Penn GSE AI & Education Symposium 2026
A recap of the First Penn GSE AI & Education Symposium, April 24–25, 2026
There is a familiar ritual in academic conferences on artificial intelligence. Speakers assemble on a stage. Slides advance. Experts explain, to a roomful of nodding professionals, that the future is arriving faster than expected and that we must be prepared. The audience absorbs. The audience disperses. The future, presumably, continues arriving. What happened at 3700 Walnut Street on April 24th and 25th did not follow this script — and that deviation was not incidental. It was the argument.
The first-ever Penn GSE AI & Education Symposium was, by design and by conviction, organized by students, for students, and about something that most technology conferences are perhaps terrified to foreground: the human being. Not the human as user, not the human as stakeholder, not the human as recipient of intelligent systems — but the human as the irreducible center of any serious conversation about what education is and what it is for.
The Medium Was the Message from the First Moment: Kicking off the symposium with some actual kicking!
Before a single panel convened, before a single slide advanced, participants were asked to line up and dance.
This was not whimsy. One of the symposium’s student organizers and lab member, Li Jiang, introduced a collective line dance — instructions projected on screen, Sweet Home Alabama playing — and what began as an unexpected request became something more instructive than most keynotes manage. Attendance was not required. Participation happened anyway. By the end, the room had already learned what two days of programming would reinforce: that agency is not bestowed by an agenda but claimed in the moment, and that the conditions for participation are themselves a pedagogical choice.
Day One: The Existential Condition First
The first day’s opening panel refused the usual conference courtesy of easing in gently. Moderated by Dr. Seiji Isotani of Penn GSE and joined by faculty Dr. Sharon Ravitch, Dr. Ross Aikins, and Coalition for Educational Excellence Director Tomea Sippio-Smith, the session was staged deliberately as a classroom — not a lecture hall, not a stage-and-audience arrangement, but a space in which attendees were addressed as co-constructors of knowledge rather than consumers of it.
Some people left. This is worth saying plainly, because it is also part of the story. A format that demands active presence will not retain everyone, and the organizers made no apology for it. Those who stayed did not sit back down into the familiar posture of conference attendance. They remained — across both days — as participants in what one might fairly describe as a democratic knowledge-building project, unusual in its seriousness and in the degree to which it treated its attendees as capable of genuine thought.
The content matched the form. The panel moved immediately into questions that most ed-tech events treat as background noise rather than central subject matter: AI and morality. Climate literacy and its entanglement with algorithmic systems. The invisible institutional logics by which schools and organizations approach AI not from curiosity or vision but from fear. This last point was given formal weight with the introduction of the AI Fear Index, a new initiative that attempts to name and measure what most policy conversations prefer to leave vague. Fear, perhaps, is not a bug in the adoption of new technologies — it is often the most honest response to them, and the most suppressed.
The day also featured a session on XR in education, led by our very own Luis Gaitan, whose work in the lab represents one of the more rigorous attempts to ask what immersive environments can and cannot do for learners when the hype is stripped away.
Day Two: Whose Intelligence? Whose Students?
If the first day established the existential stakes, the second day moved into contested terrain with equal seriousness — and with a lineup that Gabriela Gambi, Public Policy and Innovation specialist at the IDB, correctly noted something was historically unusual: our main Plenary Panel about Tech was composed entirely of women.
Sarah W. Newman of Harvard’s Berkman Klein Center, Lydia Logan, Vice President for Social Impact and Education at IBM, Gabriela Gambi, and Dr. Janice Gobert of Rutgers GSE brought together perspectives that refused to stay in their assigned lanes. The conversation moved from classroom practice to workforce development to the most fundamental definitional questions: What is artificial? What is intelligence? When we speak of AI in education, whose AI are we discussing, whose teachers, whose students, and in which part of the world?
This last question deserves to be underlined, because it is the one most frequently treated as a footnote in conversations that take Silicon Valley as their default geography. The panel made clear that the global distribution of these tools — their availability, their cultural assumptions, their language capabilities — is not a technical problem awaiting a technical solution but a political and ethical condition that shapes every claim made about AI’s educational potential.
The Gallery: Twenty Answers to Questions Still Being Formed
Running alongside the programmed sessions was a student project gallery featuring our ongoing student research projects — a format that can easily become perfunctory but here functioned as a genuine intellectual exhibit. The work spanned adaptive learning systems - Intelligent Tutoring Systems, AI education laboratory projects, bias and algorithm auditing, AI in teaching and learning, and the intersections of language, information, and literacy.
What distinguished the gallery was that it was not a showcase of finished solutions but a documentation of live inquiry. Attendees did not observe from a respectful distance; they interacted, questioned, and challenged. This is how research ought to be presented to a public — not as concluded findings delivered from authority but as ongoing work that remains genuinely open.
The Nonprofit Voices: Human Flourishing as the Brief
Among the most grounding contributions of the symposium were two voices from outside the academy. Lauren Flood, Public Relations and Communication Lead at Subject to Climate, and Eric Shepherd, President and Executive Director of the Foundation for Talent Transformation, each brought perspectives rooted in specific human missions where AI is neither the subject nor the solution but a potential collaborator.
Shepherd put it with a directness that academic discourse sometimes talks around:
“Working with your students, both at the table and during the workshop, was a highlight. It was especially interesting to see them interact with our AI coaches in Chinese and have the responses come back seamlessly in Chinese. I hadn’t seen that in action before, and it added a new dimension to the experience.”
The editorial note worth adding here is that Shepherd was not expressing amazement at the technology. He was expressing something more significant: surprise at encountering students who already knew how to make the technology serve human ends across cultural and linguistic boundaries — who had moved, without fanfare, past the phase of being impressed by the tool into the harder work of using it well. That is not a minor thing. Most institutions are still in the impressed phase.
The Research Layer Work
The symposium’s intellectual depth was further demonstrated in the contributions of doctoral candidates and faculty whose work sits at the most contested edges of the field.
Daniel Noh, LST Doctoral Candidate, along with lab member Sijie Mei, brought sustained attention to education justice — to the question of how technology either illuminates or obscures the elusive evidence of genuine learning and understanding, concepts that have resisted capture for as long as people have been trying to capture them.
Dr. Amanda Barany of Penn’s Center for Learning Analytics invited the room to consider AI as a partner in qualitative research — a proposition that challenges one of the more persistent myths in the field, which is that computation and interpretive human inquiry are in fundamental tension. Her work suggests otherwise, and the implications for how we understand learning at scale are significant.
The AI Instructional Games offered a solutiuon: evidence that core computer science and AI concepts can be taught without reliable internet access. In a field that sometimes confuses access to tools with access to ideas, this is an important corrective.
An invitation to start
A recurring conversation about AI has been about what the tool designers say they value and what they actually delivered. The Penn GSE AI & Education Symposium was a small but genuine attempt to question that gap, not by declaring it closed but by organizing two days in which the structure itself made the argument.
The danger of powerful technologies is not that they fail but that they succeed — that they reshape the environment so completely that we forget to ask what we traded away. The question the symposium put into the air at 3700 Walnut was not whether AI belongs in education. That question, perhaps, has already been answered by the market. The question is whether education can drive AI — whether the values, the slowness, the uncertainty, the irreducible presence of a human being trying to understand something, can survive contact with systems optimized for speed, scale, and certainty.
Two days: twenty posters, twelve breakouts, two panels, one line dance. We don’t have a conclusive answers, but perhaps this could be the beginning and the symposium is an invitation for us to start thinking and having that critical dialogue.
And the work continues.
Nam Nguyen.