DOWNLOAD PDF TRANSCRIPT

00:00:03 

[Joanna Szoke] 

So yeah, so the title of my talk is Agents Assistant and Us. And who's this? Who's actually doing the work? Well, that's already. This kind of question is something that I'd like you to ask yourselves while you're listening to me today. 

And thank you for the introduction. I just also want to show you the COVID of the book in case you're interested because I just added that just before the talk to the ppt. But let's get started. I actually want you to imagine a situation. So we'll start with some imaginative activities. 

So I'd like you to imagine that you're in an escape room. I like going to escape rooms if you haven't been yet. It's basically kind of a free time activity. Usually people go together in groups and, and they need to figure out the way to get out of the room. So it's usually a themed thing. 

00:01:03 

In this case we've got a pirate theme and there are all kinds of tasks and riddles and things to solve and that's how you can get out. And if you got out, you can be happy because you managed to succeed. But let's imagine that here's this little agent with you, AI Agent Jack Sparrow, who's very helpful. And basically you don't even have to think about anything. It basically just walks along or like next to you and does everything for you. 

As you can see, I'll do everything. You'll get out in no time. Now how much? And feel free to add your ideas in the chat. How much would you enjoy a game like this? 

So there's this little agent who does everything, helps you and answers all the questions, solves all the riddles. You don't have to even do anything and you're out. So would you enjoy this kind of thing? And the good thing is I again kind of lost my cursor. So I don't know what you're writing in the chat, but I might be able to get it back somehow. 

00:02:23 

[Sheila Moreno] 

Do you want me to read the chat for you? 

[Joanna Szoke] 

No, no, I found it. Thank you. I kind of wiggled around and now I can see and. Very good. 

I'm so happy with these responses that you wouldn't like this kind of game and you would find it boring. Exactly. But we're going to think about it, like think about why and find the reason why you would probably not like this kind of game. 

Oh, somebody actually said students in Asia would love it. Now that's something to think about. So let's focus on our little agent for now agent Jack Sparrow and let's focus on what agentic AI is. So agentic AI or you might, there are these terms, agentic AI or AI agents. They, this basically refers to an AI system or software that can independently make decisions and take actions to achieve specific goals. 

00:03:26 

And the key terms here are autonomy and independent planning and adaptation to a changing environment. So what that means is that you just set them off and they can learn from experience, they can adapt to the environment, they can autonomously solve tasks and they can also communicate with each other. So for example, when you try so called agentic browsers and there are now several of them, e.g. chatGPT Atlas, Perplexity Comet and so on, and you can, you can try them, they basically work on the basis that if you set a task they are not going to stay within for example the ChatGPT window or the ChatGPT tab in the browser, but they can jump around taps. So for example, you say okay, I want to travel to Mexico, organize all of this so it will start searching for information, browsing airlines, doing everything that it needs, maybe even. 

00:04:50 

And that's the other thing, connecting to your calendar, checking when you're free, connecting to your email, maybe sending out emails that dear colleagues, I'm on leave because I went to Mexico. So all of these independent actions are going to be organized and carried out by our little agent. Now that's, you might think that this is amazing. You might already think like this is maybe a bit too much, but let's, let's just look at what, what else it can do. I want to show you this screenshot. 

This was, this is actually a video shared by Anna Mills. You will find the link to this at the end in the references section. So I just want you to look at what ChatGPT Atlas is doing within a learning management system. A learning management system is like Google Classroom, right? So in this case we've got Canvas and I'd like to draw your attention to this part here. 

00:06:00 

Answering question six. Also. 

Sorry, I don't know why my cursor is doing weird things, but I managed to get it back. So look at the, the things that I met, I circled. So it says answering question six. It narrates the process like okay, so kind of narrates the thinking process. And I also want you to look at this part so I'll select true. 

It basically narrates the whole process, how it goes through the entire test solving procedure. And basically this AI agent can just go into an elearning platform and complete an online course from start to finish. And you as a teacher will not see anything suspicious. To you, it will basically look like a student was solving everything. You will never find out that the agent is doing the work behind a student's computer. 

00:07:24 

Oh, somebody's saying that the agent is infiltrating my computer. Maybe, maybe this is like the machine is fighting back because I'm sharing all the secrets now with you. So let's go back a little bit to our definition. So we can now see that AI agents are powerful little things. They are still not there yet, so they make mistakes. 

Even in this video, if you watch the whole thing, you will see that at some point the agent cannot locate a question, so it managed to answer it, but it cannot click the answer because it cannot find the place where to click. And it's not like it seems to us that it's not a big deal. But for the system to put this all together and make it work, maybe it is difficult, difficult. So there are still some mistakes, but it's a capable little thing. However, we need to also look at some other terms here. 

00:08:31 

And you may find a lot of expressions terms on the Internet, several sources and teachers and consultants talk about AI assistants. They also, we need to mention things so called shareable chats and we need to mention custom GPTs. Google calls them gems and Microsoft Copilot actually calls them agents. So the issue here is that all of these expressions seem to overlap. Let's clear this up and understand what's actually going on. 

So what we can say about AI assistants is that it's usually what we can also find in things like Magic School, for example, if you're using websites like this. So this is something that we can call like the feedback writing assistant or report writing assistant. You can do this in ChatGPT or any chatbot. You just create a specific prompt that defines the core task and you can just set that as a start and basically use that chat conversation for the same thing again and again and again. So for example, in this report writing prompt you can say you are a teacher in eighth grade. 

00:10:05 

You need to write reports or year end reports for students who blah, blah, blah, blah blah. So you set all the details that you need and then you say, okay, so we got, I don't know this term's grades. Here are the grades, write the report and then you repeat the same thing the following term and the next term. So that could be an assistant. You specified the task and it basically does what you ask it to do every single time. 

Now shareable Chats. I think they are a great idea for working with students or giving them additional activities that they can do at home. And I, I'm, I'm exploring this kind of thing. So for example, you create an argumentative essay developer or helper. It's basically a step by step dialogue style activity that you can share with students and then they can have a go at it.  

00:11:12 

For example, let's look at this argumentative essay developer. The prompt says your task is to ask the student what their title or their proposed title is. Then ask them what their thesis statement is. Suggest some ideas for a better thesis statement. If the student selected or picked a final thesis statement, then ask them to write the first paragraph and then start arguing with them so that they can practice writing argumentative paragraphs. 

In the end, give some feedback to them. So basically you build up a step by step prompt for an activity that you can share because ChatGPT and Gemini and Claude conversations can be shared with students and they can then use this mini interactive task on their own computers in their own way. So that's also a thing. And custom GPTs or as copilot calls them agents, they are not really the agents that I discussed previously. They are still pretty limited, but they can be useful. 

00:12:44 

They usually have uploaded knowledge and they complete, or they complete a specific task. For example, you can see it here, a school frequently asked questions chatbot. You upload all the documents that are important for your school and instead of a human assistant or a human receptionist, the chatbot can function as an, as an agent. But it's still limited because all it can do is search the uploaded sources and give a response. It's still going to be better than a customer service chatbot that for example you experience through Ryanair. 

But it's still limited. So it's not as complex as the agents that we saw before. So let's go back to our little Jack Sparrow and let's think about a student version. And I could see in the chatbot that or in the chat box, I'm mixing up these terms now that somebody mentioned that students in Asia would love this kind of thing. So that's my question actually. 

00:14:07 

What do you think? So let's imagine an agent that would complete a student's essay writing task from start to finish. Research, writing, cross referencing, creating charts, PPTs, personalized reflections, everything in just a minute, maybe 10 minutes. Now, would students like this? What do you think? 

And it could be, although I kind of, ah, I'm getting back the chat and you're saying definitely Yes, I think so too. Definitely yes. But the question is why? And I'm using the nihilist Penguin meme here. I don't know if you're familiar with this meme, but why? 

So if you think back to my first question, you said that you wouldn't enjoy an escape room experience if there was this AI agent doing all the work for you. But, huh, Some answers are coming in, but students would probably like an AI agent that would complete this essay writing task for them. So what's the difference? And some of you mentioned laziness. Could be.  

00:15:32 

Yeah, I mean that's one way to put it. But let's see if we can reframe that idea that all students, maybe not all of them, but we tend to say that they are lazy. 

We basically want to have some sort of intrinsically motivating task. So something that is truly motivating. To be honest, an essay writing task is not truly motivating. I mean, maybe I can explain it to myself that I need it because I need to get to university, but honestly, who enjoys an essay writing task? And then another thing we all need is a sense of achievement. 

So for example, in the case of the escape room task, I went there because I wanted to be there, I wanted this experience, I want to enjoy myself. So there's the intrinsic motivation and there's sense of achievement. Because if I manage to complete the individual tasks, I can, first of all, there's always some kind of reward. If I manage to unlock a little locket or I manage to turn on a light, there's instant feedback and gratification. But also at the very end, I get this big applause and ovation like, yay, you solved the task.  

00:17:03 

So there's sense of achievement. So what can we, how can we apply this to teaching? And I'm going to share some tips with you. So this won't be a super. I mean, I can only give you guiding ideas. 

I won't give you specific task ideas, but sort of just starting points. How we can perhaps push our students in the direction of doing the thing so we want them to do the work and not the AI agent to do the work. So one way to do this is that we could create or bring into our classrooms truly communicative tasks. Basically, the main, the fundamental concept of communicative activities is that there's a real need to communicate because there's some kind of information gap. So they need to communicate with each other because one knows something that the other doesn't know.

00:18:11 

And also there's some Kind of authentic context. Usually, if you think about it, that's what we get in course books. Like talk to your partner about your weekend and then, okay, why should I talk to my partner about my weekend? What exactly are we doing here? What's the point? 

And that's one of the things we can do is to look at these coursebook activities and rework them. Maybe with the help of AI why not? We can, we can use it for idea generation that I've got this activity. Can you make it more communicative? Truly communicative. 

00:18:57 

Another thing we can do is task based learning, which is in a way similar to this communicative task idea that we start with an authentic task. Students need to solve something so there is some end goal. They need to, for example, maybe put. 

Put together a box or a piece of furniture or some. They have to do something that has a clear result and the way they get there is not specified. So they can use their linguistic abilities and knowledge and try to solve the task. So there's again some sort of achievement at the end and true authentic communication. Another thing that we can do is project based learning, which again means that we create something that's worth doing or hopefully is worth doing. 

00:20:09 

That's why it's a good idea to know our students and know the age group we are teaching so that we can give them a project, or again ask AI to design a project for them that they would enjoy and that creates intrinsic motivation because they would like to be involved in such a thing and has some kind of end result, like shoot a video or create posters that then we can show to the whole school or write project proposals that we can send to the Council of Europe or somehow we can make them see. Because this generation is very much aware of why should I be doing something that has no purpose? Or why should I be doing something that I don't see the value of? You might be thinking like that is weird, but this generation is pretty much thinking in this way, like, I want to see how this pays off and how I'm going to get some valuable learning out of this. Project based learning could be a solution. 

00:21:28 

Again, escape room. But in this case we don't have to take them to a real escape room unless you have time for that. We can create escape room style activities within a lesson. So basically it's kind of like you set up stages and students work in teams, they solve certain activities and you hide codes and other numbers and things behind the activities. It's great for revision. 

And in that kind of Thing again, you create a situation in which they, they feel that what they do is explain exciting and there's some reason for doing it. And my final idea would be interestingly. But that's my personal experience. Let me know what you think about this in the chat. Multiplayer quiz games. 

00:22:28 

Now what I have in mind, I'm going to show you some examples in a second. Is that for example, Kahoot. But Kahoot is maybe getting a bit boring by now. They, there are these activities and let me show you what I have in mind. Maybe too much information at first, but if you look at number five, maybe you. 

You tried quizzes which now changed its name to Weight Ground. And also nearpod has this kind of thing. There are, these are multiplayer quiz games where students can compete with each other as they are solving the quiz and they need to climb a mountain or they need to fight each other. So basically these incorporate video game like elements to make the quiz more exciting. And what I found personally is that my students love this kind of game, love these games also they love the option to repeat the quiz again and again and again so they can compete with themselves as well. 

00:23:45 

They can see like, okay, Yesterday I got 50 points, today I got 100 points. And this somehow makes it more motivating for them and more rewarding for the other ideas. I also included some links that you can click on interestingly, like for truly communicative tasks. I still find, and I honestly have to say that Jill Hetfield's Intermediate Communication Games, which was published in 1990, insane. But it's still one of the best resource books I use. 

It's full of motivating, truly communicative activities that my students consistently like. And the other links as well take you to great ideas and resource books and places where you can get some ideas. But feel free to use AI for this kind of thing. And hopefully basically with this approach, I'm not saying that we can avoid AI agents because we can't. This is now a new thing. 

00:24:55 

And the new thing, AI agents are probably going to be everywhere this year. That's what they propose. They being influencers in AI technology. So what this really means is we need to think about online exams, tests, online discussion forums, for example. Because AI agents are going to be amazing at these. 

They can just pretend to be students and solve these activities. So if we are aware of all this, we can design activities that are truly motivating and exciting and hopefully our students are going to take part in these. So this was supposed to be my final point that in a way technology is not going to stop. So all we can do as teachers is be aware of what new developments there are out there and try to plan our lessons in a way that can keep our students with us, and they will then do the work, not the agents. So thank you for being here and I'm. 

00:26:19 

I'm looking forward to any questions. 

[Sheila Moreno] 

Thank you, Joan. Wonderful session. Well, as you can see, teachers were highly participative all of the session. They were really, really into it. 

Very interested. Guess what? We do have some questions.

[Joanna Szoke] 

Amazing. 

[Sheila Moreno] 

Let's see, the first one over here is by Professor Barazi. 

What's one simple needs analysis step that can improve the activities we plan in order to make them more communicative and authentic? 

What would you recommend, Professor Barazzi? 

00:27:05 

[Joanna Szoke] 

Let me. What's. Oh, I can see the question now. What's one simple needs analysis? 

I'm not sure I'm. I fully get the question, I think, but I'm guessing that in order to make them authentic and truly motivating, we need to just be aware of the needs and interests of our students. I think that might be the question here, that we need to figure out what they are going to use this language for. So whether they know the problem is when. And I kind of saw, I think, a comment in the chat box about that. 

If parents make them study the language, then it's a difficult case because then we need to try and find something that could make the whole learning process relevant for them. What I usually try to do is I ask them where they use the language, and eventually it turns out that they do use the language somehow, for example, in video game forums or YouTube comment sections. But once we find the authentic place where they use the language, we can then try and focus on that kind of context as a motivating environment because we need to find their connection with reality. Usually lessons lack this connection, not because the teacher's bad or the lessons bad, but because somehow there's no connection between the system, the requirements, and. And what happens in the class. 

00:28:53 

And that's. I know it's a bigger problem, but we need to somehow overcome that. 

[Sheila Moreno] 

Thank you, Joan. Thank you. We have another question over here by Professor Fernando. 

He asks, does using AI change how we learn or how we think? 

[Joanna Szoke] 

Yes. I would say yes. There are already several research articles and studies looking at this. There's a thing called cognitive offloading, which is often used in connection with AI and generative AI using AI, Basically what this means is that we want to solve the task super quickly.  

We just Ask you Beatty, or now we can just give the task over to an agent and we zone out because it will do it for us, but then that will slow down our mental processes. It won't make. Or it's like not working a muscle. Now, the problem is that humans typically want the easier solution, so we will always go in that direction. And somehow we as teachers need to push our students in the other direction. 

00:30:12 

[Sheila Moreno] 

Yeah, yeah, true. Thank you very much for that, Joan. We have lots of questions. Let's choose another one over here. And it says, wayground doesn't work well with dyslexic students. 

It gives points based on how fast your answer. It doesn't work for me. What could you suggest to Professor Miguel in these cases? 

[Joanna Szoke] 

Yeah, that's. 

[Sheila Moreno] 

I think. 

[Joanna Szoke] 

Well, it's interesting because I. My experience is that Wayground seems to be the most special needs friendly platform. It has lots of accommodations for special needs students. And I haven't tried it yet, but I think you can like, cancel out that option. This kind of speed favoring part, I think. 

I don't know. But yeah, that's a problem, and I don't really know. Well, if you cannot cancel that out, it's. It. It is a problem which I cannot solve at the moment. 

00:31:20 

[Sheila Moreno] 

Yeah, sure. Well, thank you very much for your input. I. 

[Joanna Szoke] 

Oh, I think Sheila, I lost you. I'm not sure. 

[Sheila Moreno] 

Hello? Yes. Yes. 

[Joanna Szoke] 

Okay. Yes. 

[Sheila Moreno] 

Then we have a. Hindering AI is hindering our creative process. What does this imply? 

[Joanna Szoke] 

Well, What does this imply? I mean, the downfall of humanity, I could say. But what I typically think in this case is that it's not necessarily AI that's hindering our creativity. It's. It's always something that. 

00:32:12 

So we sort of have these things in us and in ourselves that here's now this tool that brings out this problem, which we probably had before as well. 

And I just think that we need to sort of. Well, I won't be able to give a good answer because we cannot just accept that this is the case. Because as teachers, we want our students to think creatively and. And work their brains and learn. 

But at the same time, I mean, okay, we can do this with our students. And what about the rest of the world? Because AI is now new to everyone and we are all at the same level in this kind of thing. We can deal with our students, but the adults also outsource a lot of their thinking and work to AI and no one can really influence them. And they are a huge majority.

00:33:19 

Sure. Unless again, intrinsic motivation. Unless they would like to keep their creativity. And I do see some movements for this. There are some people who say, I'm not going to listen to AI generated music. 

I'm not going to appreciate AI generated images. I want authenticity. So there are some movements. 

Yeah. I hope that people will find this important. Of course.

[Sheila Moreno] 

Well, really valuable, John. Definitely. 

And, well, we really appreciate you being here. It was a fantastic session. As I said before, teachers were highly participative. Really into the session. 

Chat box is just absolutely crazy. Yeah. 

[Joanna Szoke] 

Yes. 

00:34:12 

[Sheila Moreno] 

Thank you very much for that, Joan. And, well, if you could give us a last insight to wrap this session up, what would you say? The teacher. 

What would you tell to the teachers?

[Joanna Szoke] 

I think keep it authentic. I would say I just believe that, yeah, we might think that, okay, AI is ruining our teaching or our lives, but as long as we try to find this connection with reality and make our students see that what we are doing is actually, or it makes sense, they will, I hope, stay with us and follow our lead and do the tasks we set them.

[Sheila Moreno] 

Thank you very much, Joan. Totally agree. 

And now look, all the teachers are going crazy with reactions, so they loved your session. Thank you very much. Thanks again. Thank you.