Voices from the Moodle Community (MoodleMoot Estonia 2026)

Hi everyone. Um, good morning. My name is Denise. Um, I'm a UX designer at Moodle HQ. So, as a UX designer, part of my job is to talk to people. I will talk to students, teachers, administrators, developers um to understand what they need and how we can design better for them. So if you want to be part of this community and help us shape the future, please scan this QR code, we would really love to talk to you and you get to also try new things with us. So yeah, um, as I've seen AI develop across the Moodle ecosystem with all these new tools and plugins, integrations, amazing things being built by the community and across the Moodle HQ teams, a question kept coming back to my mind. So what should stay human and what struck me is that we are past the point where we consider AI as a feature because right now it's bigger than that. It's changing the way teachers teach. It's changing the way we build courses and most importantly it's changing the way students learn. But most of the conversations are about what AI can do and very few people are stopping to ask what AI should do and almost nobody's asking the community directly. So I decided to and that's why I conducted a series of interviews with people in the mule community whose work I genuinely admire and respect and some of them are even here in this same room. So I started and talked to Marcus Green who's been building um one of the most widely used AI plugins in our entire exist. How cool is that? I also talked to Hiy from University of Finland of Helsinki sorry who's been studying and researching how AI changes the way we build software specifically how we build with Moodle. I also talked to Essio from Alto University who oversees their model based learning environment and gets to shape how these AI tools get rolled across alto universities. And if that wasn't amazing enough, I also talked to people in Bulbaria, the people from Vex. Um, don't tell the Ministry of Germany of Education of Germany that I said that, please. Um, and they're deploying AI tools to 1.8 million students across more than 6,000 schools, which recently I learned that probably that's bigger than the population in Estonia. How interesting. And I also talked to Andrea and Davidid, two developers who build the actual connection layer between AI providers and mood in Italian universities. And last but not least, I talk to people in North America. I talk to all Rachel's who is um retired teachers bu building plugins with the help of AI every day. And I also talked to um Jennifer and Jansen, two classroom teachers. And these people are experiencing AI in their classrooms whether they asked for it or not. So 11 voices, five countries, and I asked them more or less the same four questions. And before you think any of this was scripted, none of them got the questions in advance. So what you're about to hear are the real honest and unscripted reactions. Um some of them couldn't be recorded because of technical issues and some of them because of preference. So for those voices, you will get to hear their words read by an AI voice, which for a talk about what AI can help with, felt kind of right. So I collected hours of conversations and distilled them down to the very essence of the responses. what you're about to hear are there um yeah are these people's thoughts and what I discover about what this community thinks uh should state human and what AI can genuinely help with and remember all of this defined by the people who build with AI teach with AI and make decisions about AI every day we're going to start understanding how these people um understand AI's role, not the hype, but how it affects their daily work. And already here, we'll get to see some tensions that I wasn't expecting. And then we'll get to hear what should stay human for them. Um what are these people protecting? And to be honest, this turned out to be the most emotional part of the research and turn deeper than I expected. And then we'll get to hear what excites them. What do these people think it's possible if we got this right? And here even the most cautious of the voices had the clearest vision of what AI can help with. And finally, I would like to share with you what's happening on the ground. Um the real stories, the we real guard rails, the real um practices happening around the classrooms or around the repositories. Some of these stories would be really inspiring, but some others heartbreaking. So, let's get to it. I started every conversation by asking everyone when you hear AI in education what comes to mind and this already was a complicated question. There's so many ways where AI is used and can be used in education that that sentence doesn't have in anything at this point to me. I think my large based AI is like it's like a partner learning system and you can interchange that enhances your learning and worst case scenario something that you all your your efforts to which prevents you from learning >> um I'm just using it more like a uh as if it's another colleague to just check or maybe to brainstorm and then take it from there I'm more experienced if I can decide to use it or not >> so by asking what does AI in education means to you I got a lot of different answers. Of course, we know now that AI in education, that phrase alone is too broad to mean anything at this point that AI is like a aspiring partner or that it's like a colleague that you can check your work with and that it's safe in experienced hands but can be dangerous in inexperienced ones. And what's great about this is that it reflects what each of these people cares about. An administrator needs precision, researchers is adality, and teachers is risk. And there's a lot to unpack here, but I will also like to dig into one idea that came across every interview, trust. Because once these people started using these tools, the trust question got very concrete. If you were to go to chat and say I just made up this joke who's there and then you say is this a really good joke I think you'll find it always says it's a good joke because it's designed so teachers need to be aware of that tendency of these tools and regards they need to try to work against that tendency >> and I would like us to take a look again about what Marcus said. Thank you Marcus. Um because apparently your aspiring partner is designed to tell you that you're doing great because it's designed to flatter. And when it doesn't flatter, it can fail in ways that aren't always obvious. Like a teacher who told me about wanting to make quizzes for her science course. And everything went fine until she noticed a pattern. two months ago and every single answer on the list will never return it back to me. And I asked why and it was like oh you know I have skewed into trying to do the the answers so that they're not a b weird thing that popped up that I've never seen before. >> The problem with these is not that they're inaccurate. The problem with these tools is that most of the time they are accurate and therefore people think that they're always going to be accurate and that >> so these tools um need a human to detect the failures. But here's what that actually means. It means that we need to treat every output with suspicion. Even when it's right 90% of the time, especially when it's right 90% of the time, because that's the moment that you stop checking. And the moment you stop checking is when someone comes and clicks the save button without looking. And even if you stay on top of everything and stay vigilant, these things, these tools keep changing underneath you. Two weeks of holidays, you're thinking what happens now. It's fascinating. >> Sorry. And then for European institutions trying to keep up with that pace, there's a structural problem that most of AI conversation ignores entirely. I think no other company offers the same reliability and power of service that US companies offer. So we are forced to use the US company. We are forced to use models that doesn't provide all the kind of features or quality that commercials keep us uh in many senses. >> So when it comes to understanding AI and trusting AI, there's a pattern. We know that AI is designed to flatter. We know that AI fails silently. And the real danger lies um when it's right often enough to make you stop checking. And that's why it's safer when a human is paying attention. Um, but there's a deeper problem underneath that because even when you're paying attention, the the field is shifting and changing um at a pace that it's hard to follow. And for much Europe, you're either forced into using US tools or forced into using tools that are limited because these models weren't built for your language, for your region, or for your laws. So, we're left with a harder question. Given everything that we can control, what are we refusing to give up? What are these people protecting and from what? And this is where the conversation turned quiet because when I ask what should state distinctly human, what is something AI should never touch? The answers came slower, but they also came from deeper. They just type in music. I think that's the main relation to what you're creating. And we also do not want to remove the social experience and the individual flavor. the teacher adds to a course. >> Teacher has a responsibility and you can't that responsibility to and if you do then then the whole of educational context in general because I think teacher always has to have the responsibility for learning and the wellbeing of their students. So the most important thing right now or one of the most important things to teach to teachers and students is to not trust AI. So always what AI said, what AI generates. >> So so far we're hearing that people are protecting ship. Uh they're also protecting the individual flavor that a teacher adds to a course. They're protecting the ability to question the output and we're also protecting the idea that responsibility should stay with a human. But what surprised me is where the conversation went when I pushed deeper. Because when I asked what makes teaching meaningful in a way that AI can't replicate, I didn't get a principle. I got a student named Billy. I know that really struggled with this concept and he finally gets it. I can relate. Oh my gosh, this is great. So game day after school. I'm so glad to see that you got this work done and that only build. So that that is not data and definitely that is not a student record. That's months and months of being in a particular room with a particular student. And for the youngest learners, one voice drew the sharpest boundary that I've heard of the entire conversations. So, two teachers, different classrooms, different ages, but they're kind of protecting the same thing. Jennifer is protecting what happens when you get to know a student well enough to see them not their grades but them. And then Jensen is protecting the environment where that type of knowing can even form. And underneath all of this um a new well another thread that run into every conversation data. So who owns it and who can see it and what happens when it leaves the room? Our data is our training other model. They don't send our data and they don't make aggregated data. >> AI is obviously not sharing that information beyond feels like a perfect violation. >> And it's not just about what data leaves the room. It's also whether the systems that we use are being honest about what they do. decision. We are using other way around. >> AI data is being sent to it. >> So what are we protecting? Um, again, we're protecting other ship, not losing the relation to what you're creating. We're protecting the relationship with Bailey. We're also protecting the ability to judge what AI gives us. And we're protecting the idea that responsibility should stay with a human. We're also protecting data and how our in our systems work with honesty. So different words defined by different rules but they are all circling to the same idea that the human has to stay in the loop as the author as the judge as the one who takes responsibility and for the systems that we build. We need to protect uh our data and we need to be honest and transparent about what our systems can do. Um this is what I meant when I mentioned that this part of the research went deeper than I expected because I was asking looking for principles but then people gave me names they gave me boundaries. They gave me lines they're drawing for themselves. This is not theoretical. These are things people are holding on to right now. But holding on doesn't mean standing still. So what do these same people hope it's possible? And here's what surprised me because the same people who just drew the sharpest boundaries also had the clearest vision of what AI can help with. So when I asked if AI collaboration worked perfectly, what would it look like? The first thing that came up was about the students. >> The biggest opportunity is to get more and more students into an active role in lessons. Most of the time today is spent being passive by just listening, reading or watching something. And even the most cautious voices of the research, you remember that teacher who just told me no AI in elementary period. Well, he also had a moment where AI felt unambiguously right. He told me about how one of his students use AI to create music. Not to skip a step, not to cheat, but to actually create something. And if students become more active then what happens to the role of the teacher? >> The role of the teacher will switch more. >> But the real value of the teacher became to design learning experiences setting meaningful tasks guiding critical thinking. and also helping students use AI tools in a responsible way. >> You want to have AI do those minding repeat. At that point, the teacher is relieved from the tenure of having to spend those hours. >> And once the teachers get these hours back, something else becomes possible. something that education has been chasing for decades. >> Personalized learning help provide content that certain students need more students less. At the moment we have education for this learning but they are so complicated to teach. So again, we're hearing what's possible. Active students, teachers are designers, mentors, um relieving the tedium from the teachers and the possibility to create personalized learning paths. This is not a technology vision. This is a teaching vision. And this come from the people who just told me what AI shouldn't touch. So the ambitious and the boundaries aren't in conflict. They are two sides of the same instinct. But then if teachers are switching to mentors and students are becoming more active. Um how do we know what students are actually learning? And this is what where people got very specific >> an AI practice tool that generates questions. all course materials that lets students write draft answers and gives focused feedback highlighting what is correct and what is missing. The goal would not be to give a grade but to help the student understand what they need to improve. So not a grade but an understanding of what to improve and that it's a meaningful distinction and it's not how u it's not only how we give feedback it also changes the way um assessment looks at in the first place. >> The evaluation is shifting from evaluating the final product to evaluating the entire process. Ask the students, okay, you generated that report. What was the prompt that you used for the AI to generate the prompt? What materials did you provide the so to understand and assess the process? >> And underneath all of these visions, active students, teachers as designers and rethought assessment process, there's a question that I will also like us to ask. What should AI ultimately free people to do? >> We want people to come to campus and and meet each other and AI can help help in learning. We understand what we are actually teaching. Some tasks done by AI and some done by people in the room. It's not the threat, it's the understanding. >> So the vision uh what's possible? What do these people think is possible? We have active active students, teachers, our mentors and designers. The possibility to create personalized learning paths um rethought assessments that follows the entire process and then what happens in the room can stay human. But if we look at it, none of this is new. This vision has been articulated for decades. What's new is that AI might made it possible at a scale. So the vision wasn't really the hard part, right? Then what is and this is where the conversation turned honest when I asked people um what are you using AI right now? What surprise you? What's working? This is what they told me. The take up of the use of these AI tools by the teachers is very low. I'm a bit less concerned for two reasons. One, that's the nature of teachers. Teachers just want to do their job. Don't be surprised if they don't rush to it. Don't be worried that they rush to it, but make sure you monitor their use so that you find out why. And part of the reason that they're not rushing is because we're asking teachers to become something that they never really sign up for say that teachers shouldn't have to be prompt engineers. So I've been building a way of sharing prompts prepared prompts with good characteristics because why should a kindergarten teacher or or a maths teacher learn this thing that >> No, it was definitely trial and error. I very quickly learned that the more specific you are, the better the results you get. >> And then the teachers who are figuring it out, they're becoming the most cautious, >> especially the ones very experienced in digital teaching, those teachers tend to be very hesitant and very thankful for the amount of control. >> And meanwhile, in the other side of the classroom, we have the students. students know better how to use some tools than than the vast majority of the teachers and this is the great opportunity that we face today. So what's happening right now? We know that teacher adoption is low and when you dig into why it's not mystery teachers aren't prompt engineers. Nobody prepared them and nobody taught them how to use these tools. So they're figuring it out. trial and error in their own time. And then the ones who figure out, the ones who become experienced, they're being the most cautious because they understand the risk. And meanwhile, in the other side of classroom, we have the students who are getting ahead of the teachers. So remember the vision that we just mentioned where students are becoming more active and teachers as mentors. Well, guess what? It assumes that everyone is ready and they're not. Not because they're resistant, not because they don't want to, because nobody prepared them. And honestly, this isn't just about teachers. Nobody in this room got a transition manual for AI. We're all figuring this out together in real time at the same time. Um, and I would like to share also one more story of what's happening in the classrooms. >> A lot of teachers, especially in my building, are going back to close the computer. Everything's going to be paper pencil. >> So, when you don't know if students are learning, you remove the technology. This is what's happening right now. But there are also other stories of AI that can be really inspiring and that look different than expected about one teacher who was teaching prototyping and students couldn't do a coding. So they do the coding with AI so they can focus on the prototyping. So AI has helped to balance that content in actually what they want to learn. So in this case, the course became more itself because AI removed the hard part and some people aren't really waiting for the rules. They're just putting the rules into the tools themselves. >> Teachers will have access to the students prompts. This allows them to discuss what were you typing in here. Did you believe the response? >> So this team is building the wall drills into the tools themselves. But tools alone aren't enough. Someone have to prove responsible work >> and and we try to use some democracy and training for the AI just to prove that you can use AI in a decent way. >> So none of these stories are about AI arriving. They're all about people deciding. The teacher decided that AI could help with the part that wasn't important. The big team decided to put the word rails into the tools themselves so teachers can see what students are typing in the chat bots. And then uh Davidid decided to prove responsible work for these tools. So nobody had to decide for him. And I want to close um with something that I haven't been able to stop thinking about. A teacher who told me um seeing how her students are writing into the chat bots, not homework questions, but something else entirely. >> I've seen kids typing prompts in. I'm having this problem with my friend Sally and I don't know what to do. And it's just kind of like, oh, let's get you talking to a counselor about that, not your computer. It's a kind of a double-edged sword. It's yay, good for you that you're being proactive, but you're sad that you're still not doing the human connection thing. You're going straight to your computer. >> So, what's stay human? I don't know. And none of these people also have like a clear idea. But we know that AI is a sparing partner. It's not an authority. And the minute of we treat it as one, we lose something. That language problem is real and that AI is not open source for everyone equally. We know that responsibility cannot be outsourced and a human needs to stay on the loop paying attention. And then we know that experienced teachers are the most cautious and that's a signal that's not a problem. And then the biggest risk isn't being replaced by AI. The biggest risk is uh leaving anyone behind either teachers or any of us. So um Marcus said something in our conversation that portrays the Moodle opportunity better than anything. But the thing that Moodle is doing that other people haven't done is give you the keys to the castle. You can choose the external system that you use. You can turn it all off. Just turn off the AI until you're ready. Just turn it all off. And we continue that openness by giving you full control. >> And then Hike left me with a challenge that I will also like to leave with you. It's very easy to always use the latest model with the best capabilities instead of looking at long-term developments and what kind of the model use is sustainable. So that's the challenge for the model community. >> That's a big challenge. Um so I've been sitting with this re research for months now and idea ke an idea kept coming back to me that nobody no if no one of these 11 people said directly but I realized that the decisions about AI aren't really about AI they are about what we want AI in education to be for whether we treat AI as an end itself to make things cheaper faster more automated or whether we created for a more fundamental mean which is the flourishing of every learner. That's the choice underneath every other choice and nobody gets to make that choice alone. So you're here because you partner with Moodle, build with Mood, deploy Mood, teach with Moodle and you will also have your own version for this story. So I would like to leave with you these questions. what should stay human um what can AI help safely help with and how can Moodle support this collaboration. Thank you so much for being here. Thank you to everyone who gave me their time. This talk belongs to them but now it belongs to you. >> Thank you Denise. Okay. So, we have a lot of time for questions. So, please raise your hands and mic will be given to you. Do we have Oh, okay. Uh thank you very much for an awesome talk. That was uh very eye opening and I might say I notice a few uh inongruencies in some of the ideas that are presented. And one that strikes me as a a big inongruency is the idea of say letting the AI create learning paths for our students but then keeping the teacher in the loop as uh you know the the the deciding and and guiding and curating the content for the students in their learning experiences. How do we actually reconcile these two seemingly quite opposing ideas? You have any ideas for that? That's a really good question. Um I think um at the end it's it's a thing of collaboration and get to know until what point you lead AI. Um as Marco said, you have the keys to the castle. You know what to do. You know where to put the control. Teachers also should be responsible if as they build the relationship with students. They should be able to determine which is the best way to do this. But I don't know if I can answer your question, but I hope I can at least u give you something. >> Well, thank you very much. Yeah, it's it's just that's always been nagging the back of my mind for about the last year or so. Um and also there seems to be an abs acceptance of the idea of an AI creating a learning pathway for our students but the AI itself interacting with our students is something we seem to shy away from. So I'm sort of feeling that that's a bit of a surprise. Did you notice any of your participants not wanting the students to directly interact with the AI or did were they all more open to that sort of an idea? No, I think uh students are all up to AI. They're probably they're not using it correctly and that's why we need teachers to be um teaching critical thinking which is going to be the most important thing in the future from now on. And that's why rethought um assessment process because we need to evaluate everything not just the output. we need to be uh every step of the way involved of what's happening. And I think it's it's just the way to integrate AI. It's it's just going to change the way we learn and the way we teach and and especially for teachers. There's there's a big responsibility there. >> Thank you very much. >> Okay, maybe more questions. >> Thank you very much for your presentation. You raised a lot of uh food for thought for me. uh regarding digital competence of teachers. Technically we should uh be skillful, we should have knowledge and the right attitude about artificial intelligence. But so far I have more questions than answers. how can I use it in meaningful way? And uh and a lot has been raised about threats and uh a little less about vulnerabilities of uh AI integration and even less about the consequences what uh eventually it it would bring. And during your presentation, I was just reflecting to what I can can uh associate uh AI as such. And to me it looks like it's it's a kid uh and the father is a IT programmer and uh the mother is education and they speak different languages and this kid is like learning from both and just playing somewhere in neighborhood and uh eventually we know that this kid is going to grow up teenage years and adult ages and yeah from from TED talks you can learn that eventually it's it's going to be super AI and and what's going to return back to this family. So the the sooner we understand how we can talk to this kid who is still learning, the better will be the outcomes and and the less threats and dangers and and terrible consequences. Thank you. >> Thank you. Thank you for your thoughts. >> Okay, maybe more questions. We have still time one from my side. you are a UX designer at the HQ. So how this survey and all these answers insights in reflected your work model vision in general because I think it's very important feedback which you get from your users. >> I think it's uh to start a conversation. It's not nothing is set in stone yet. nothing is defined but it's good that in Moodle we're starting the conversation of where we want to go with AI and one of the our goals or our north stars is as I said the flourishing of every learner so even if it's like we don't have the answers it's good that we are have this into consideration that AI should have like a first human approach where a human is always in in the loop and vigilant. So I think it's just like a wakeup call that we need to be there present and paying attention. >> Thank you very much.