Chris Colley: 0:13

We're here at ShiftED podcast, I'm reaching to a fellow Montrealer from TELUQ University, which is our online university here in Quebec, and Valérie Pichet is an expert in AI and teaching and learning, and we're going to ask her some questions today about the transformation of our schools with AI. It's been happening a lot. So, Valéry, merci, thank you for joining me today. I hope you're doing well.

Valéry Psyché: 0:54

Yeah, thank you so much for inviting me and yes, I'm doing pretty well actually.

Chris Colley: 1:01

Excellent, but we'll go nice and slow and have a cool conversation. Before we start, valerie, I'd like to ask the guest, kind of just as an introduction, what were some key moments that brought you to where you are here today. In education, what were some schools you went to or people you met or experiences you had that helped you get to where you are here today?

Valéry Psyché: 1:26

yeah, that's a very good question. Actually, that's uh, there's some very important key moments in my, in my life, the, for example. I can just talk about some of them. Uh, I started my study in the in in France, in Grenoble. I was doing the equivalent diploma of a bachelor and a master. And then there was this possibility to go to Quebec to have just an exchange, student exchange for six months and to study at Université Montreal for this time, to learn a bit how you guys teach and to have some courses in international context. And then I decided to stay there and to continue my study in Quebec because I enjoy the people, the life.

Valéry Psyché: 2:25

And then I was doing at this time I was studying physics, which is very far from education, but after my so I did another master and then I decided that I wanted to do a PhD.

Valéry Psyché: 2:43

But I wanted to change my orientation and there was this new program it's about computer cognitive informatics at University of Quebec, which is artificial intelligence joined with cognitive science, and so it was very interesting program, very uh, innovate, an innovation in a innovative program actually and uh, I decided to, to jump in this and I never regret. So I had uh, two directors, so one in computer science and another one I chose in education. So that's where I started to uh, to uh. You know, like uh study, not study at pages, more like research actually. Um, but we have a couple of courses to have to align our knowledge because we had to be uh uh aware, not not only about computer science if we're coming from computer or technical science but we have to be aware about what was the main concern of people from humanities. So it was very, very interesting. And then I started to be interested in everything that was based on AI in education and technology, enhanced learning and all that stuff.

Valéry Psyché: 4:15

So yeah, that was a shift in my career in my career, my career.

Valéry Psyché: 4:25

I never changed. And I was my co-director co-supervisor was in TELUC, so that's where I discover distance learning as well.

Chris Colley: 4:41

And your experience with distance learning. How do you find it different from classroom teaching Like? What are some of the main differences that you experience when you're preparing a class for online versus, maybe, preparing a workshop for a live in person or doing a lecture? What do you find the big differences?

Valéry Psyché: 5:01

It's quite very different and, to tell you the truth, for many years I couldn't conceptualize teaching or learning without thinking about distance learning. So it took me a while to explain more to people who are doing just teaching in presence what were the differences. But there are quite some big differences. For example, for us especially, what we're doing is asynchronous distance learning, which means that we don't do synchronous, we don't do like a webinar to teach our students. So all the material has to be prepared in advance and it has to be prepared in a way that will be engaging and motivating for the students, because they will be alone when they will be doing the following their course. So it means for us a lot of preparation. So doing a teaching for us it's doing learning design, and after that, when the learning design is done, so we implement it in a platform, an online platform, and then we do the same thing as people do in the classroom. So we evaluate the student, we answer their question and all that stuff. We supervise them. But this part of learning design, for us it takes like a team. This is how we do it in TELUX. So we need, for example I will write all the instructions very clearly we have to think about strategy, pedagogical strategy. It's very important, as well as the activity that we choose, to maintain the student motivated, but as well to capture his attention and to be sure that he will gain some learning. And when we finish that, usually we produce a very big documents could be like 100 or 100 or 100 or 200 sometimes documents, and with all the resources.

Valéry Psyché: 7:38

Of course, all the resources have to be validated, which is very important when you're doing distance learning in the way we do it, because we cannot violate any rights. So we have to have the list of all the resources of the course and we have a team who just do that full-time validate that we are okay for the rights that we use in the course. We are okay for the rights that we use in the course, you know, and as well. But we need to have somebody, a linguistic expert, because we cannot put nothing that have been well written, because everything is on the web, it's on the platform, so, and we have graphic, a graphic designer and a programmer that programmed the. All it's so one course, it's a. It's like a that programmed all it's so one course it's a. It's like, uh, it's, it's a learning environment. It's just uh, yeah, so that's the way we do it yeah, and do you guys work with moodles at all?

Chris Colley: 8:39

like I was first introduced to that moodle idea last year and we built a course around it. And you're very right about the engagement part, like keeping it so that the students don't get bored or like, oh you know, start looking at their phone or whatever else how? And I always kept asking, like people that had built Moodles before, you know, like online courses where you just yourself pace, it's asynchronous and you have certain chapters you go through and you do certain activities, and I kept asking what makes online courses the most engaging for kids? Is that a fair question? How do you design your course? Because I know that you have to think about that design a lot when you're creating these courses. How do you bring engagement in when the students aren't with you and then also not doing it at the same time? They're all kind of doing it at their own levels.

Valéry Psyché: 9:36

How do you?

Chris Colley: 9:36

create that in your course.

Valéry Psyché: 9:39

We have different strategies. For example, we have a studio actually, so we can't make some video, so we have a team to help us to that. So we prepare some video. We are split some concepts so the student can see us, that he can replay the video they wanted to. And we use as well, for example, some in a notice strategy that I use. Sometimes I invite some experts and I ask them to answer some questions. So we film them so the student can have access of high quality experts explaining some concepts. I use that a lot, explaining some concepts. I use that a lot and as well, interactive activity. But we have to be very careful because we are very aware of using inclusive access for everybody. So what we do is we have to respect the WCAG, which is the web guidance guideline for the accessibility. So that means that sometimes some of the interactive exercises that we want to use have to be compatible with that. So we have some constraints sometimes based on that. Yeah, exactly, so this is a strategy we use, yeah great ideas.

Valéry Psyché: 11:13

Yeah, we change the media so not to just text, video, audio and all that stuff to keep them engaged.

Chris Colley: 11:21

Yeah, yeah, no, that's cool. That's what we did. We just tried different kinds of media and different kinds of small tasks. We made everything pretty short too, so it never felt like it was going on and on forever.

Valéry Psyché: 11:37

No, that's something that we use a lot. It's a broad journal so they can have questions for reflection that they have to fill in their journal and then they do need. So we have different questions that we ask them to answer and it helps them to write their essay. They have something to submit for evaluation. And the last thing that we use a lot we usually have a blog associated with the course but we notice that students sometimes they are shy to interact in the blog because they commit themselves. You know they have to sometimes present themselves or talk about some subject and they are like, maybe shy. So we encourage them.

Valéry Psyché: 12:30

They have some bonus for participating and we encourage them to comment the text that all the students write. So there are some interaction in the course event. It's a very flexible approach that we have in the way we do distance learning. A student, for example, can start whenever he wants, but doing this kind of stuff, at least it's connected with all the students, even if they are not in the same step of the course. There's always been somebody. Actually there will always be somebody that is starting at maybe pretty around the same time.

Chris Colley: 13:18

So yeah, right, right, interesting. And then AI comes into the picture.

Valéry Psyché: 13:24

Yeah.

Chris Colley: 13:26

How is that helped or not helped in designing and building your lessons and interacting with your students? Can you give me a couple of examples of where it really helps you a lot and helps save time, and maybe one or so where it doesn't help so much?

Valéry Psyché: 13:47

yeah, it's a tricky question because, of course, yeah, that's who does a good point there's some disadvantage, but, um, actually, uh, we are facing a lot of efforts uh to to review our courses because of generative AI, because it's been democratized almost two years ago, more than two years ago, in November 2022. Learners have access to this kind of tools and they use it a lot to generate content to help them to write their essay, their summative activities. So this is a problem, a problem of intellectual integrity and plagiarism plagiarism. So, um, actually, we have, we have to review the way we ask them those questions to make more personalized uh, make them to uh doing like, for example, more like, uh, what we call um experience, experiential uh learning, like called theory about. So they have to to contextualize their the answer based on their experience and, uh, this kind of stuff can't help. Or we, we, we have a we have to change the way, the strategy, the way we, we uh inquire them, their knowledge, uh, in order to avoid the plagiarism. So it's a lot.

Valéry Psyché: 15:25

It's really demanding for the professor team because they, you know, like you have to review all your courses, change the question, change the strategies, and but fortunately, we have help from technological experts, techno-pedagogical experts that we have at TELUK, who help them with that. Put in place some strategy, for example, even give them information about how to use AI, what they can do, what they cannot do. You know like to come back, because sometimes they will use it not knowing what they can do what they cannot do. You know like to come back Because sometimes they will use it not knowing what they can do what they cannot do, but if the rules are more clear.

Valéry Psyché: 16:14

The clearer it is, the better it is for everybody. So we start to see, in all the universities actually even the ones which are not doing the distance learning, they start to put some some framework about the, the, the acceptable use of AI or what is not acceptable. So, yeah, I cannot say that it will help me. It could help me actually, but I don't use it a lot like this. But I know some teachers will use it to help them doing a planning of their teaching or to be inspired with some ideas, and that's what gives me the ideas of developing a research.

Valéry Psyché: 17:06

I develop actually a tool that is called Ecopadia and this tool has an instructional design, a learning design scenario implemented with AI assisted scenario implemented and with AI assisted. So the teacher who don't know to be followed to do this planning, he could use this tool, for example, to write a scenario with planning. Planning is a teaching lesson, so it's very useful. We use it and we test a lot with different contexts, actually in Quebec and as well in the West Indies, to see how it can be used in different contexts, and we implemented some prompt to help. So we have nine or ten prompts implemented, so we control the prompt. So that's why the teacher don't need to write a prompt, because it's a challenge for them, right? So you know, like a prompting education, it's something new, a challenge for them, right to uh you know, like a printing, it's something new.

Valéry Psyché: 18:26

Not everybody knows how to do it and uh, so we um and they can. After that they can do in a follow-up. So they can. There's a prompt implemented for each part, for example, uh, the part when they have to describe the general information or the pedagogical information or the learning objectives or all that stuff. So there are different programs and if they are not satisfied with the answer, they can make a follow-up inside the tools. So this is one tool that could be helpful for teachers to help them to build their teaching courses.

Valéry Psyché: 19:04

I have done research which is more like oriented to the learner, so we implemented the chatbot in a MOOC. So MOOC it's distance learning courses, but they are not credited, so there's no supervision. There's no um supervision of teacher in the MOOC. So the MOOC is free but it's not credited. So the people, they register and they have to follow the, the, the activities by themselves, but they don't have nobody to talk to if they have a problem or they are they question.

Valéry Psyché: 19:44

At the beginning we had some forum, but when we stopped moderating the forum because it takes a lot of resources to moderate the forum for more people to have thousands of people, we realized that it was not possible to moderate anymore the forum and when we stopped that, so there was no more interaction, even between two teachers, two learners, for example, I would say. So what we did, we added, we tested chatbots, ai chatbots, but, as you know, chatbots sometimes they give false information, so it's not very useful for students. So what we did? We integrated some technology inside to control the way he behaves and to control the way he answers. So there's a different way to do that and we have tested.

Chris Colley: 20:41

So that's a personalized tutor, basically what you were building.

Valéry Psyché: 20:46

So if a student felt they weren't getting it, they could ask more questions, yeah yeah, yeah, I see it like more a learning companion, because a tutor, intelligent tutor yes, sometimes they call it. People associate charge GPT as a tutor, but it's not an intelligent tutor. If you know about the technology of intelligent tutoring system, you will understand that chat GPT is far from an intelligent tutor. Intelligent tutor it's an environment that can give feedback, yes, for the student, like Chatbot will do, but they will do it with the knowledge of the learner. So that means that they've been programming the way to follow step-by-step what the learner is doing and adapt itself, not just to the interaction, the conversational part, but as well other parts, for example, change strategy, pedagogical strategy based off cognitive diagnostic of what the student is doing. For example, they will adapt, showing different contents based on an error that the student will do.

Valéry Psyché: 22:22

So it's really more adaptive that chat GPT can do. Chat GPT is really dialogue and we try to make it more like a tutoring system, like we're doing, like, for example, we had a technology called the RAC which is a multi-generation technology which allows us to feed the chatbot with the knowledge, with the knowledge, the domain knowledge, like we do in the intelligent tutoring system. So we feed him with okay, you have to know all of that, to answer well to the student, instead of choosing some information out of the way, so you orient it towards the source knowledge, rather than it going and picking wherever its model is Exactly.

Valéry Psyché: 23:11

And, as well, we tell him how to behave as well. So that's true that it's another part that we do in the intelligent storytelling system. It's tutorial expertise. So we tell him how to behave. To do that, we will implement some pre-programmed prompts to tell him what is his role, for example, and how he has to behave with the student. And if the student did that or that, this is the way he has to behave. So maybe one day we will have some kind of intelligent tutorial based on generative AI. The missing part is more like the learner's model. So actually, for the moment, we don't really have a learner's model. So actually, for the moment, we don't really have a learner's model. We don't know what. We don't register the good, the good points, the bad points, the good, you know the error, the misunderstanding, all that part. And Chatbot doesn't react based on this misunderstanding for the moment. But it will come for sure. I have no doubt about that. Right right.

Valéry Psyché: 24:32

Yeah.

Chris Colley: 24:32

Somebody told me too, valerie, that that's the power of AI to come, that with whatever is being inputted, it can give you some suggestions back. For example, you know I had a history course. I was confused. You go into the chatbot to try to clarify and then the chatbot will tell the teacher about where the misconceptions were, so that the teacher can adjust his or her teachings based on the information that the student's feeding into the chatbot or the query that they put in. Is that something that TELUS is looking to as well, so that the teachers can get some of that information to help inform where the students are getting the info and where there's misconceptions or they don't have an understanding of it? Is that coming down the road, you think?

Valéry Psyché: 25:26

Yeah, yeah, yeah, that could be if we For the moment, yes, so the way it works, it's like if we register, if we keep, we stock all the information, the interaction, so we call it educational data learning analytics. So if we make some learning analytics based on the interaction, the trace that students keep in the system, we can effectively improve the correct misunderstanding of the learners and, as well, improve the learning design of the course. So the next, we can improve the teach for sure.

Valéry Psyché: 26:19

But we have to do this. We have to do the learning analytics part and it's with some questions, some challenge, because some institutions they are very, you know, they don't feel like they. There are some issues about privacy and data privacy, confidentiality of student data. So not all the institutions are at the same level of reflection based on that. But it would be very helpful if we are more like a standardized way to analyze data and to use them to improve the course. That's why you have more information to overcome dropout and all that stuff.

Chris Colley: 27:14

Yeah, absolutely. All that information is so helpful If it's there too and you can have access. But, like you said, privacy issues are still kind kind of and we're kind of in the youth sector as well, so there's age limits on who can have access and who can't. Where do you see this AI? Going down the road, like if you had a magic ball that you could look through. Where do you think we're headed with AI in education? Oh, there's through. Where do you think we're headed with AI?

Valéry Psyché: 27:44

in education. Oh, I'm going research since many, many years. Ai in education is not new AI education. There's a society, international society, very active in research, and so there's different orientations. I just evaluated some articles yesterday for one of the conferences. There were, for example, people who are interested in how to use AI to generate, for example, this diagram to help in mathematics, for example, to translate some situation, so a mathematic situation, and exercise, and make some visualisation to help the student understand some concept, for example. Other example would be like how can AI can help with, maybe have some framework to cover this kind of special needs, with some tools that can be used to generate some activities that are more helpful for people with special needs, more helpful for people with special needs. Um, another example that I had, um yeah, so I don't have all in my life.

Valéry Psyché: 29:31

I have uh, but but anyway. So that's kind of example I like your differentiation, though.

Chris Colley: 29:34

I think that I think that that whole um thing that we have within education, where kids are all learning at different levels. It seems like AI can kind of level that playing field a little bit, where you can adjust certain things for certain students, but they're all still learning the same stuff.

Valéry Psyché: 29:53

Yeah, it's what we call the adaptive learning, so the tools adapt to the need, to the different needs. So usually this is what we can do with artificial intelligence in education to build some kind of system that, based on the misunderstanding of the student, can provide different kind of exercise or a different path as well. So maybe the rhythm is too quick, so to make it more slow, or to show a lesson that when it's needed, because it isn't a concept, or to give an exercise that is more like the same kind of exercise, same level, but with another example, another explanation, another way to about this concept so that the student could understand, or to make it more contextualized, based on where the person comes from, what we call a contextual system or system sensible to the context or the culture, for example. So, yep, so there are some actually different, a lot of things going on with the AI and that community, actually different, for example, tools that have been developed, architecture as well that exploits, exploits, neural network or large language models, and to it's, it's. It's a lot of tools that could be useful for teachers, for learners as well, different kind of public, great, yeah.

Valéry Psyché: 31:49

So I think there's another shift, I think it's an educational shift. I think with AI generative AI that will come. We will see that there's a shift like we cannot teach the same way, shift like we, we cannot teach the same way. And there's a over hyper use of AI by the student and the way you use it, sometimes they will lose knowledge or not acquire some competencies. So there will be it's we are, there's a shift going on that I think that that will address more like integrated generative AI in the teaching. That means to show the student how to use it well, to make to learn things or to make exercise to, instead of don't let them use it and they will just copy past and don't learn anything.

Chris Colley: 32:55

so, yeah, exactly getting the students to know what it's all about. Super important. Um Well, valerie, this has been fascinating. Such great insight. I love your examples that you offered, too, and the future does look bright. I mean, there's a lot of stuff happening and, like you said, there's a big shift going on right now, but I think people like yourself have done research and thought about this and you're making us all a lot smarter. So thanks for sharing some of your insight today.

Valéry Psyché: 33:25

My pleasure.

Chris Colley: 33:27

It was great talking to you and have yourself a great day.

Valéry Psyché: 33:30

Thank you, Chris. Bye-bye.

Chris Colley: 33:34

Bye-bye.