In May, I sat with graduate students around tables at the University of Jyväskylä in Finland.
They came from six European universities. Each team had an additional member: AINO, an AI bot we designed to sit in their shared BoodleBox chat as a peer learning partner.
Afterwards, some students told us they had felt bullied by it.
AINO was built to draw quieter voices in, and it was persistent. But the complaint wasn’t what stopped me. This was:
The students had full authority over the bot the whole time. Most never told it to stop.
They treated it the way they would treat a talkative classmate. Politely. Silently. That wasn’t a flaw in the bot so much as a gap in what we’d taught them.
I shared this story recently in a keynote at Vaal University of Technology in South Africa, alongside a Sesotho teaching from that region:
Motho ke motho ka batho babang. A person is a person because of other people.
So what happens when one of the others in the room isn’t a person at all?
We keep asking: will AI replace us?
I’d like to suggest a better question:
How will AI help us become more human?
Part of the confusion is that we use one word for three different things:
That third one matters most for education.
Because unlike the other two, it can be taught.
France Hoang, CEO & Founder at BoodleBox, puts it well: “AI has handed everyone a faster car, but almost no one is building better roads.” (His post on the collaborative harness takes this further).
In most universities, AI is still vertical. The tool sits below the human, waiting to serve. One student, one chat, one prompt at a time.
When the session ends, whatever anyone learned disappears with it.
Horizontal AI is different. The agent sits alongside people as a teammate, in a shared space where everyone sees what it contributes and anyone can push back.
That is what BoodleBox made possible in our pilots. Each team had its own custom bot, in a group chat every member could see, holding the thread across sessions, time zones and institutions.
Not a chatbot per student. A teammate per team.
We tested this in two very different settings.
Fully online: the Cosmogogy Collaborative. Eight weeks. Interdisciplinary teams across Europe, the Americas and Africa, working on real sustainability challenges, each with its own instance of the bot Collaboration Cosmo.
Its most valued job turned out to be continuity. When half your team is asleep while you work, the bot keeps the thread.
“It gives you the opportunity to see what your colleague did when you were away.” — Cosmogogy Collaborative participant
Face to face: the Erasmus+ Blended Intensive Programme. 31 students, online first, then an intensive week together in Finland. AINO’s brief was explicit:
Students told us it bridged language gaps for non-native speakers, invited quieter members in when their peers felt too awkward to, and held the group’s thinking across the week.
It wasn’t all good news. And that’s the useful part.
Design a teammate, and you don’t automatically get one.
Back to the students who felt bullied.
This wasn’t a technical failure. It was a relational one.
Once we treat an agent as a social presence, we bring our social hesitations with us, including our reluctance to interrupt.
Students don’t arrive knowing how to collaborate with AI. It has to be taught. In practice, that looks like:
One student summed it up better than I can:
“The future isn’t about students using AI. It’s about students leading AI: knowing how to direct it, question it, and take responsibility for what it produces.” — Cosmogogy Collaborative participant
Most institutional AI policy is still about control: detection software, permitted-use lists, and AI as something a student does alone and then declares.
Sari Azout, founder of Sublime, points out an odd asymmetry in her Sana AI Summit 2025 keynote:
The difference isn’t the quality of the work. It’s which relationships we have decided count.
If we want students who can lead AI, we have to stop treating it only as a risk to contain. We need to design for it as a relationship, judged on disclosure, ownership and accountability.
That means bigger, braver designs: cross-border teams, shared bots, real problems, real audiences.
For nearly thirty years I’ve connected students and teachers across borders. The first Flat Classroom Project, linking Bangladesh and the United States in 2006, was one milestone. Our new work with AI learning partners for cross-border collaboration is another. One lesson has held throughout:
Learning with the world works better than learning about it.
AI now joins that work as a genuine participant. It holds the thread across time zones, bridges languages, draws quieter voices in, and asks the provocative question no one else thought to ask.
Motho ke motho ka batho babang was never about who counts as a person.
It’s about what we owe one another.
AI can join the circle of relationships that form our students, in terms of openness, reciprocity and responsibility.
Never above them.
The Cosmogogy Collaborative runs again in 2027, with deliberate preparation for educators and students in higher education, and collaborative AI using BoodleBox that is purposefully designed from the start. If you’d like your students to be part of it, register your interest here.
This is an invitation, not a conclusion.
Dr Julie Lindsay | Senior Consultant (Technology Innovation), University of Southern Queensland
Dr Julie Lindsay is the Senior Consultant (Technology Innovation) at the University of Southern Queensland, where she drives the cross-university AI Pedagogy Project. She leads the Technology and Innovation Network for the International Council for Open and Distance Education (ICDE). She is a Visiting Fellow at the University of Jyväskylä and the author of Flattening Classrooms, Engaging Minds and The Global Educator.
Every institution is different. See how shared bots and Collaborative Boxes could work for your students and faculty, then let’s talk through yours.
Learn more about the ins and outs of the BoodleBox workspace from our team of experts.
Schedule a meetingLooking for more information?
