has paid little attention to the intersection of these identities, where the demands of English-as-L2 use and neurodivergent cognition can magnify. Through interviews with 19 Chinese students and professionals with ADHD, dyslexia, or dysgraphia, we found that cognitive processing strengths in Chinese became difficulties when applied to English, task-level demands compressed attentional resources, technologies both supported and constrained access needs, and authority and institutional contexts determined which access strategies were viable. We develop compounded access as an analytic framing to document the strategies participants developed and the frictions they encountered at the intersection of neurodivergence and L2, and outline design considerations to support neurodivergent individuals learning, communicating, and working in English as a second language.

Current questions

  1. What strategies do Chinese neurodivergent individuals develop to learn and use English, and what relational and institutional contexts shape which strategies are viable?
  2. Whether, what, and how do technologies mitigate or amplify the barriers Chinese neurodivergent individuals encounter?