“AI as a Catalyst for Inclusive, Personalised, and Ethical Education: Empowering Teachers and Students for an Equitable Future.”

For Snigdha, whose research focuses on accessibility and inclusivity in AI-driven learning environments, the conference was a chance to situate her work within a broader global discourse on ethical and equitable technology in education.

AIED 2025 was organized by the International AIED Society and hosted by the University of Palermo, Italy. Snigdha’s participation was supported through a conference scholarship recognizing promising early-career researchers in inclusive AI education.

This recognition comes after a long and determined research journey, and we spoke to Darshana to learn more about her experiences and reflections.

A Scholarship and a Mission

Receiving the AIED 2025 scholarship was both an honor and an opportunity for immersive learning. Along with travel support, it came with a volunteer commitment, allowing Snigdha to contribute to the event’s organization while gaining an insider’s perspective on how a major international conference operates.

She spent her first days greeting attendees, helping with registration, and assisting session chairs — tasks that quickly turned into opportunities for networking. On Wednesday, she supported the Early Career Panel, where established researchers shared candid insights into their professional journeys — from the unpredictability of postdoctoral paths to the balance between research depth and practical impact.

“Being a volunteer made the conference feel immediately accessible,” she reflects. “It helped me connect with researchers informally, observe how large conferences function behind the scenes, and appreciate the collective effort that sustains our academic community.”

Inside the Sessions: Learning from the Field

By Thursday and Friday, Snigdha was helping facilitate sessions on instructive assessment methodologies — a topic closely related to her research interests. These sessions opened rich discussions about how AI can measure, support, and enhance human learning without reinforcing bias or inequity.

“The most valuable learning often happened between sessions,” she says. “Quick hallway conversations, helping presenters debug slides, or listening to how they handled unexpected data challenges revealed the real craftsmanship of research — the decisions that rarely make it into papers but define their integrity.”

Through these moments, Snigdha gained a deeper understanding of what rigorous, context-aware evaluation looks like in practice — lessons that continue to shape her own methodological approach.

Presenting LipMOOC: Synthetic Content for Accessible Lipreading

The highlight of the week came on Friday, July 25, when Snigdha presented her paper titled “LipMOOC: A Modular Synthetic MOOC for Lipreading”, co-authored with Vinay Namboodiri and Prof. C. V. Jawahar, during the “Accessibility & Neurodivergence (2)” session.

The paper explored whether synthetic talking-head videos can effectively replace or complement real video recordings in teaching lipreading — potentially expanding access to learners with hearing impairments or limited data resources.

Her presentation described the project’s design framework, experimental setup, and evaluation strategy, while addressing how the model handles viseme variations across languages and contexts. The audience engaged deeply, raising questions about cross-lingual generalization, fairness across learner subgroups, and robustness in noisy or low-light conditions.

“The discussion was incredibly constructive,” Snigdha recalls. “It helped clarify where we need stronger ablations and how to separate the influence of video fidelity from instructional design.”

Several researchers later approached her with collaboration ideas — from testing LipMOOC in new linguistic settings to adapting its pipeline for cross-cultural accessibility studies. These exchanges, she says, “transformed the talk from a presentation into a starting point for shared inquiry.”

Keynotes that Reframed the Field

Among the most influential moments were the keynote lectures by Kristen DiCerbo and Maria Mercedes T. Rodrigo.

DiCerbo emphasized the importance of designing AI tools that are usable and intelligible for teachers — foregrounding classroom realities over abstract technical metrics. Rodrigo, meanwhile, championed context-aware evaluation, urging researchers to measure what truly matters for learning rather than what’s easiest to log.

Together, their talks offered a framework that resonated deeply with Snigdha’s work:

“Align your claims, your measures, and your deployment context.”

She has since adopted this principle as a guiding lens — ensuring that each methodological choice in her projects reflects real-world relevance and educational value.

Conversations, Mentorship, and Community

Beyond formal sessions, AIED 2025’s greatest strength lay in its community. Coffee breaks, poster sessions, and informal dinners became fertile ground for exchanging ideas. Snigdha connected with peers exploring topics like bias audits in learning systems, differential item functioning, and tutor-facing analytics — collaborations that continue even after the conference.

Mentorship, too, emerged organically. Senior researchers shared practical advice on project scoping, managing long-term collaborations, and maintaining a sustainable research pace.

“They shared not just encouragement but operational wisdom,” Snigdha notes. “How to pace papers, how to choose collaborators, how to sustain impact beyond publication. Those conversations were invaluable.”

By week’s end, she left with new collaborators, annotated reading lists, and a clear sense of where her research fits within the global landscape of AI in education.

Learning to Communicate Research Better

One of Snigdha’s key takeaways from Palermo was the importance of clear, transparent scholarly communication. Observing seasoned speakers handle critical questions gracefully — acknowledging limitations while defending core claims — offered a model for her own practice.

Since returning, she has begun refining her presentation style and documentation practices: emphasizing study design before results, annotating figures with precise parameters, and explicitly listing failure modes alongside successes. She is also developing a “threats to validity” checklist to guide future papers and reviews.

“These habits may seem small,” she says, “but they’re the foundation for reproducible, honest science.”

The Palermo Experience

The University of Palermo provided an ideal environment for learning and reflection. The conference venue — Edificio 8, Viale delle Scienze — was thoughtfully organized, with smooth logistics and a welcoming atmosphere.

Outside the sessions, Palermo’s layered history and Mediterranean charm offered moments of respite. Evening walks through the old city allowed space to process ideas and conversations from the day.

“The setting itself encouraged reflection,” Snigdha recalls. “After long technical sessions, walking through Palermo’s streets gave me perspective — it helped me connect the dots and return each morning with sharper questions.”

Looking Ahead: From Reflection to Action

As she looks toward AIED 2026, Snigdha’s goals are both ambitious and collaborative. She plans to:

• Extend the LipMOOC study to include cross-cultural replications and multi-lingual datasets;
• Share open-source materials that support replication and transparency;
• Mentor first-time conference attendees; and
• Contribute to community initiatives on accessible evaluation practices.

She also hopes to co-author future studies with emerging researchers she met in Palermo — continuing the culture of openness and support that shaped her own experience.

“The scholarship that enabled this trip was more than financial assistance,” she reflects. “It was a vote of confidence that carried me through the preparation, presentation, and all the conversations that followed. I intend to pay that forward.”

A Turning Point

Looking back, Snigdha describes AIED 2025 as a quiet but decisive turning point — one that deepened her methodological clarity and strengthened her confidence as an emerging researcher.

“AIED 2025 wasn’t transformative because of one big moment,” she says. “It was transformative because it offered exactly what I needed — constructive critique, credible exemplars, and a generous community. It reminded me that progress in AI for education isn’t only about better models; it’s about careful design, transparent evaluation, and the relationships that make sustained collaboration possible.”

As she returns to her work, Snigdha carries forward not only new technical insights but also the conviction that research excellence and inclusion must grow together.