WEBVTT 00:00:06.000 --> 00:00:06.900 Good morning from Wales, UK! 00:00:21.000 --> 00:00:21.900 Kaya from Western Australia! 00:00:28.000 --> 00:00:28.900 Hello everyone, good evening from Townsville, Queensland AU 00:00:31.000 --> 00:00:31.900 Hello from Wurundjeri lands, Dandenong Ranges, Victoria 🙂 (Swinburne) 00:00:32.000 --> 00:00:32.900 Hi from eastern Sydney in Australia on lands of the Bidjigal and Gadigal peoples of the Eora nation. 00:00:48.000 --> 00:00:48.900 Hi from sunny Dudley in the West Midlands 00:01:43.000 --> 00:01:43.900 Hi from Auckland! 00:01:53.000 --> 00:01:53.900 Hi from Adelaide 00:01:56.000 --> 00:01:56.900 Good morning from Hatfield, England! 00:02:03.000 --> 00:02:03.900 Hi from wet Sydney! 00:02:15.000 --> 00:02:15.900 Yes 00:02:16.000 --> 00:02:16.900 yes! 00:02:17.000 --> 00:02:17.900 Yep. 00:02:20.000 --> 00:02:20.900 yes 00:02:22.000 --> 00:02:22.900 yes 00:02:22.000 --> 00:02:22.900 Yep 00:02:27.000 --> 00:02:27.900 Yes 00:02:30.000 --> 00:02:30.900 Yes 00:02:31.000 --> 00:02:31.900 marginally.. would like detail 00:02:34.000 --> 00:02:34.900 AI does a job interview! 00:02:35.000 --> 00:02:35.900 Yes 00:02:39.000 --> 00:02:39.900 didn't see the news... 00:03:13.000 --> 00:03:13.900 I have marked a few deeply fake assignments this week! 00:03:29.000 --> 00:03:29.900 😆 00:03:40.000 --> 00:03:40.900 https://www.youtube.com/watch?v=oxXpB9pSETo 00:03:45.000 --> 00:03:45.900 https://www.theguardian.com/technology/article/2024/jun/05/ai-researchers-build-future-self-chatbot-to-inspire-wise-life-choices?CMP=share_btn_url 00:04:50.000 --> 00:04:50.900 For those just joining, here is the link: https://www.youtube.com/watch?v=oxXpB9pSETo 00:05:18.000 --> 00:05:18.900 Scary! 00:05:26.000 --> 00:05:26.900 Cool! 00:05:33.000 --> 00:05:33.900 creepy! 00:05:45.000 --> 00:05:45.900 would fool me 00:05:59.000 --> 00:05:59.900 it looked slightly off but if you are only casually paying attention it would pass. 00:06:01.000 --> 00:06:01.900 It was close but not quite right. 00:06:04.000 --> 00:06:04.900 What "is" Morgan Freeman? 00:06:16.000 --> 00:06:16.900 I have seen a few of them now, and while you can tell but it could easily be some sort of weird lag on the video? 00:06:37.000 --> 00:06:37.900 Imagine students using it to do their recorded presentations.. 00:07:05.000 --> 00:07:05.900 @Laurine Hurley exactly 00:07:07.000 --> 00:07:07.900 @Laurine Hurley don't jump the gun 😉 00:07:22.000 --> 00:07:22.900 I know Morgan Freeman's voice much better than I would most students 00:07:46.000 --> 00:07:46.900 https://link.springer.com/article/10.1007/s10462-024-10810-6 00:07:50.000 --> 00:07:50.900 Deepfakes of real people are being criminalised but how do you enforce these laws? 00:07:55.000 --> 00:07:55.900 👍 Andrea 00:08:11.000 --> 00:08:11.900 I have been reflecting on deepfake pedagogy and applications and implications. Taking me to places of very 'deep' thinking. 🙂 00:08:12.000 --> 00:08:12.900 @Andrea 😆 00:08:27.000 --> 00:08:27.900 Seeing is believing? 00:08:37.000 --> 00:08:37.900 This just announced in Australia https://ministers.ag.gov.au/media-centre/new-criminal-laws-combat-sexually-explicit-deepfakes-05-06-2024 00:08:58.000 --> 00:08:58.900 It has been happening in the Indian elections just past. 00:09:14.000 --> 00:09:14.900 @Andrew Yes, quite a few cases of countries bringing laws in to restrict the use of these 00:09:59.000 --> 00:09:59.900 it's really scary thinking about the US election, where many already seems to have lost touch with "real" reality, and could believe much 00:11:05.000 --> 00:11:05.900 It takes the meaning of Dont believe your lying eyes to a new level 00:11:22.000 --> 00:11:22.900 2 00:11:32.000 --> 00:11:32.900 @JCobham yes 00:11:34.000 --> 00:11:34.900 VASA 1 video Link: https://www.youtube.com/watch?v=fTAuzFzMt5Y 00:12:00.000 --> 00:12:00.900 Watermarking sounds good in theory but while the tech giants may comply there are plenty of rogue actors who will ignore any regulations https://www.theverge.com/2024/2/13/24067991/watermark-generative-ai-deepfake-copyright 00:12:47.000 --> 00:12:47.900 Creepy eyes in those people 00:13:07.000 --> 00:13:07.900 the facial expressions! 00:13:29.000 --> 00:13:29.900 impressive 00:13:35.000 --> 00:13:35.900 There was talk of avatars that could attend Zoom meetings and even make decisions for you so you don't have to attend 🙂 00:13:57.000 --> 00:13:57.900 That's the dream 😉 00:14:09.000 --> 00:14:09.900 @matt.. I reckon that's been happening, given some decisions I've seen!😩 00:14:15.000 --> 00:14:15.900 I have seen a digital panel member at a conference instead of the real person 00:14:17.000 --> 00:14:17.900 It's there now. 00:14:38.000 --> 00:14:38.900 wait until the scam factories start sending your grandparents video calls featuring your uni profile page as a video to ask for money. 00:14:45.000 --> 00:14:45.900 also know a conference speaker whose digital twin presented for her while she watched online at home with a glass of wine 00:14:48.000 --> 00:14:48.900 @Laurine Hurley - we could try it! 00:15:08.000 --> 00:15:08.900 beats the old "Mum, I lost my phone" tetx scam... 00:15:12.000 --> 00:15:12.900 Why do these avatars show so many teeth while talking? Seen similar with https://www.oiedu.co.uk/ai/ 00:15:20.000 --> 00:15:20.900 @Jacqui Young 😆 00:15:56.000 --> 00:15:56.900 @Carol Bailey it's the looking too much at the camera without blinking much that gave me the creeps. 00:16:18.000 --> 00:16:18.900 Can deepfakes support mental health needs in some cases? 00:16:29.000 --> 00:16:29.900 @Emma Duke-Williams (Dundee) 👍 00:17:01.000 --> 00:17:01.900 I can see this having benefits for the many who freeze when speaking public/class.. those with English as additional language who feel self-conscious... 00:17:21.000 --> 00:17:21.900 I created a talking digital version of myself this morning to narrate slides on some pre-lecture study material. 70% satisfied with result (from a free app) 00:17:24.000 --> 00:17:24.900 I had a student asking me recently about using a text-to-speech tool for their presentation as they have a speech disorder. We approved it, as long as they told their UC and added an acknowledgement for the software. 00:17:49.000 --> 00:17:49.900 I would if I could change my face and voice...😍 00:17:54.000 --> 00:17:54.900 @Geri Harris which app?? 00:17:55.000 --> 00:17:55.900 It comes down to what you are marking against in a rubric. 00:17:57.000 --> 00:17:57.900 and the students will do the same. so we have deepfake lecture talking to deepfake students. 00:18:22.000 --> 00:18:22.900 I already have a student with a very good digital twin of himself 00:18:41.000 --> 00:18:41.900 @Mathew Hillier then have a GEnAI essay graded by AI ... 00:18:48.000 --> 00:18:48.900 yep! 00:18:58.000 --> 00:18:58.900 @Carol Bailey Vidnoz 00:19:32.000 --> 00:19:32.900 We're changing our recording policy to reflect ownership of avatar rights of academics once they leave the University 00:19:36.000 --> 00:19:36.900 and as we are being increasingly driven (in some places anyway) to have polished presentations (no idiosyncrasies...), this may have an imperative that many of us don't actually want 00:19:36.000 --> 00:19:36.900 If it means you don't have to do your hair and makeup... 00:19:55.000 --> 00:19:55.900 Will giving authentic real-time presence will be as rare as receiving a snail mail letter? 🙂 00:20:08.000 --> 00:20:08.900 @Amy Milka .... 00:20:25.000 --> 00:20:25.900 @Mathew Hillier I am genuinely struggling to find assessment methods that truly protect academic integrity. I really am trying everything! Short of asking every student to participate in a face to face interview I do not know! 00:21:02.000 --> 00:21:02.900 incentivising disclosure of AI usage by including criteria on rubrics is helping but only a little 00:21:06.000 --> 00:21:06.900 But it's clear that it's an AI assistant that's supporting the students. 00:21:15.000 --> 00:21:15.900 Does all of this just compound the ongoing depersonalisation of education? 00:21:18.000 --> 00:21:18.900 @Geri Harris yes it is becoming challenging. online interactive oral assessments were a boon during the pandemic, but now with deepfakes those are now less safe. 00:21:45.000 --> 00:21:45.900 @Natalie Lafferty #2 It certainly feels like it. 00:21:46.000 --> 00:21:46.900 @Emma Duke-Williams (Dundee) good point. Transparency.... 00:22:43.000 --> 00:22:43.900 In the future, would you have a choice of AI Assistants to explain in other languages? 00:23:17.000 --> 00:23:17.900 @Emma Duke-Williams (Dundee) , I believe this is already available 00:23:33.000 --> 00:23:33.900 I meant BUV's not the software generally. 00:23:36.000 --> 00:23:36.900 conversational interfaces are going to revolutionise how we think of user experience in software and web design 00:23:52.000 --> 00:23:52.900 JISC ideas for assessments in an AI enabled World. https://repository.jisc.ac.uk/9234/1/assessment-ideas-for-an-ai-enabled-world.pptx 00:24:02.000 --> 00:24:02.900 @Geri Harris I think we have to rethink what we are trying to do with assessment. We can't map existing approaches to these technologies. 00:24:08.000 --> 00:24:08.900 if it relieves us of remote proctoring, I'm (a bit) all for it! 00:24:17.000 --> 00:24:17.900 I wonder if students will have better engagement with synthetic media format feedback? 00:25:07.000 --> 00:25:07.900 @Rachel Forsyth (Lund, Sweden) Hear, hear! Rachel. Let us break the paradigms of assessment. 00:26:54.000 --> 00:26:54.900 well, It'd be better quality than the sloppy one my student did for a placement...not that this is good! 00:27:28.000 --> 00:27:28.900 Written reports are no longer authentic, sorry. I have raised 11 red flags for case study assessments I’ve just marked where the answers are clearly AI generated text and visuals. When the students write to me via email that is NOT how they write. It’s the same as contract cheating 00:27:29.000 --> 00:27:29.900 What impact do deepfakes have on audio authentication technologies which are currently being adopted by banks, telcos and call centres? 00:27:32.000 --> 00:27:32.900 forged report 00:27:55.000 --> 00:27:55.900 can a "voiceprint" be faked? 00:28:03.000 --> 00:28:03.900 @Geri Harris yes, chalk and cheese 00:28:30.000 --> 00:28:30.900 oh dear! 00:28:46.000 --> 00:28:46.900 https://bankingjournal.aba.com/2024/02/challenges-in-voice-biometrics-vulnerabilities-in-the-age-of-deepfakes/ 00:28:55.000 --> 00:28:55.900 @Emma Duke-Williams (Dundee) Yes, we are exploring creating these in other languages as well-very easy to do through the app. However, the biggest benefit is probably the ease of closed captions and making these correct 00:29:40.000 --> 00:29:40.900 👍 @Mike Perkins 00:30:01.000 --> 00:30:01.900 back to old school oxbridge discussions with your tutor as an assessment approach!? 00:30:27.000 --> 00:30:27.900 I have 100 students - don't really fancy 100 face to face interviews! 😮 00:30:31.000 --> 00:30:31.900 (face to face) 00:30:52.000 --> 00:30:52.900 @Mathew Hillier I can see the beancounters LOVING that.. 00:30:58.000 --> 00:30:58.900 small group discussions with 5 students instead of marking 5 essays? 00:31:12.000 --> 00:31:12.900 Geri I agree - what's the answer though is the question 00:31:15.000 --> 00:31:15.900 I like that idea @Mathew Hillier 00:31:24.000 --> 00:31:24.900 training to spot deepfakes is only a temporary measure till the technology improves and it becomes impossible to do so... 00:31:34.000 --> 00:31:34.900 that is a good idea Mathew - my only concern would be whether it has inclusion issues 00:31:50.000 --> 00:31:50.900 Griffith's interactive oral assessment as scale just keeps looking more attractive 00:31:52.000 --> 00:31:52.900 https://www.spotdeepfakes.org/en-US 00:31:52.000 --> 00:31:52.900 next semester I am having to do that, following model of Shahper Richter, Patrick Dodd and Inna Piven from University of Auckland re oral assessments - look them up 00:31:55.000 --> 00:31:55.900 in my experience group discussions can lead to a power imbalance among some students...etc 00:32:02.000 --> 00:32:02.900 Yes, and you may be reading autocue... 00:32:03.000 --> 00:32:03.900 one person dominates the conversation 00:32:08.000 --> 00:32:08.900 @jasper - we have been conducting oral viva's with 1000s of students - just need marking team - no more expensive than marking an essay 00:32:10.000 --> 00:32:10.900 having a rubric to mark each as you go or just following . similar to online interactive orals (re griffth uni). 00:32:21.000 --> 00:32:21.900 👍 00:32:27.000 --> 00:32:27.900 that's interesting Jacqui - is that a response to GenAI? 00:32:28.000 --> 00:32:28.900 I agree that reconsidering how we assure learning, how we assess learning, is imperative. 00:32:29.000 --> 00:32:29.900 @Laurine Hurley - do you have a link to that - if it was at the start of the session, I was a little late. 00:32:42.000 --> 00:32:42.900 yes @Sel Martinez Pacheco (Torrens University Australia) it's only temporary 00:32:50.000 --> 00:32:50.900 Mathew will have it (A previous TA session) 00:33:04.000 --> 00:33:04.900 Ah - I'll look through the archives, thanks. 00:33:11.000 --> 00:33:11.900 @jasper no but very authentic assessment - and no issues with AI 00:33:24.000 --> 00:33:24.900 i'd love to hear more about it actually. It's quite a bold approach! 00:34:05.000 --> 00:34:05.900 One area that I have concerns about is English placement testing (standardised entry tests) - these ones are probably highly vulnerable to deepfakes 00:34:10.000 --> 00:34:10.900 remote testing centres again? 00:34:18.000 --> 00:34:18.900 Would students prefer a more easy to understand/follow deep fake of a lecturer than a lecturer who is difficult to follow understand? 00:34:26.000 --> 00:34:26.900 @Jacqui Young - yes I worked with a team grading a year 1 project - via Viva (and a comprehensive form on the screen). It was, in many ways better, than essays; as limited to working day & we built in gaps to ensure graders had comfort breaks. 00:35:12.000 --> 00:35:12.900 IMHO, Orals can adapted given the general approach has been shown to be scalable at reasonable cost compared to written exams. Authentic online oral assessment - an examination replacement https://transformingassessment.com/events_30_april_2020.php [link fixed] 00:35:19.000 --> 00:35:19.900 @jasper - video recording of interaction with another student utilising skills taught & then reflection on video & additional questions 00:35:32.000 --> 00:35:32.900 Jim Cook interviews William Charles Wentworth - University of Sydney https://youtu.be/p0ptVToHhes?si=4s0HNd07O_8eviSc 00:35:43.000 --> 00:35:43.900 I agree, there's inclusion and accessibility to consider with every option so as to not disadvantage students 00:35:45.000 --> 00:35:45.900 @Emma Duke-Williams (Dundee) - yes very time effective... 00:35:45.000 --> 00:35:45.900 @Emma Duke-Williams (Dundee) . mathew has put the link in 00:35:48.000 --> 00:35:48.900 Thanks @Jacqui Young 00:36:19.000 --> 00:36:19.900 https://arxiv.org/abs/2404.15601 00:36:35.000 --> 00:36:35.900 https://teachwell.auckland.ac.nz/resources/assessment/interactive-oral-assessments 00:36:48.000 --> 00:36:48.900 https://app.heygen.com/share/61c5963017cc4744843229d866279653 00:36:51.000 --> 00:36:51.900 👍 00:37:02.000 --> 00:37:02.900 sorry broken link http://transformingassessment.com/events_30_april_2020.php 00:37:31.000 --> 00:37:31.900 watch Mike's link first. 00:37:42.000 --> 00:37:42.900 You won't have heard me enough to know this - but the deepfake of my voice made me sound a lot more fancy than I actually am! 00:38:39.000 --> 00:38:39.900 I showed the video of Jasper to people who know him well and they were convinced that it was him 00:38:59.000 --> 00:38:59.900 😕 00:39:01.000 --> 00:39:01.900 the accent seemed to move a bit towards the end... 00:39:10.000 --> 00:39:10.900 kind of scary that Mike now has the power to clone me 00:39:11.000 --> 00:39:11.900 Very convincing 00:39:15.000 --> 00:39:15.900 It is in the eyes that the fake is revealed...(window to the soul!) 00:39:19.000 --> 00:39:19.900 If you hadn't given the disclaimer, I wouldn't have known for the first part! Even after thinking about deepfakes all this time 00:39:20.000 --> 00:39:20.900 The eyes blink but don't move 00:39:22.000 --> 00:39:22.900 How good are deep fakes for those who prefer lip reading to captions (and can they ever do sign languages (and know Auslan/BSL/ASL/etc. 00:40:06.000 --> 00:40:06.900 The tech will keep improving. It'll be an arms race like with GenAI of the creators and the detectors. 00:40:07.000 --> 00:40:07.900 @Emma Duke-Williams (Dundee) They're largely useless, since they don't use proper lip placement, just enough of an approximation to pass muster at a glance 00:40:10.000 --> 00:40:10.900 universities can tell students that if at any time in the future, even after graduating' they discover the student 'cheated' /lied then their degree will be invalidated 00:40:21.000 --> 00:40:21.900 @Brett Cullen - that's what I thought. 00:40:34.000 --> 00:40:34.900 so if we consider GPT3.5 compared to 4o... 00:40:43.000 --> 00:40:43.900 then it would be very convincing 00:41:05.000 --> 00:41:05.900 Chegg a shareholder?😞 00:41:10.000 --> 00:41:10.900 haha 00:41:24.000 --> 00:41:24.900 so even if deep fakes can't be detected now they may be detected in future 00:41:41.000 --> 00:41:41.900 👏 👏 👏 00:42:21.000 --> 00:42:21.900 @jasper I would like to clarify for the video that I have removed the video and audio deepfake from our platform so I can now no longer clone him! 00:42:31.000 --> 00:42:31.900 @Mike Perkins imagine the legal battles and cost, not to mention the detective work to hunt back... 00:42:47.000 --> 00:42:47.900 Yes, I think that the Pandemic was a bit of glitch for education, AI is a total revolution 00:42:47.000 --> 00:42:47.900 🥳 00:43:06.000 --> 00:43:06.900 @Emma Duke-Williams (Dundee) Lip reading is a great one-I don't know! Also something that could be helpful if it works for supporting inclusive access to materials... 00:43:07.000 --> 00:43:07.900 I've seen a lot of disgruntled students who've been accused of plagiarism or cheating thanks to Turnitin AI's 'detector' 00:43:17.000 --> 00:43:17.900 Which is mostly garbage 00:43:35.000 --> 00:43:35.900 Brett you are totally correct. We are just wrapping up a project which investigates some of these issues...some real injustices have taken place 00:43:51.000 --> 00:43:51.900 will the recording be available? 00:43:52.000 --> 00:43:52.900 often teachers going rogue 00:44:04.000 --> 00:44:04.900 I really hope the various student guilds are across the advocacy for students, it's horrible how easy it is to trigger the AI detection software 00:44:05.000 --> 00:44:05.900 and jumping to conclusions based on the output of zerogpt etc. 00:44:06.000 --> 00:44:06.900 So glad we opted out of TII's AI guesstimations. 00:44:10.000 --> 00:44:10.900 I think you are on the money Matt. The current model of education will be upended and we will be transitioning and redefining what we do for many years to come. 00:44:14.000 --> 00:44:14.900 Not to mention ESL students using Grammarly etc. for polish 00:44:27.000 --> 00:44:27.900 Potential for identity theft at a higher level 00:44:52.000 --> 00:44:52.900 Similar witch hunts as ever from the zealous among us who are determined to catch out their 'cheating students'... sad 00:44:55.000 --> 00:44:55.900 would be happy to connect with anyone on LinkedIn who wants to chat further or follow our posts https://www.linkedin.com/in/dr-jasper-roe-sfhea-894571102/ 00:44:57.000 --> 00:44:57.900 At our institution we don't allow detectors, or at least, they couldn't be used in a disciplinary process. 00:44:59.000 --> 00:44:59.900 https://arxiv.org/abs/2403.19148 00:45:21.000 --> 00:45:21.900 AI detection is an arms race that nobody can win 00:45:31.000 --> 00:45:31.900 Agreed 00:45:57.000 --> 00:45:57.900 There's some interesting techniques using frequent words like 'delve' and 'demystify' that were noticed 00:46:00.000 --> 00:46:00.900 For every detection method there is a new obfuscation method available 00:46:10.000 --> 00:46:10.900 Andrew - absolutely. 00:46:39.000 --> 00:46:39.900 and in the meantime, how much effort goes to waste in this punitive, catch&punish paradigm 00:46:39.000 --> 00:46:39.900 Authorship tools are more effective - tracking the writing style over time and looking for big deviations as opposed to steady improvement 00:46:54.000 --> 00:46:54.900 https://link.springer.com/article/10.1007/s10462-024-10810-6 00:47:51.000 --> 00:47:51.900 @Andrew agree. 00:48:00.000 --> 00:48:00.900 Great point. Any assistive program is being penalised at the moment 00:48:21.000 --> 00:48:21.900 Yes: Authorship tools are more effective - tracking the writing style over time and looking for big deviations as opposed to steady improvement. 00:48:25.000 --> 00:48:25.900 Requires a whole rethink of assessments- I like the phrase- 'assesses the process not the product'..... so maybe more authentic /in person assessment types/ and more programmatic type assessments are areas for educators to consider. 00:48:52.000 --> 00:48:52.900 Project based assessment that tracks performance over the semester, with iterative, repeating assessible 'check-in's are the way to go 00:48:56.000 --> 00:48:56.900 @Jo #2 Our paper discusses those issues in terms of inclusive education 00:48:59.000 --> 00:48:59.900 Check this alternative: https://nor.education/ 00:49:02.000 --> 00:49:02.900 But they're incredibly time intensive and don't scale well 00:49:13.000 --> 00:49:13.900 Assessment as learning - more formative - this is something we've been calling for for years - now another reason to do it 🙂 00:49:30.000 --> 00:49:30.900 @Jodi Sita #2 I agree - I suspect you, like us, have been talking about process being far more important than product for many years - finally many others are realising that it's key. 00:49:46.000 --> 00:49:46.900 Thanks everyone. 00:50:19.000 --> 00:50:19.900 @Mark Bassett Some kind of ... studio based multi-credit point unit perhaps ... 😉 00:50:20.000 --> 00:50:20.900 agree re detectors.... but on the other hand, when speaking (empathetically) with students,. so many own up to using it and tell us why.. it is helping us target better help and resources and trying (the hardest bit) to get teaching staff to scaffold, be clear and be reasonable in what and how they assess! 00:50:21.000 --> 00:50:21.900 A poor tool for definitive detection but one way of starting a conversation with a student. 00:50:53.000 --> 00:50:53.900 @Paul Oliveri agree and @Laurine Hurley yes that's the other side of the coin honestly - I have had many cases of that 00:50:54.000 --> 00:50:54.900 Our new mantra: assess the process not the product. 00:51:07.000 --> 00:51:07.900 IMHO, no Ai detector result should be used as the sole basis to accuse a student. it should always be moderated by a fair due process. (which is how TI also says it should happen). I like Kane Murdoch's 'courageous conversations' approach (also covered in a previous TA webinar). 00:51:35.000 --> 00:51:35.900 that should be BAU, @Mathew Hillier 00:51:45.000 --> 00:51:45.900 @Brett Cullen - what we should always be spending most of our time on - teaching and developing students' evaluative judgement. The facts have been Googleable for decades 🙂 00:51:46.000 --> 00:51:46.900 Yes Matt 00:52:23.000 --> 00:52:23.900 CRADLE recently had a seminar on evaluative judgment.. most illuminating 00:52:29.000 --> 00:52:29.900 @Matt Townsend - Cardiff University - yes it would be a good thing. it will (and should) require rethinking a lot of learning outcomes, curriculum and assessments. 00:53:06.000 --> 00:53:06.900 There needs to be seismic shifts in assessment modes at Universities, but that effort is costly and risk inducing - both of which Uni management aren't super enthusiastic about 00:53:10.000 --> 00:53:10.900 https://open-publishing.org/journals/index.php/jutlp/article/view/810 00:53:12.000 --> 00:53:12.900 Do universities keep assessments for 10 years+? 00:53:13.000 --> 00:53:13.900 Let's not make academic integrity into WADA 00:53:13.000 --> 00:53:13.900 @Laurine Hurley yes - Margaret Bearman et al recently had a paper in Assess & Eval on this too. 00:53:42.000 --> 00:53:42.900 To what extent is the quality of synthetic media determined by how much one pays for it? In terms of student use, are we looking at economic disparities as well as gaps in technical knowhow? 00:53:54.000 --> 00:53:54.900 Reminds me of this news: https://www.aljazeera.com/news/2024/1/3/why-has-harvard-president-claudine-gay-resigned 00:53:54.000 --> 00:53:54.900 yes, that's the paper they used 00:54:04.000 --> 00:54:04.900 yes - rethinking the learning outcomes1 00:54:17.000 --> 00:54:17.900 @Carol Bailey absolutely. 00:54:20.000 --> 00:54:20.900 Yes- it will be very interesting in the future looking back at lots of things (not just students cheating- but lots of things in society at the moment)- and having a future lens that detects AI use that was not acknowledged - that was used in this period. 00:54:31.000 --> 00:54:31.900 Thanks for a great presentation and discussion, have to leave a few minutes early. Looking forward to connecting in the next one 🙂 00:54:34.000 --> 00:54:34.900 Bearman et al.. https://www.tandfonline.com/doi/full/10.1080/02602938.2024.2335321 00:55:03.000 --> 00:55:03.900 Technical knowhow differences is something we can support students to enhance. £$€ differences are harder. 00:55:10.000 --> 00:55:10.900 Thank you for a great session. A lot to process and reflect on! 00:55:11.000 --> 00:55:11.900 Thanks Mike and Jasper - a challenging thoughtful session. 00:55:17.000 --> 00:55:17.900 Thanks for a great discussion 00:55:22.000 --> 00:55:22.900 Thanks for sharing! 00:55:22.000 --> 00:55:22.900 I missed a chunk but love these conversations 00:55:25.000 --> 00:55:25.900 My LinkedIn https://www.linkedin.com/in/mgperkins/ 00:55:26.000 --> 00:55:26.900 is this the 'real' Mathew? 00:55:33.000 --> 00:55:33.900 Great session. If it WAS actually you! 00:55:34.000 --> 00:55:34.900 I have found this resource useful- from Flinders Uni: https://staff.flinders.edu.au/learning-teaching/good-practice-guides/good-practice-guide---designing-assessment-for-artificial-intell 00:55:34.000 --> 00:55:34.900 😃 00:55:53.000 --> 00:55:53.900 Great session Jasper and Mike, much to think about and follow up on. Thanks for all the links that have been shared by everyone too. 00:55:59.000 --> 00:55:59.900 @Jodi Sita #2 Flinders used our AI Assessment scale for this 🙂 00:56:03.000 --> 00:56:03.900 Thanks- great session. 00:56:29.000 --> 00:56:29.900 👏 00:56:34.000 --> 00:56:34.900 its great @Mike Perkins - I have been showing lots of people that scale 00:56:34.000 --> 00:56:34.900 👍 00:56:37.000 --> 00:56:37.900 Thanks very much 00:56:41.000 --> 00:56:41.900 thank you 👏 00:56:41.000 --> 00:56:41.900 thanks 00:56:43.000 --> 00:56:43.900 Thank you, very informative 00:56:43.000 --> 00:56:43.900 Thank you all