 as well. So welcome everybody. Really delighted to be here, to share with you our work on AI in classroom. And I'm here with my colleague Neha Gupta. Neha, do you want to briefly introduce yourself before I make a start? Yes, so hi everyone and absolutely delighted to be present on today's webinar. So I'm Dr. Neha Gupta. I'm assistant professor in information system management and analytics groups here at Warwick Business School. Prior to my academic career, I have been a practitioner in IT industry, you know, as a data scientist and software developer, but I have been in academia for the last three years and very aggressively we have been working on this AI task force project, the findings of which we want to uncover as part of this presentation. So I was very fortunate to support my senior colleague Dr. Isabel on this project, so that's me. So you will hear from me in the second part of the presentation, but over to you, Isabel. So yeah, similar to Neha, my background is in industry where I worked in technology projects. So I led a technology project mainly in digital payments. So hence my interest in AI and I start off with giving a contextual overview. Then we discuss the opportunities we established from our task and finish group and how it's linked to pedagogy. And then I give some basic examples and Neha takes this on and with more advanced examples. And then depending on time, I might just be online and some of you will have heard it before, but I developed an AI based formative feedback tool. And for those of you who has not been aware of this tool, I'm very happy to discuss it at the end of our meeting subject to us still having time for it. So as an outset, I would like to outline that there is no single AI and we are aware of this and we are using AI as an umbrella term. So it's really because I'm aware that some people find it important to point out that there's not a single AI and that I do agree with it. And we're looking at different variations here, but we are using the word AI as an umbrella term for you as you can see this onion graph for various aspects. And please do put questions in the chat if you have any or raise your hand if you have any questions. So we believe that AI has the opportunity to democratize education and we call it the pedagogic paradigm 4.0. So here AI education students work together as partners on having a clear vision and the vision and one challenge being really about fairness. In our view, it's about trying to use AI ethically and to the greater good of actually trying to improve current perceived weaknesses in education. And one of those perceived weaknesses as we know from NSS data across the sector is formative feedback and fairness of assessments. And that's the reason why we focused on formative feedback from the start because we believe there is something we can actually improve. And using AI is not standalone. It has to be embedded within processes and pedagogy. And you see here an extract of an article we recently published. We say that actually AI has to be embedded in such a way that students actually want to listen to you. So from the start, students might have the perception because they're younger and they see themselves as more digitally able. So we want to acknowledge that actually students use technology on a daily basis. Perhaps not in the way though that actually we want them to use technology as part of this teaching or as part of the preparation for their future work. So we acknowledge students' knowledge of technologies while actually acknowledging as well that there are different aspects that they might not be aware of. So there's something they can learn. You then might want to position yourself as an expert because we find ever so often that students actually want to be taught by an expert. So it's about you demonstrating your knowledge and discussing hot topics from industry and how they apply to the technology that you want to convey. So students are inspired and it's only then that you can then move on to facilitate a role where actually you encourage students to explore AI. And here we come to an article that Neha and I are about to submit for a special issue on Genitive AI is that we actually find that at the moment students do not use the full potential of AI because they don't really explore it for themselves. They want to be told but we have to tell them actually that they can explore and develop their own understanding first. But this role as facilitator and conveying what AI is for them to explore needs to happen only after you've established first what students expect you to do is showing your expertise and acknowledging that they are actually quite good in using technologies generally. And by following this sequence you might get more buy-in with students conceptualizing, visualizing and shaping the unknown future. With a group of over 50 members from across Warwick and also other institutions within Europe and worldwide. And we have, would you mind sharing the article? Okay, thank you for your interest. I'm going to send the links but to not cut my task. So I'm going to do this when Neha talks. I'm going to post some of the links to the articles and let me just, and this is the slide I'm showing here, the link, the link I have to hunt. I know it's the link I didn't have to hunt because I just, this is the link I shared with the other. But the link I shared before, so somebody here and you can see it. And so thanks Neha for doing it. And here we are looking at the moment at the slide of our resources because I really, really encourage you to spend time on our resources and share them going forward and making sure that actually students use the site as well. And so we have colleagues of mine for big modules of over 500 students. They use our resources to get students to start to think about AI and how to use it in an academic way. And you can find our resources. There's a full report on the top left. And in the right column, you have, for example, a student presentation for students, a student report for students, a student report for academics, academic advice for academics. So you have a mixture of resources that you might want to get your students to engage with, or you might want to engage yourself with. But it's six months work of many of us. So do spend time on the resources. They are up to date and hopefully you make use of it. But you see broadly that we have split the report into six chapters. We have first the AI Enhanced Learning Environment, where we try to explain why we believe that using AI in education is a good thing rather than moving back, let's say, just to unseen exams. We have a chapter, and this is headed by Neha, on the teaching of the future. We have a chapter on ethics and academic integrity. So our decision at Warwick was to leave lecturers the choice of whether they want students to embrace AI, to use AI if they want to, or if they want to forbid AI. And you can see our rationale in chapter three. And we have a chapter on assessments and principles of using AI for assessments. We have a chapter on feedback and dialogues with AI. In that one, you also have the discussion of formative feedback provided by AI. And then we have a chapter which looks forward thinking about possible changes in higher education, to environment more agile, and developing a supporting infrastructure. And so we have six broad chapters, but each of them is in three different parts. One part is produced by students only, one part is produced by our working groups, and one part is produced by external contributors. So in total, you'll find there are 18 subsections. So, absolute amazing nighttime read, so don't miss out on it. And at this point, maybe we can have some feedback from everybody here, is because I'm going through easy examples. And the reason I mentioned easy examples is that we find actually, and I was just recently at the same with doing an internal presentation, and we asked who of colleagues had actually themselves used generative AI in different forms. And actually it's a very small proportion of colleagues who have actually tried out different forms of AI themselves. So my question here to the audience is, I give you the example of Dali and Heijen. Who of you has used Dali for hard work, and who else you have tried Heijen? Anybody? Just write in the chat or raise your hand. Nobody? Okay, so if you haven't, so Michael, perfect. And so Michael, you, and Andrea, you both have used it, and Mike as well. And so for those who have used Dali, and I think it's easy to use, and I assume you agree with it. The other thing is Heijen. Heijen is a new way, so you upload your picture. Thanks. Most have used Dali, good. And so if you haven't used Heijen, you find here three examples of me producing a birthday message. Actually for my godson. So you see my first recording, and then you see the translation into French, all done by AI. And it's all a question of me typing in my birthday wishes, and then it's then translated and it's produced as a talking picture. So you see the picture I have on the first page here of this presentation, and that picture starts to talk in a, I think I chose British accent rather than American accent. For those of you, and I see nobody has used here Claude yet, it seems. And I think you should really start looking at Claude. I'm really, really impressed with Claude. I also like Bart, and none of you mentioned Bart. Bart is so easy to use. So, and do sign up for them. They all are free to use up to a certain, you know, degree, because after all you give up your data, and you also have often a limit of what you can do. But, you know, and each about, okay. And so Gamma, do you want to tell more about Gamma? Because I haven't tried Gamma yet. Do you want to unmute yourself to share the insides of it for slides? Okay. So, yeah, Nia, are you going to talk about Gamma afterwards with your PowerPoint presentation? Yes, I think there is a link I have provided to Gamma saying that, you know, and I will show the chativity way of creating presentation also, but Gamma is a tool or, you know, a app which can help you generate that. There are multiple ways to do the same thing, and Gamma is one of it for PowerPoint. Yeah. Thank you, MSC. I'm quite happy to see that a lot of people are actively using AI here. Yeah. So, Nia, your turn. So, you can now continue with the more the advanced versions. Yes, yes, that's correct, Andrea. I'm just reading the chat messages, you know, and I think Andrea is quite right in saying, but saying that, I think, yes, it produces non-reliable information, but for me it's about, you know, how we can productively manage our time, and even if, you know, we get some courses slides to begin working with, it saves, you know, some amount of your time. So, the more we start getting used to generating stuff automatically, and then refining it later on what we want to put our experts in that, then I think, you know, that should be the way forward in my view. So, yeah. So, thank you, Isabel. I think, yeah, so, for your introduction and, you know, for sharing your knowledge. So, from my side of things, I would like to share four use cases, prompt engineering. I'm sure if, you know, you are using Dali and CharGPT extensively, you might have come across prompt engineering, but I will touch on that. And then there are other two or three tools which I have used myself personally, which I think they are good interest in use cases and teaching context, which I would like to share with some fellow colleagues here. So, that's the overall broad agenda, you know, for my doc here. Yes, I think you're right, Andrea, you know, Adobe and Microsoft, they are already, I think, embedding plugins in their applications, which can allow you to do that. And I don't see that in your future, it could be months, we will get the latest update, who knows. So, that's absolutely right. Next slide, Isabel. Yeah, so, prompt engineering. So, how many of you know about prompt engineering? I think most of the people we have got five, yeah, co-pilot is good, MSc for coding kind of things. I've seen it a bit, Andrea says, okay, so let me just, okay, Mike has done micros about prompt engineering. So, this is a very useful resource for prompt engineering. There is a full dedicated website, you know, if you want to spend some time as part of your CPD, it is a continuous professional development activity, learn prompting. So, there are a lot of clues and guides and examples there on this website, which tells about how you can go on getting better at writing prompts when you are interacting with these generative AI models. So, one of the well known method, which is called short prompting, this is to do with, you know, when you write your prompt to a generative AI tool and you provide specific number of examples in which you want your output to be seen as that's called short prompting. So, you write a basic prompt and then you refine it that this is the kind of tone I'm looking in the text or this is the kind of example I want. I want you to bullet points the stuff, you know, in the generative AI or large language model is written in the output. So, that's called short prompting. So, you can continue from a very coarse looking shallow prompt and going on to the fine key prompt, you know, so that's that's something which is that bit is something which comes with practice. The more you start writing prompt in this way, the more you will get better in writing prompts and get better output from these large language model. You can use it for structuring the data, as I said, you know, writing emails, different writing style bullet points, paragraphs, giving example, and for image prompting, as Isabelle mentioned, open AI, a Dali tool is also the one which kind of works on this prompting kind of mechanism and you can refine the prompts to get a better output. So, on this particular website, I have seen, you know, there is list of some example products. I think one of the links you will find on this website that which all products, which all language models or generative AI tools take prompting as an input to generate an output, there's a big list. And that becomes, I think, a window to many other tools, which we don't know generally, you know, because I think part, chagivity, drama, being, they are the common ones, but then behind the doors, there are many other which are available out there and how do you know which are these ones, so this can also kind of an encyclopedia of it. Do you think users need to understand how AI is actually creating output, as otherwise they think of it as magic? I think that's right. I would say, Mike, but I think there are two audiences here. As educators, we ourselves need to have some sort of basic understanding of it, how the output is generated, but it gets into really machine learning and deep learning side of things and how these large language models are written and how the data is actually processed, you know, internally. So, I don't think all educators coming from different discipline backgrounds are, for myself, I understand up to fair bit because my background or my training is in computer science, but I think it is hard, but I would say there is a very useful resource. I can share the link later on with Edith. I had seen a slide from Dominic from Oxford University. I think he is the technology support person and he has got some very interesting slide share. I think I'll be sharing that link in one of his presentation, in one of the upcoming slides. There, he kind of talked about in at least five, six slides how exactly the output is being generated. So, from a layman point of view, you can get some sense of, you know, how the data is being combined. So, can I just come in here, just briefly. So, first of all, the University of Oxford. So, Dominic's page is linked on our page. So, you can see it in additional resources at the bottom right. So, if you go to our page, you can see the input from the University of Oxford. Second, we're coming here to do students need to understand it. We also need to think who is going to teach this. So, because if you think about our various modules, if you're doing, let's say, a third year undergraduate student, you as the lecturer on a topic, let's say, accounting, wouldn't want to then start to explain how AI works. Because actually, students might have heard it before, they might not, but it's not really the topic you want to cover. So, on the one hand, I don't think it is the lecturers role to now start to all explain AI. It's more like a university decision. That's why we have our AI resources there, is that students actually have to take some ownership of engaging with AI if they want to understand the detail of machine learning. So, I don't think, coming back, I don't think the lecturers role to start to now talk about it. I do think students need to know how the basic understanding, and I've written two teaching cases, for example, to explain the basics of AI. So, do look at my, you know, on the page, there are two teaching cases, and you can use them if you wanted to discuss the basics of AI with students. Yeah, I think Mike is saying we need to understand the limitations. Yes. I think the way we can share that knowledge with our students, in my point, we can give them examples, you know, examples where an AI goes right, examples where AI goes wrong, you know, and that's how we kind of communicate, you know, the barriers, as Isabelle said, or the limitations of it. But opening up the black box, I think that that cannot be possible because people are coming from different disciplines, isn't it? And then coming back to the limitations, exactly those two teaching cases I wrote are talking about ethics and the ethics of AI. So, do look at those teaching cases and possibly use them in your teaching or share them with colleagues. Yeah. Yeah. Yeah. So, I would say I agree with it. So, I think continuing on our slide, Isabelle, can I move to the next slide, please? Yeah. So, as I mentioned, you know, example, prompt engineering. So, in this one, you know, as part of writing the report, you know, I was like, you know, let me play the chat GBT because I wanted to reword a certain paragraph or kind of rephrase it using chat GBT. And you can see, you know, in this particular example, the rephrase output which was generated by a chat GBT for a particular paragraph, which I inputted to chat GBT is kind of very ambiguous in its meaning because in the paragraph, if I can read it out, you know, the one which I wanted to rephrase it, I wanted to rephrase the discussion of this is trying to focus on the potential use of AI and some of the commonly available AI tools that can be employed by educators and students alike in a collaborative manner. Okay. So, the question here was educators and students collaborating with AI. And I wanted AI to rephrase it. AI completely changed the context of it in this case, chat GBT. And it's a, you know, the discussions entered around AI potential use of readily available tools for collaboration among educators and students. So, it's not about AI collaborating with students and educators. So, AI on one side and educators and students on other side, however, the meaning completely is changed. So, the more you play with the prompts, you will see such examples coming in your own teaching practice. And what I tend to do, I tend to take the screenshots of this and save them in my learning material. And on the beaks, when I'm going to teach a little bit about AI, how they are going to use AI, I do share these examples as limitations of, you know, of AI. And I think, you know, knowing the art of prompting, these are another three examples that have used personally, you know, you can give a prompts like, you know, proofread my writing above fixed grammar, spelling mistakes, make suggestions that will improve my writing clarity. So, here, you are doing short prompting, you are providing specific instruction to chat GBT. You can also use 80-20 rule in prompting, meaning you want to say, insert a topic, I want you to learn about the topic. Can you identify most important 20% learning from this topic that will help me understand 80% of it? So, that's like, you know, extracting the gist of maybe a big page or two pages of an article and generating new ideas is something which I've heard, you know, most of our students have started using it as prompting to ask, you know, new kind of ideas, what can I write about, you know, particular topic, what are the some of the initial ideas and that can be a starting point. You know, removing some barriers and thinking. So, that's prompt engineering. Next slide, please, Isabel. The second one, I think most of you are already going the way to interact, change to be more conversational rather than prompting. It can, yes, but I think I already see prompting as the interaction in my new mic, you know, because when I specify examples, when I define my language, the way I talk to the AI tool, I feel I'm interacting with my AI assistant. So, I think we're already moving in that direction. So, you just in response to Mike's, Mike's question point here on the conversational tone, and I looked at how others prompt and surprised how many people still use words like please, could you please outline, so they really develop this conversational tone and make it very human-like. And then I keep saying, why bother to write please? But anyway, so yeah, I can see Mike where you were coming from. Yeah. No, what I'm really getting at is, I mean, this prompting, it's based on the fact that the way the human, the humans have built, you know, the question answering and on top of the large language models. And, you know, it's sort of like a black box, really. People are, it's an art, not a science. And surely, you know, in the future, you know, a more conversational model which will say, could you make some slides about this? And the AI will come back and say, okay, how do you want me to make the slides? And rather than this weird way of writing prompts, trying to work out how to interact with a black box. And I'm imagining that prompt engineering, people are teaching it at the moment, perhaps by next year, I'll be out the window. But then Mike, it's a bit like what I think a lot of no code, no code environment, isn't it? And it still takes some time, doesn't it? You know, we talked about no code, low code or low code, no code for some time. And we're still, I think, using code. So it's not as when you said, you know, it's, I think progress isn't as fast as we imagine. I think it's only the prompting that will change the shape of it. I think we are already using part of conversational element in prompting already. And slowly, slowly as the time progresses, you know, Gartner technology, kind of cycle, if you look at, sometimes we think it's, we are in a hype mode, and then it drops. As Isabelle mentioned, I think two years ago, there was this hype, you know, no code, no code is meant to replace everything, but people are still coding. So I think it will take a while before it becomes completely conversational. Okay. So I think, Gama, thank you, Mike, you know, it's really nice, you know, we are having this interactive session, and people are really engaged, you know, with the material which we are sharing. So thank you for that. And moving on, Gama, I think, as many of you mentioned, you have used it. So Gama is an app which can be a web app, you know, you can use to generate presentation slides automatically. But you can also do it in chat. So if you know, VBA is visual basic code, which is available in most of the Microsoft suite of products. It's a code, it's a coding language, visual basic. And if you ask chat to give you a code that can generate a PowerPoint presentation about a particular topic. So for example, in this screenshot, as you can see, you know, if you give charge, you can say, give me a VBA code that generate a PowerPoint presentation on a topic of a lesson plan to teach AI challenges and opportunities. I need, and I need four slides again, you know, short prompting just to specify the stuff. Then you will see charge, you will actually generate the code for that. And it'll even tell the steps you need to do in your PowerPoint. So if you mind, Isabelle, I can I share my screen because I want to give the live presentation of this how I have done it. Let me share my screen, give me a sec. So I hope you are able to see the PowerPoint slide. I'm opening the PowerPoint slide here. Yeah, we can. Yeah. So what, so the, so charge it gives me this instruction here, it gives me the code here. I, I just, can you still see my screen? No, I think you know, we only see the PowerPoint. No, no, no, I think I want you to see my entire screen. Just a sec. Because I'll be entire the screen. Yeah, I think I'm on. Can you see my screen? Yeah, I think you can now. So if I go here, and if I, so you go to your PowerPoint, which is like this, and you just say alt F11 here, and that's what I'm doing. Alt F11. So this, this is the, you know, the visual basic window, which gets open, you just simply follow the instructions, click insert. And you're inserting the code here. So you just open here, you copy paste the code. And what does it say in the module you created a new module copy paste it, and then you close the BBA editor. So this is the window wherein you have, so you just close this visual basic editor. Now you have copy pasted the code there and then you run alt F8. So these are all shortcuts. So you say alt F8. And then you can say, you know, this is automatically this module was created when I copy pasted the code. And if I say run, the four slides are generated. So this is really course, you know, but then you can refine your prompt, but you have got at least five or 10 or 12, whatever you specify in the prompt, and it will give you some basic slides. So I can start, go into each slide and start editing and, you know, do more stuff. So I think chat, if it is able to give that code, and you can use that in your, you know, if you want to create a presentation about a basic presentation about any topic. So Isabel, can we go back to the slides, please? I'll stop sharing. So there is a question. Do you need to provide additional information prompt for the slide to be accessible? No, no, no, no, no. It's just that you just asked chat GPT about this code, which is a visual basic code. It gives you the code, follow the instruction, and that's it. You are done really. So you don't really need a lot of coding experience, you know, and you can still play the PowerPoint on that. Is that okay? I did. Has that answered your question? Thank you. I was thinking of accessibility from the user point of view. If you had a screen reader, for example, screen reader, I didn't get that. So the PowerPoint, yeah, sorry. I was just going to answer that. If you're using the standard PowerPoint template, it'll be accessible. Thank you. Thank you. Next slide, please. So yeah, so these were the steps I, as I, you know, just showed in the demo. So next slide, Isabel. Next one, please. Yeah. And the next one, which I have used. Yeah. What about one? Responding to emails, marking. Responding to emails, I think there is this example, you know, paraphrasing. So I don't think I have tried with any AI, automatic AI respond to my emails yet, but it's really good in terms of paraphrasing and, you know, in terms of writing nice emails, I've tried that, but not automatically I have tried marking. So then it depends on your institution. And at Warrick, we have a clear instruction that we do not use AI for marking. We also, lecturers are not allowed to upload anybody else's assignments onto any form of generative AI. So we have a clear no for marking policy, no for AI for marking. And of course, your institution might handle this differently. But I would go with your institutional guidance. And I generally think at the moment, and from my work on using AI for formative purposes, it is, it is, you know, we always discussed about automating versus augmenting. I do think in the long run, we will be able to augment the marking process. But I do think it has to be a collaboration between the human marker and AI that actually work together. And AI can make the marking process quicker. But I don't think AI is that they're yet to actually and accurately provide a mark that students believe they get based on the teaching you have provided. So and we are very happy to discuss it. And one of the teaching cases I wrote was actually using AI for marking. And if you look at it, it's even, even if the percentage difference is small, and the stakeholders, the students, the greater public would not approve of an institution to move marking to AI. And last point on this, I do think as a first step, we have to differentiate between low stakes environment, let's say, assessments that are worth less than 20% or 20% less of the module mark and high stakes environments, which is like the main over 80% of the mark. I do think if you want to start using AI for marking, we have to be upfront with students about it and develop something in a low stakes environment, rather than just lectures, trying to upload essays that are worth 80% of the students mark and then coming up with the feedback and marking at the moment in a non structured way. Just on that, there is nothing stopping a student putting the marking rubric into AI and uploading their essay and seeing how it marks it according to the rubric. Absolutely. And then that comes back to using AI to actually improve, because then you students use it to engage further with their assignment. So absolutely, students can do it, but for us lecturers to do it is next step. So I think I responded in the response was for marking, I do think students should use AI to get forwarded feedback. Thank you, Isabel. I think, yeah, so the next one, another tool which I have used and I found it, I think just, yeah, sorry, Isabel. So there is a, you know, a YouTube link if you want. I think in the interest of time, I would not like to give a demo, but let me just very quickly share my screen. So this is a speech texture actually, yeah, so this is a speech texture. And if you see, if I press, So we, at the moment, we cannot hear the sound, but normally you would hear the sound and then you can see how the video gets straight away transcribed. It's a bit like when we use teams and we allow transcription, so that's similar speech texture. So yeah, so there are online tools. Thank you, Isabel. There are online tools available, you know, which can help you write some lecture notes, you know, if you are hearing a podcast or, you know, and there are important points, which you think you might need to include it in your lecture to share as an example, or as an industry use case with your students. And if you're not using teams, for example, or so there are things like speech textures, which can help you generate text to do it. Okay. Oh, that's good to know. Thank you, Mike. I didn't knew that Google Chrome has a free extension. So that's an excellent share of knowledge, you know, which can allow to do the similar stuff. Isabel, back to slides, please. Thank you. Yeah. Next one, please. Yeah. And then it's part, it's coming. This is then excerpt from our report, you know, we have shared a lot of public tools as on, you know, a month ago, or as of today, which we are available out there in the market, which can be used, you know, to enhance your education practice. There are some people to follow on LinkedIn, which I personally follow, John Betty, where I is one group, super human, ethylmolic Wharton professor, Wharton school professor in United States, he's quite active in sharing his voice with regards to whatever, you know, new progress is happening in January day where I may be following his LinkedIn page and AI French triad is another one. So these are the four ones where I see, you know, a lot of top voices on LinkedIn and what is being conversed on the generating AI piece. There are active people, you know, sharing their knowledge. So, you know, that can be a good idea to follow similar kind of influencers on LinkedIn pages. Next one, please, work, business school, the research cohort from ISMA information system and management analytics group to which myself and Isabel belong, you know, our PhD students came up with this cheat sheet AI for research. And here we share some of the links, which are there, you know, or some of the tools which can be used for literature generation, analyzing the literature, you know, literally guiding your research forward if you are in process of writing a paper and how you can utilize this one. So all that knowledge is captured as part of this cheat sheet. Please don't go with the word cheat sheet is, you know, a vocabulary which is used a lot of time in the science and in technological domain, you can find Python cheat sheets, Python is a programming language or our cheat sheet, see, people just use it, that it's a summary of concepts in one page, you know, and someone few years ago came up with the term cheat sheet and you know, just following on from then, but it's basically a summary of important concept on one of the pages, how you can get it done. What is what policies on using AI tools? Isabel, would you like to take that question, please? Would you know? So actually, so the policy for students, they have to sign, you know, their documents and the thing is, it's a bit like before you could have somebody to proofread your documents. So I do believe that at the moment we cannot, we know that there's no way that we can actually check and students writing that, you know, all attempts by the sector and not just all attempts by the sector to detect whether students have used AI, F and failed. So what we say is, you have to make sure that your assessment is authentic enough and provides them, is measured by things that do not depend on rephrasing using any form of AI. And so the policy is that actually we don't know if students use it and they have to sign documents where they might say that they haven't used it, because you remember I said that Warwick has a three prompt approach and lectures can actively encourage students to use AI and lectures can accept the use of AI or lectures can forbid and if lectures forbid the use of AI, it's up to them then to try to devise assignments where they can detect the use of AI. But generally the policy is so students, students, you know, we cannot deter students from doing so because it's just non-detectable. But one thing I find is that actually we worked on the same research because in May I had encouraged students to use AI for their assignments and we found that most students just use it exactly for the rephrasing and for forgetting the wording. But actually what students are assessed on is critical thinking for example. But critical thinking didn't come across you know if you just get things to rephrase you don't get the high marks because it's not going to show and demonstrate critical thinking. So what we have to do is to say to students look you can use AI to rephrase but make sure that actually you bring in your own IP and your own ideas to actually be able to demonstrate critical thinking. And so let's see the thanks Nia for sharing the link. So that's it from my side. I think you know I just wanted to share some examples which where I have started getting engaged with some of these AI tools and obviously there's more to it and fellow colleagues might be using it in a different way or similar ways. But it was just you to share what we have been doing here at Warwick and share it with our colleagues based on the practices we have adopted in our teaching curve. So over to you Mr. Ben, that was it. So the last part was on the tool and we can briefly discuss it afterwards but first to the respect to the question on the effect of the use of AI tools. So when I developed this AI formative feedback tool the students who used the tool and based on you know some analysis did six percent higher mark. So if students using the tool beforehand gave them a higher mark however we do have to differentiate between correlation and causation. So it's maybe not causation maybe it's not the students who use the tool that get a higher mark but it's mainly the students who are engaged in their learning who would have got a higher mark happen to use the tools and thus got a higher mark than other students. So let's be careful between correlation and causation. Absolutely Mike and equity issues yeah and but interestingly and I am you know those who have the paid version on charge gpt and actually and if you if you use a combination of let's say Claude and Bard and the free version you can actually receive actually have quite good support without paying. And yeah we as a school actually look into those things that we do look at it but at the moment we don't provide everybody with funding for getting charge gpt forward but I think they would be not right either because as I said Claude and Bard do offer alternatives so why would we go with charge gpt just because it's at the moment the paid version is better than the free version there. Any other questions and then I'm conscious that it is that before that we want to finish on time let me just briefly show you some slides and so this is and we have done this formative feedback tool and and the idea here is to bring different different parts together processes people and data and it's an additional report and it uses various form of technology it doesn't use gpt it uses bird as a transformer and the reaction of staff was mixed and AI ethics is important for us and we have the advantage that we actually know that the data we collect is not kept anywhere but we delete everything so we really really proud of this tool and it works as a formative tool so it's not involved in high stakes assessments as I mentioned before I believe that actually we have to try things out in a formative or low stakes environment first here you get the presentation that the type of features their students get and it's well perceived by students students appreciate it and the environment the IT environment might not be necessarily there to actually use AI tools in full yet but this is maybe to come so that was in a nutshell the part of formative feedback tool I did previously a presentation as part of an edith group on the formative feedback tool so you might have heard me talk about it but it's just to say genitive AI is not all and there are other forms of AI that are important and that helped us to democratize education and especially by providing a tool such as this which is free for all students we are bypassing the equity issues we've discussed before so um that's really the end so any last so any last questions otherwise we finish five two on on time any questions can you guys if I'm so I didn't try I didn't try exercise yet because I teach topics only with a word and I finish using mainly um qualitative research you know I haven't tried it either but I think I'm tempted you know when Michael says you're converting excel to csv enough to plot I think I'll give it a go Michael getting some ideas from here you know I haven't tried it as yet the big issue generally and this is last my last point is and then it's always very difficult to pitch AI talks at the right level it's it last time we did you know and I did something internally and we I felt we pitched at a too high level this time we and now I think actually with you as an audience we should have actually made it you know pitched it higher it's very difficult to predict and thank you for staying with us thank you to Isabel and Niha for this really engaging talk I think lots of conversations in the chat so thank you so much and everybody have a really good afternoon thank you bye bye thank you thank you very much thank you all thanks for listening thanks for having us