ChatGPT: To ban or not to ban?

and over the next month as I said we're hosting three webinars of which this is the first to create the space for this sector to Grapple with these unanswered questions around artificial intelligence after Easter we'll also be releasing some more guidance to support providers to continue on this journey of sort of adaptation and transformation in the face of AI I should say that my name is evil I head of public affairs at qaa and when we jumped up the title to this webinar uh to ban or not to ban that felt like a really relevant question that everybody was asking um and people are still asking it but a matter of weeks later we know now shortly that we will see generative AI integrated into software like Microsoft Word and Google Docs so in a sense the title sort of already seems out of date and it just shows how fast this is moving so I can't promise that this event will answer all of the questions that you might have on AI and what it means for the higher education sector but I've no doubt it will deepen the thinking and discussion on the issue for you and if you walk away with at least one New Angle or implication that you hadn't thought of before then that will be success certainly in my eyes and I hope you will get that out of it today so to help us grapple with this we are thrilled to be joined by five fantastic speakers we've got Michael Webb who is the director of Technology analytics at jisk and co-leads just National Center for artificial intelligence got Dr Phil Anthony who is head of Technology enhanced learning at the University of Kent Stella Maris who is director's assessor at St Andrews University and a member of qaa's student strategic advisory committee Dr Irene Glenn dinning who is the academic Integrity lead for Coventry University group and Dr Bronwyn eager who's a senior lecturer at the University of Tasmania who has very kindly joined us um late in the evening over in Australia so our speakers are going to speak for about eight minutes in turn each giving their response to the question to ban or not to ban do feel free to put questions in the chat as we go and we'll try and squeeze in as many as we can in a q a before we finish though it might be a bit of a squeeze so forgive us in advance if we're um rushing that bit we'd love to hear your thoughts on Twitter as well and if you want to use the hashtag qaa chatgpt to follow along from and connect with other people on the event then please do that and a heads up that we are also recording this uh for future use so I don't want to take up any more time uh I'm delighted to hand over to our first speaker Michael from jisk take it away Michael hi everyone so I'm just going to share my screen okay hopefully that's worked so I'm gonna look at two angles is it possible to detect chat GPT and is it possible to ban which I think we've already agreed it's not but we'll delve into that a little bit more so let's start with detecting um so is it possible to detect AI the answer to that is sort of which might not be terribly helpful but it's worth delving into this a little bit more um so no system today can conclusively prove that text was written by Ai and I think that's quite a useful thing to understand as a starting point all that the systems can do is give some kind of probability that it might have been written by AI I'm just going to go into the three main approaches that we see it's useful perhaps to understand that and then we can see why these are unreliable so there are three main techniques that are generally um people look at on this one is looking at writing style the next is machine learning classifiers and the last is watermarking so I'll start with writing styles and this one gets proposed either as something that humans can do or something that machines can do so we've probably observed that chat GPT and the like have a default voice that's very particular it's kind of bland it's kind of authoritative um our CEO described it as mansplaining as a service that I quite liked um so it's got a very particular voice and some people say I can spot that a mile off two problems with that first of all actually in practice um researchers saying that most people can't I know I can't I've been fooled by it many times when people have sent an article and then it says down the bottom I but also we can change the voice you you'll know you can get it to write like a pirate or something no essay is going to be written like that but it can change um software like gpt0 tried to measure this more scientifically looked at something called perplexity which is the randomness of the text and measured that um very easy to defeat lots of false positives the next mechanism is a more traditional if you're familiar with this sort of thing machine learning type approach so this is where you take a whole bunch of essays that a human has written or a whole bunch of texts that humans written you're right take a whole bunch of text that's written by AI or written by chat GPT you then train a model so that it can then when it sees new text either work out whether the whole text or just portions on the text were likely to be written by AI open AI had a crack at doing this and released a system and it was catastrophically success rate of about 27 um accuracy in terms of spotting loads of false positives if you tried it it was almost like a random number generator so either this is a really really hard problem or it wasn't in open ai's interest to actually get it to work very well we've probably all seen the figures from Turner in that are much higher in test scenarios so they've talked about a 97 accuracy it's worth thinking about that in a one percent false positive rate in controlled experiments so not be able to test turn it in one but any other one that we've come across again very easy to defeat just by changing the text so it no longer feels like it was written by AI the last is watermarking now I mentioning this because you may have heard it mentioned as the solution no products um today actually have watermarking um but there is some research in this space and they roughly proposed that what you could do is have a essentially a secret way of splitting the words that AI generates into two groups we call it a green group and a red group and what it will do is it will only select or almost always select words from the green group we won't know because we don't know how it's split and there are enough words in the language that um it will still read normally but you can see on the image on the right you can visually see on Watermark text against Watermark text yeah so you can even see as a human it looks different um it looks there are going to be problems with this one it looks like paraphrasing might just defeat it there's an argument so you have to paraphrase so much that that would um change the meaning second is that we sing a trickle down so it relies on the producer of the technology to release a model so we're relying on the person creating it to release tools to do it in the meantime these models trickling down so anyone can run them you can run a basic chat GPT actually on a small laptop or phone using technology for a few months ago so we'll see a whole Market of non-watermark text right and so each of these techniques can be defeated and they also give false positives okay so they will say text is written by AI when it's not and this can happen for a number of reasons some people actually just write like chat GPT you've all met them haven't you um so they will get flagged but it's also there are some nuances around the edge perhaps English as a second language some sort of text like um lists bullet points things like that just happen to look more like it was written by Chachi PT so we'll see detectors on the market uh we know it's coming in turnitin they will provide a bit of information but it's not conclusive can we ban it so we've already seen the answers no we never could okay um AI was already in word we maybe didn't think about it but grammar Checkers spell Checkers the generative summarizer all sorts of other things in word already used AI next word prediction that's been there for ages was AI couldn't have banned it we've probably all seen the Microsoft co-pilot announcements coming in it's going to be built throughout the whole thing so it's going to be the way we write if we start trying to investigate everything that something like now detector throws up as um it's academic this practice we are going to get into completely unfeasible situation that is my Whistle Stop eight minutes thank you thank you so much Michael um really clear and useful to to useful framing to sort of set us off I am going to hand straight over to our next speaker in the interest of Time Dr Phil Anthony from the University of Kent so Phil if you want to uh build on what Michael's already told us thank you I'll just quickly share my screen okay hopefully you can see that okay um so hi everyone thank you my name is Phil Anthony I am thrilled to be here today to talk about embracing chat GTP so hang on a second as you can tell from my first slide I'm going to argue that um I'm going to adjust the cons oh sorry hang on um as you can tell from my next slide over the next eight minutes I'm going to argue that as a sector we should be embracing um chat GTP basically and to back this up I'm going to start by addressing the concerns raised over the use of chat GTP in education and hopefully allay some of the fears um around this before moving on to the benefits so that's what I'm going to argue but what I really wanted to say today is that these ideas are not just my own I've formed them with the help of input from colleagues from across the sector um so where do I get my information from well you might recognize Michael here and as some of you may know I run the digitally enhanced education webinar series and we've recently hosted two webinars focusing on chat GTP and both of these webinars attracted around 1400 attendees each and I'm excited to share some of the insights that we gained from that webinar with you today during the webinar we also opened the padlet which you can again see on the slides here um so people could share resources ideas and voice their concerns and the paddler's been used really well with around 200 contributions so I'll try and drop the link into the chat so you can all have a look if you'd like to so before we can talk about embracing chat GTP I really think we need to acknowledge that lots of people have got concerns about it um about chat GTP or other large language modules um specifically in education now here's a list of the main concerns there are more but we've only got eight minutes and I've found via the webinars and the padlet board now we could fill the entire webinar today talking about each of these quite easily but we've only got eight minutes so I'm going to try and do my best to summarize them now so whenever chat GTP is discussed we've already kind of brought it up one of the first things that comes up in conversations is that students will use it to cheat however I would argue that if students were allowed to use this technology to help their learning with the caveat that their open and transparent with how they use it they're going to be less likely to use it to cheat after all why would they need to they don't they don't need to hide in again as we've already kind of discussed Microsoft kind of a four-star hand on this one now announcing recently that they're bringing this into all of the tools that we now provide our students so PowerPoint Excel um Microsoft teams all of this stuff which is called copilot so I believe the way forward has to involve an evaluation of this an evaluation of how we assess our students so for example looking at where written assignments really are needed and if allowing students to use this technology just isn't possible and they're where the there will be cases where that is the case then we need to have a look at the assessment type and maybe go for something that's a little harder for chat GTP to do and again there's lots of information around this so it could be reports on Independent research activities reflective responses that bring in personal insights um you know reflective portfolio style assessments that are built up over time these are all great assignment methods the students love doing it's really kind of you know the authentic assessment style you know it's the things they want to get stuck into so using these are actually harder to do with this technology anyway so maybe we need to have a rethink next up it's quite a big one so Chachi TP won't help students become independent and critical thinkers who can solve problems on their own I recently saw a quote that I thought was brilliant and that was that teaching critical thinking and problem solving isn't enough if we ignore the real world in which students will actually use those tools I think that's so important and as I pointed out in my last slide AI is being placed into all of these tools that we use in the workplace so surely we should be helping our students to learn how to use these tools in a critical way so how do we do this well I believe it has to start with having discussions with our students about these tools and helping them understand what it can do but also what it's not so good at now I understand that this is where kind of a whole load of stress comes in as the majority of us with the exception of Michael here are not experts in AI so having the thought or the thoughts of having these conversations with students could be really daunting especially if you feel that they're more clued up in it than we are now here is where institutional level guidance is really needed to help provide a framework for those discussions and you'll see on the screen here a screenshot of some student level guidance that I've created for students at Kent but we also have star fading guide star facing guidance as well to to help guide staff or at least give them something to go on I think that's so important so next up we have bias and hallucinations groups together now regarding bias Michael's done a great talk on this um it's important to note that chat GT's be uh gtp's responses are and I'll trip over on that one are generated based on the data it's it's trained on which can lead to biased responses now this link means that the large language model May reinforce or amplify stereotypes or provide skewed perspectives on certain topics so I think it's really crucial therefore to remind students about this potential bias when using chat GTP specifically in an education setting now on to my next point if you've had a play with chat GTP I'd encourage you to if you haven't you've probably experienced the hallucinations and found that sometimes it generates responses that are not factually but factually accurate or relevant in any way now it's important to note that this is a limitation of the tool that we need to make students aware of okay it's still a really cool tool and again linking to my previous Slide by doing this you can emphasize the importance of critical thinking and the decision making and most importantly responsible use of this technology and again that's where the conversations come in so this is a concern I've heard a lot at the webinars in the paddler and a lot by social media and in conversations now while chat ETP is a really cool tool and some of the other tools as well it's not going to replace us why well these tools they lack emotional intelligence don't they if we look at research from social psychology it's consistently showing that humans have a basic need for social connection and interaction I think we can all recognize that it's referred to as belongingness and it's linked to lots of positive outcomes such as increased motivation engagement and academic achievement Studies have also shown that students who have positive relations with their relationships with their teachers are going to be more likely to be motivated and to learn and perform better I think we can all kind of realize that so to put simply AI will not should not and indeed cannot replace the teacher because of what is unique to the teacher and that's our Humanity isn't it so I'm nearly done I promise regarding the benefits of these Live Language modules um I've listed a few here there are loads more um but I'm starting to run out of time so I think if we just go by the sheer number of people now signing up to use these tools I think we can see that there are clear benefits there but later on I'd encourage you to have a look at this video here um that was at the last webinar this is Ed Mahmoud he's the Kent Union president and in this talk he really highlights just how much Church UTP would have helped him during his time at University it's a really powerful message and I'd encourage you to have a look at it so in conclusion in education we aim to provide our students with the skills that they're going to require when entering workplace so why would we limit the use of chat GTP one of the most exciting advancements in technology we've seen in years I think I think it's so cool and I think most of us here today do so surely it's got to be our jobs to help students to learn how to use these tools so let's make that amazing thank you so much Phil um and speaking of the student perspective actually our next speaker we have the brilliant Stella Maris who is the director's assessor at St Andrews University and is on qa's student strategic advisory committee so Stella if I can hand over to you that would be fantastic so um can you guys see my screen we can see something in front there you go perfect yes right so you can see the PowerPoint yes and nothing else good yeah okay awesome um Michael said my name is Stella Maris and I'm director's assistant University of Saint Andrews I will be covering how we Foster better relationships of trust and partnership with students in the process of implementing chat GPT which as has already been established is probably unbannable um I'd also like to note that I think a lot of the points that have already been raised I've noticed I've raised as well um hopefully there isn't too much repetition so um as we continue to navigate the rapidly evolving landscape of technology and education it's important we consider how we can effectively incorporate those new tools and platforms into our learning spaces this can help us better engage with and support our student Nation um one technology that has been gaining Traction in recent years is our chat Bots and conversational AI with chat GPT being a particularly notable example and the one that we're focusing on today so um we I'm going to start off by discussing how we navigate the technological landscape in education so despite growing up in a time of Rapid technological Advance development I'm still astounded by the fact that we now have access to Technologies like chat GPT which just a few years ago might have been thought of as something only future Generations would benefit from decades or even centuries from now um though I consider myself among the uninitiated in the rapidly developing technology landscape technological landscape I can't help but feel both excited and nervous about how often I find myself reevaluating my own perceptions of our species capacity for Ingenuity and inventiveness it's so thrilling and daunting to witness the limits of what we once thought possible being pushed further and further every day as Albert Camus once stated our contemporary science has progressed to such an extent that it's useless to be unscientific with every new height we reach our species undermines and ridicules the entire concept of self-imposed limitations and I'm seriously considering abandoning the practice altogether I genuinely believe that the biggest mistake the higher education sector can make is to underestimate progress in the technological sphere our institutions are supposed to provide pedagogy that propels us to the Forefront of innovation and progress as John Wheeler once famously stated our challenge is to keep alive the sense of the possible to keep alive the sense of the mystery that has always been there the current iteration of chat GPT represents the beginning stages of what's likely to continue to be developed into an even more powerful pool that could revolutionize the way we teach and learn with future updates bearing little resemblance to our current pathological practices perhaps fear be the enemy of progress is to consign higher education to a cycle of consuming our own potential in the pursuit of protecting ourselves from the possibility of diminishing relevance so how do we Embrace chat GPT in higher education well to get the most out of chat GPT we must we have to see it not only as a tool for educators but also as a tool that can be co-designed and co-implemented with our student community by working in collaboration with students increasing their confidence and organ tensions we can more efficiently adapt the tool to meet their needs and provide them with personalized support for success after all students will be the ones using these tools on a daily basis and will be best placed to know what would be most beneficial for them as learners this collaborative approach also gives us the chance to gain insight into how students May intuitively respond to the tool such as what kind of questions are likely to ask what prompts they consider to be academic dishonesty and what kind of Engagement would the tool gives greater accessibility to information while still stimulating and challenging them intellectually we need to rethink what we Define as academic dishonesty in light of these technological advancements especially as they perform more powerful and prevalent as these tools evolve they may circumvent and have been explained it's been explained how they circumvent current methods of detecting academic misconduct and it's crucial that we have an open dialogue with students to better understand their perspectives on the ethical use of these tools so how do we involve students in co-designing and implementation so partnering with students in their development and implementation process of chat gbte can increase their engagement with the technology and the sector as a whole by placing students at the center of their learning we can create a more personalized and focused approach to education which benefits academic success and future career possible um we also misrecognized that the use of artificial intelligence does not reflect the students intelligence or capability rather AI models like Chachi PT can be valuable tools for increasing accessibility to education making it easier for students to engage in information and learn about the world around them this is especially beneficial for students with neurodiverse profiles who may struggle with traditional methods of learning and so by redefining our approach to academic Integrity in the context of these new technologies and partnering with students in their development and implementation we can create a more inclusive and empowering educational experience for all learners incorporating chat Bots and conversational AI into our education system doesn't have to be seen as a threat to traditional teaching instead it's an opportunity to enhance and supplement existing practices in order to better meet the needs of students for more backgrounds and learning profiles with the use of these tools we can create a more personalized and engaging learning experience one that can help break down the barriers that have traditionally made education and accessible to many individuals so how do we so sorry I missed out a section on this but this this section is on enhancing accessibility with AI modules and not concerns just yet by reducing the workload of Educators and updating course content GPT can allow for greater attention to be paid to incorporating diverse perspectives and ideas thereby contributing to a more inclusive and Equitable education system that values the experiences and perspectives of all students in addition the 24 7 availability of chat GPT can provide support for students who may face additional barriers to assessing those educational resources so students who have work or family responsibilities or disabilities that thoughts can also help bridge the digital divide by providing access to educational resources for students who may not have access to traditional resources or support systems so students in underfunded schools or rural areas or International students who may face language barriers or cultural differences and the integration of those Technologies in education can as I said earlier promote equity and Social Justice by reducing barriers to access and providing support to students who may face additional challenges so access to Quality education should not be limited to those with the means to pay for it sorry I don't know why it's drawing on the screen I'm going to ignore that um and nor should it be reserved exclusively for those who Excel within a particular framework of learning by embracing technological advances like Tech TPT we have an opportunity to reimagine education as a more inclusive and accessible space for all Learners and that's not to say that traditional pedagogy should be abandoned entirely but rather that it should be adapted and complemented with using Tools in methods I'd like to acknowledge that advocating for embracing chat Bots and conversational AI in higher education is not to ignore the reality that that technology can be a double-edged sword while other tools can offer numerous benefits including increased efficiency and personalized learning experiences they can also present challenges and potential ethical concerns that warrant examination and mitigation so how do we address ethical concerns so things like privacy security and reliability so on privacy and security concerns institutions can Implement strict policies and protocols that prioritize data protection while also ensuring transparency in how student data is collected stored and used additionally they can provide students with information on their rights including their ability to control the types of data that are collected and how that data is used and also provide them with the ability to opt out of certain data collection processes to address concerns about reliability and accuracy of tech GPT they can provide students with their information on how the AI powered system works and the research and development process that has gone into creating them so that can include details on the types of data that are used to train the system the methods used to validate its accuracy and the ongoing efforts to improve its performance um some as was raised earlier May worry about chat GPT replacing human instructions instructors sorry or reduce the quality of their education but institutions can communicate that the AI based platform is intended to complement rather than replace human instructors and that the primary role of chat is to provide additional support to students not to replace the important interactions and insights the human instructors can provide so institutions can also emphasize the ways that GPT can provide personalized responses to student queries even outside of typical office hours can I do a time check really quickly 30 seconds is that all right I might need a bit more than 30 seconds but I will break through the rest sorry um okay so I'll move on to the section on redefining academic dishonestly so a lot of the public discourse soundtrack GPT has centered on anxiety is about its potential to facilitate teaching I'm exciting this framing of the issue demonstrates a fundamental mistrust of students that I think is counter-intuitive to establishing that Foundation of trust and partnership instead we need to consider how we can work with students to promote academic Integrity while also leveraging the benefits of chat GPT for learning institutions can Implement robust measures for detecting academic misconduct so as mentioned earlier plagiarism detection software or reviewing assessments for inconsistencies and irregularity but we need to recognize that no system is foolproof another way to combat concerns about academic misconduct is by providing clear guidance on how chat GPT can and should be used so that can involve creating a set of best practices for integrating the technology into learning and assessment progress processes and communicating these guidelines effectively to students might also be beneficial to involve students in the development of those guidelines um I don't really have a lot of time to run through the rest of it um but I want to stress at the end of this that I do think that the benefits of Chachi PT are significant and by embracing technological advancements like chat GPT institutions can create a more inclusive and accessible educational environment that prioritizes learning as a as a societal good sorry that was quite fast fantastic no thank you so much Stella a really compelling and persuasive case for why this has to be done in partnership with students um and some really brilliant ideas in there so thank you for that um up next our penultimate speaker is Dr Irene Glenn dinning who is academic Integrity lead for Coventry University group so Irene to ban or not to ban thank you very much and thank you for inviting me to come along this morning in my eight minutes I'm going to try and share with you something about what I've been working on with my colleagues for the last 12 months um can you see my screen just to check first of all I hope so okay so um without further Ado what I'm going to do is to focus in on um the the um academic Integrity side of of the discussions but before I do that I want to try and broaden the discussion this is not just about one tool it's not just about chat GPT there are many different types of tools that use artificial intelligence and they've been with us for a very long time they're still with us they're going to be with us forever we know that so what I have on this slide are but roughly divided into four types of different kinds of tools that we're talking about so first of all we have uh type one is tools that Trump that translate or transfer between media uh the different kinds of uh text so we're looking at text to begin with um so speech to text tools text-to-speech tools predictive text also translation tools and like colleagues and I have long been concerned about student misuse of translation tools using translation where they should be writing in in English so that's the first time so we need to think about what is what we should permit or what we shouldn't permit in terms of use of those types of tools zero point in Banning these tools and with us have been with us a very long time the second type of things like spell Checkers paraphrases Granite Checkers again they've been with us forever um and they will they're getting better and better students can misuse these tools so so you need to be very clear with students what they're allowed to do and what they should not be doing and where that red line lines type three is what we've been talking about so far in this morning's um session essay box text generators and we have to remember it's not just a fact check GPT it's there are lots of these these tools um at quill bot perplexity lots of different ones and they're getting better and better they're still in their infancy but they're already very very good type four that we haven't talked about is all these other kinds of generators um they can generate programming code they can generate graphical objects artworks music solve math problems so it's not just confined to text generation so I think that's what I'd like to start with by just broadening the discussion into those different types of tools and all of these tools uh well most of these tools there is the potential to misuse the tools as well as to use them in a very positive way and we really need to think very carefully with our students and discuss it with our students um what is appropriate use and what is inappropriate use of these tools so what we did about a year ago a group of colleagues and I started to think it was really paraphrasing tools that we we started to think about so we decided we we really needed to do some investigation to find out more about the different types of tools and then that led into a consultation with the Whitey university community and now we're in the process of generating the guidance of staff and students and also making Necessary changes to regulation but even by September I realized that there was a need to do something more than just wait until we got all the guidance together so I put in place two what what I would like to turn was kind of emergency uh additions to our regulations um so these are two definitions of academic misconduct that I put in place in September 22. using technological AIDS and artificial intelligence including various types of tools without specific authorization and this was um you know perhaps a slight knee-jerk reaction but that seemed to be a really appropriate thing to do and I I stand by it I think this is actually quite a quite a a measured way of actually um talking to students and talking to staff about what it is we need to do to to control the use in in certain way and the second one is just specifically looking at translation tools whereas a assignment is expected to be written or required to be written in English if a student uses translation software or gets assistance from a third party to write it in their native language and then convert it into English that could well be considered to be academic misconduct so that was my response from September and because I only get a chance once a year to um to uh um change the regulations that I had to do in September and I may well have a kind of slightly more newer history of those for for the next uh edition of our regulations in September so we've been running focus groups now um since since January um and on these are this is just a very very brief uh list of things that we're finding um we haven't done the full analysis yet but first of all clearly as everyone's been saying banny's Banning is not possible we must learn to live with these tools in particular we need to provide guidance on what we consider to be ethical use of the artificial intelligence tools where where the red lines are where it becomes this condo but also we need to embrace them because students will be using these in the workplaces has already been said by by other colleagues on the panel we found that awareness among staff is mixed a lot the people coming forward and asking you to be involved in the focus groups generally tend to be very clued up on what's going on but a lot of staff are really quite worried and don't know a lot about it so very mixed we all we discovered that um one part of the group had all already started to do some research into the tools and they were particularly looking at the detection of artificial intelligence use and and they produced a really useful report so now we're working together with with that group of educational technologists there are different viewpoints in allowing um allowing the use and not and inevitably some staff came back and said we need to move back to exams and those kind of points of view um but we had a very useful discussion on that about the types of assessment that we should be working towards unfortunately um we're already working on a curriculum transformation and assessment transformation project in the University which is is actually addressing a lot of the points that we need to address um what was come up a lot of times is we need to think very much about um students with special learning needs and also with staff so Banning these tools would would actually um disadvantage all of those people so we need to be helping students to get get access to these tools and staff as well who would benefit from them uh clearly assessment design but it's also very similar to what required of contract cheating and there's a lot of overlap between what we've been doing to address contract cheating and and the artificial intelligence tools it came out that we really need to actually communicate with staff and students and and that clearly we've been doing and we need to do that more continue to do that as we go through um in terms of the guidelines I'm going to show you some initial guidelines that I've come up with um but we um as has already been said we'll perhaps have some scenarios and some examples on types of use but we don't want to ban them we don't want to kind of restrict the appropriate use of these and and of course it's also looking at the possibilities for detection um okay so staff guidance that's um on the use of them students should um be looking at what types of tools they may use and for what purpose uh what they may not use and why not and explain to students how they could how they should acknowledge and and what content um how they can demonstrate what content is their own content um and you know it's possible to produce a kind of template to be used in terms of students the opposite student if a student wishes to use tools they should gain permission from the module leader or supervisor specific use adding to deep water in terms of academic misconduct so um detection we've talked about a little bit and detection as it was said by the first speaker is is not a um lots of so-called detectors out there but none of them work particularly well that may change with with time um in terms of the checklist um I think most of these points have been covered we are picking cases up already um it's very difficult to prove them um but there are some it's very similar to the checklist that we use for contract teaching surprisingly um but as already been said this is not foolproof it's not a hundred percent but it does give some some Clues um so and then I'm going to stop so I'm sorry I'm slightly overrun but uh hopefully that was useful and compliments fantastic yes thank you so much Irene and really useful to hear um those Reflections sort of in context of what's actually happening in your institution as well so so that's really useful to kind of Bring It um into into kind of context for us so last but by no means least joining us from the other side of the world uh we have the fabulous Dr Bronwyn eager from the University of Tasmania so Bron take us home thank you for having me and I I mean I loved what you were just saying and I'm so excited that the conversation we solved the question of whether it should be banned in the first five minutes of the presentations tonight um and we can kind of move forward which is a delight to me because I'm obsessed with using these tools and like Irene was saying it's not just a GPT question it's so much bigger than that like I'm using it for writing for images for generating podcasts videos websites it's just embedded across so many different use cases and it's getting more sophisticated and it's um I think 3 000 tools was the article that I saw the other day of um of how many were out there that you could possibly use but I've only got eight minutes so there's three things tonight or tonight for me this morning for you guys for what I would like to reflect on and they are water a tweet that I saw and also a timeline and the first one is the famous commencement speech that was given by the author David Foster Wallace and if you're not familiar with it it's essentially that it starts off by saying there were two young fish swimming along in the water and they passed an older fish and as they went past the older fish the older fish said to them good morning boys how's the water and the two fish kept on swimming and then they turned to each other and said what the hell is water and I think we've kind of got to a point where we've been going through our journey and we haven't actually really noticed that AI has been everywhere and all around us for a very long time um and we just haven't really been paying attention to it myself included that definitely has been that some people um but I think the majority of us were going through the world without realizing that AI was already embedded across most of our devices um and until November last year AI was kind of like water and the other thing that I reflected on this week was a tweet that someone posted in response to this session actually which was saying why are we still talking about this like why haven't we moved on already from the should we ban and should we not ban conversation um and I think it's really easy to lose perspective when you're in this conversation all the time about the timeline of adoption of the technology um so chat GPT is kind of the mass Market consumer version of AI That's caused all this disruption because it is so easy to use um and it feels like it's been years since that technology first dropped in November given the journey that that we've all been on that's actually only been a few months really um and it's only been a few months in a higher education um Global conversation about what's happening um a huge percentage of students and Educators and people in the higher education sector haven't logged on to chat GPT or haven't consciously used an AI tool in their work process and I think we need to remember that there is an adoption curve and if you're in this session you're probably quite quite far ahead along in it um so I think we kind of have a responsibility to keep having these conversations to bring people along with us to go through the trajectory or through the Journey that many of us have have taken ourselves and the last thing in the last few minutes that I've got is is about Journeys as well and it's the Journey of AI and digital literacies and I was reflecting on even just in my immediate family like my father he was a a typewriter native because that was the technology that he was using so he was native to typewriters he adopted computers he got very excited about electronic typewriters and then that became normalized for him and then in My Generation Um I wasn't quite native to computers but I was native to the electrical typewriter or the digital typewriter and to graphical calculators and I saw some comments in the chat before Oh My Gosh the stir that the graphical calculator caused in when I was a student um enormous comical now um but I think with my father's generation and my generation we had a really slow adoption of these Technologies in our lifetimes which were not necessarily getting with the current um speed at which things are moving um and then I look at my students who were digital natives or they're called digital natives my younger students at least um and they don't know a world without a smartphone in it they can't imagine going through life without the instant communication um but yet many of them don't know about basic filing systems in computers or saving a file to a drive they just save it all in one spot and then search for what they they need um and then I look at my niece and my nephew who are growing up in a generation where they think that they can pinch images on magazines and they will enlarge um and I think they've grown up they will grow up as AI natives um but just like the students who don't understand the basic filing structures of computers I think it's really important that we teach the younger generations and the older Generations the underlying kind of why this how this technology works and coming back to the issues of bias and equity and stuff that are actually built into these Technologies um and just just like the files and I think a i is it's almost if we were in danger of eventually treating it like the two fish that are swimming with the water and not actually consciously engaging with it and I think it's essential that we engage with it because these skills are going to be essential workplace skills they're just going to be Baseline skills like knowing how to send an email or um using like basic Microsoft Word programs I think that's where the future should be so to the question of should we ban or not bad I think it is overwhelmingly been answered by the panel um and my my Approach would be I don't think we should even be thinking about banning or not Banning which I think we should be thinking about educating um so that people have a choice and they can consciously engage with these Technologies um and make deliberate decisions about whether they want to in the first place and if they do want to how they actually engage with them um and that we don't just whip into a world where ai ai is just ubiquitous and swoop into David Foster Wallace's water water world and that's me thank you amazing thank you so much Brian and a really sort of nice I guess reflective set of remarks to kind of finish us off remind us that AI has always been sort of creeping into our everyday and though this can seem alarming it it will just become part of us and it's you know the panel as you said are unanimous on that um so there we have it five perspectives onto battle to ban all pretty unanimous in their um in their conclusions which is AI is here to stay um the question is how we use it as a Force for good which coincidentally is the title of our next webinar which is next Friday but right now we've got I'd say maybe five six minutes um to cover off a few questions from the audience so my colleague Helena has been keeping a BDI on the chat which looking at the sort of volume of messages in there is quite the task so Helena what questions do we have yeah it's been a very very engaged chat which has been really interesting to read and a lot of support for the idea of embracing um AI I I'm gonna just do a round of three and if we've got time maybe do some more afterwards so sorry if we don't get to your question um but it's a question from Stephen on do you see the dangers of chat gbt damaging The Learning Experience so outside of assessment validity what about the process of reading and researching and synthesizing writing an essay do you think AI damages that and we've had a question from Shaylee about whether colleagues have come across any students who are really skeptical about using it and how you've approached those conversations if there's students that don't want to kind of engage with it and finally a question from George that basically says if if we're going to ask students to declare this is that not counter-intuitive if this is about to become as kind of ubiquitous as spell checker is to a student writing um an assessment amazing so three questions there if you're able to sort of summarize them and put them in the chat for people's records but what I might do panel is ask you each in turn and feel free to just respond to one that is particularly jumps out to you or you can give a couple of Reflections on a couple of the questions if you want to but I'll just maybe go in the same order that you spoke so Michael do you want to um offer some Reflections on any or all of those questions first yeah I think I'll pick up on the last one about whether it's feasible to get students to declare all use of AI I I think it's not I think we're going to have to talk about behaviors instead um people just aren't going to be aware of whether something is AIA or not it's going to become kind of so normal the war challenge analogy is brilliant I use that one again um so yeah I think we're just gonna have to think of different ways of describing this yeah great thank you Michael Phil I'll come to you next what are your Reflections on those questions oh yeah thank you I did I reply to a couple of those questions in the chat but just to quickly summarize if I take the the over-reliance kind of question I do have a concern I think the concern I have is that students will become over-reliant on the technology and I think that can I think come back to bite them sometimes maybe later on so typically assignments should be set up on a module that may be Progressive so students have to use kind of they learn skills they learn facts and things in the in the first assignments that they will need to use later on now if students rely on AI technology too much in those initial assignments then that could come back to bite them a bit later on so I think that's where kind of clear guidance and communication comes in again just to remind students of that and there's there's various different ways that you can do that one way that I have with our student guidance is that I've mentioned things like illusion of confidence where it's where a student may think they know something about a topic or know a lot about it they make like you know think that they do but in reality they've used the AI so much to lean on that they actually maybe don't know as much as they think they'd know so when it comes to an assignment where it's not possible to use this AI they suddenly go oh oh dear and they can kind of get themselves into hot water quite quickly and I think the earlier that we can get that communication across to students the less likely it is to happen but I think for me that's the main one that I kind of not worry about but I think it's something I think is important to get across early on I think we can all agree that how useful it is and think you know in the space of time we've had it just using it for work if it was now taken away I think some of us would be like oh man I could really use that tool right now to help me do this or to you know to to create that presentation or whatever it is so yeah thank you that's really helpful Stella your Reflections on any all of those questions um can you remind me of the questions I have very bad very bad attention Okay I can see the the questions in the chat so um I think outside of the learning experience the talk that I gave um focused a lot on ways that AI could be used to actually enhance the learning experience I think particularly for students with poor processing um or learning difficulties as someone who has several myself I've always felt a bit like um political damage in the pursuit of maintaining additional teaching styles and I think that the extent to which having a tool that helps you process information could help those students achieve what they couldn't otherwise is something that I think it's worth looking into and considering um I think that and this is something that was raised in my university in the context of developing our digital strategy um seeing how quickly this technology is developing and seeing that the current tools we have aren't very well placed to actually identify the use of this as well um it's really kind of jarring I think especially if you are you weren't kind of um born into a um a time where technology was as rampant as it is now the kind of Technology natives where we for those of us in my generation we kind of grew up having to adapt to um technology changing at a really rapid Pace not as quick as this um but still incredibly quickly um and I think that the concerns around you know how do we balance um you know the the benefits we can get especially for students and from neurodiverse backgrounds with the potential for cheating and the potential for this technology to continue to move it you know incredible speeds um needs to be considered in the context of maybe younger Generations are better placed to make that determination because they have that experience of adapting to those platforms so it's about including students and young people more in the conversation providing governance expertise providing expertise in terms of assessment and incorporating their their perspectives on how we adapt to the technology fantastic thank you Irene your thoughts on the questions I can see we're almost out of time so I'll be very brief thank you I respond to the first one about damaging the learning process I think it's really important that um that Educators set set the the um the rules in terms of students having to learn the basics first before they start to use the tools to help them so if I take my my own subject of computer science you can't write it you can't kind of you can't um maintain a computer program if you don't know how to program um and therefore you have to learn the basics first before you can then start to use the tools because we know that these tools are not always accurate and unreliable so I'm going to leave it there and and prep prominent coming thank you yes Braun final word I've got two seconds so mine is I'm not changing any Assessments in response to AI I'm just changing expectations about what is now possible with partnering with these tools fantastic thank you all right well we are out of time brings us to the end of the webinar um I want to say a quick apology to those who wanted to log on at the start and couldn't the qaa team uh hopped on the phone to zoom to allow you to join which is why the numbers flicked up Midway through that webinar we will be making this recording available so that you can catch up on what you missed big thank you to our brilliant speakers on providing their perspectives um who have you know shed a light on lots of the issues that I know that we are all grappling with um remember there are two more uh webinars over the next few weeks one is next Friday the next is after Easter one is about how do we use it for as a Force for good so the natural next step in the conversation the one after Easter is about what assessment should look like the links for those are um on our website but also will be put in the chat if you're a QA member your institution has access to lots of resources on the membership resources site um that help with things like how you design assessment authentic assessment promoting academic Integrity if you're not sure whether you're a member there's a list of those on our website and if your institution isn't a member and you would like to be then please do reach out to us the best email for that is membership at qaa.ac.uk and we can have a conversation about what membership might be able to do for your organization so that is all from us we are one minute over thank you so much everybody for attending we hope to see lots of you back for the next two webinars but in the meantime enjoy the rest of your day thank you very much foreign

Channel: QAAtube

Views: 3,876

Keywords:

Id: FPY0at2m69Q

Length: 58min 15sec (3495 seconds)

Published: Wed Mar 22 2023