WEBVTT

00:00:18.000 --> 00:00:18.900
<v Eleonora Moura>
Good morning from St Albans, UK

00:00:30.000 --> 00:00:30.900
<v Paul Whitelaw>
acknowledging the woi wurrung and bun wurrung people of the east KULIN nation here in Naarm (Melbourne)

00:02:23.000 --> 00:02:23.900
<v Mathew Hillier>
Hi from eastern Sydney, Australia, on the lands of the Bidjigal and Gadigal peoples of the Eora nation.

00:03:22.000 --> 00:03:22.900
<v Andreas Kopf>
Hi I'm joining in from Fiji

00:03:27.000 --> 00:03:27.900
<v jane Pittaway>
Hello all, meeting from Launceston (Kanamaluka) Tasmania (Lutruwita) on the lands of the Palawa people.

00:04:42.000 --> 00:04:42.900
<v jane Pittaway>
yes

00:05:07.000 --> 00:05:07.900
<v Mathew Hillier>
https://www.menti.com/ali24mximed5

00:05:08.000 --> 00:05:08.900
<v Fabio Aricò>
http://www.menti.com/ali24mximed5

00:05:15.000 --> 00:05:15.900
<v Lee (Leanne) Shibata [OVC Dept – TGE]>
Hello, I'm joining from the land of the Yugambeh/Kombumerri people.

00:09:44.000 --> 00:09:44.900
<v Mathew Hillier>
If your university has published information for staff/students related to use of Generative AI online, please feel free to share the links here.

00:10:02.000 --> 00:10:02.900
<v Fabio Aricò>
Yes, interesting to hear as different institutions have responded to AI very differently, from ignoring to regulating very tightly.

00:11:15.000 --> 00:11:15.900
<v Eleonora Moura>
https://dx.doi.org/10.2139/ssrn.4544010

00:11:19.000 --> 00:11:19.900
<v Mathew Hillier>
Macquarie Uni where i am located, we published as series of blog posts for staff. https://teche.mq.edu.au/tag/generative-artificial-intelligence/

00:12:48.000 --> 00:12:48.900
<v Mathew Hillier>
Initial suggested guide for use we proposed https://teche.mq.edu.au/2023/03/advising-students-about-using-and-citing-generative-artificial-intelligence-for-assessment/

00:13:05.000 --> 00:13:05.900
<v Fabio Aricò>
The University of East Anglia has a full AI Policy. I cannot share yet as it is currently in the process of approval. But it covers everything, also research and property rights.

00:13:17.000 --> 00:13:17.900
<v Mathew Hillier>
and an AI literacy framework suggestion: https://teche.mq.edu.au/2023/03/a-proposed-ai-literacy-framework/

00:14:17.000 --> 00:14:17.900
<v Paul Hunter>
University of Sydney (not my institute): https://canvas.sydney.edu.au/courses/51655/pages/general-principles-for-use-of-generative-ai

00:14:46.000 --> 00:14:46.900
<v Mathew Hillier>
Thanks Paul, the USyd stuff is good 🙂

00:15:05.000 --> 00:15:05.900
<v jane Pittaway>
Yes, treat AI output as any other reference source.

00:15:53.000 --> 00:15:53.900
<v Mike>
if edit /modify AI output how is this reported in quotes?

00:16:46.000 --> 00:16:46.900
<v jane Pittaway>
What do you mean by students remaining autonomous using AI tools?

00:16:58.000 --> 00:16:58.900
<v Tom Gislev>
Exactly, I got the same question as Mike. A lot of AI output is edited before used/published in essays

00:17:15.000 --> 00:17:15.900
<v Fabio Aricò>
I am taking notes, please keep questions coming 🙂

00:17:39.000 --> 00:17:39.900
<v Tim Hunt>
Initial Open University guidance is https://about.open.ac.uk/policies-and-reports/policies-and-statements/gen-ai/generative-ai-students?nocache=64f82801052d9


00:19:00.000 --> 00:19:00.900
<v Mathew Hillier>
We explored why ChatGPT makes fake reference here: https://teche.mq.edu.au/2023/02/why-does-chatgpt-generate-fake-references/

00:20:41.000 --> 00:20:41.900
<v Mathew Hillier>
advice produced for students https://students.mq.edu.au/study/assessment-exams/academic-integrity/ai-tools also the library developed citation guidelines

00:21:38.000 --> 00:21:38.900
<v jane Pittaway>
Surely what we want student to be able to do is generate ideas and fashion arguments?

00:22:13.000 --> 00:22:13.900
<v Mathew Hillier>
MQ library guides on gen Ai https://libguides.mq.edu.au/referencing/generativeAI

00:22:21.000 --> 00:22:21.900
<v Lee (Leanne) Shibata [OVC Dept – TGE]>
Supporting both staff and students is absolutely to be applauded Iliada! 

00:22:45.000 --> 00:22:45.900
<v Elin Weston>
Were the student perceptions of AI use reflected in their work - in other words, did it help improve their work (as it were)?

00:22:52.000 --> 00:22:52.900
<v Laurine Hurley>
 @Lee, 👍 

00:23:00.000 --> 00:23:00.900
<v Fabio Aricò>
I think that, aside from generating, students need to develop the critical ability to appraise the information they are eliciting. (It was the same with Google, Wikipedia... no news under the sun but the complexity is increasing).

00:24:11.000 --> 00:24:11.900
<v jane Pittaway>
Thank you Iliada

00:24:11.000 --> 00:24:11.900
<v Mathew Hillier>
Fabio- i think the 'plausibility' of the wrong info you get out of LLMs is a challenge for users. 

00:24:19.000 --> 00:24:19.900
<v Marion Palmer>
Thanks Iliada - great food for thought

00:24:31.000 --> 00:24:31.900
<v Iliada Eleftheriou>
AI Code of Conduct: https://www.iliada-eleftheriou.com/AICodeOfConduct/


00:24:36.000 --> 00:24:36.900
<v STUART HARGREAVES>
https://ler.scholasticahq.com/article/83297-words-are-flowing-out-like-endless-rain-into-a-paper-cup-chatgpt-law-school-assessments

00:24:45.000 --> 00:24:45.900
<v Nikki Donald, Product Innovation ~ On Gadigal Land ~>
THank you Iliada!

00:25:14.000 --> 00:25:14.900
<v Laurine Hurley>
☺ 

00:26:59.000 --> 00:26:59.900
<v jane Pittaway>
100% OMG!

00:27:14.000 --> 00:27:14.900
<v Laurine Hurley>
😩 

00:27:53.000 --> 00:27:53.900
<v Fabio Aricò>
ChatGPT performs differently across disciplines. Surely, colleagues teaching Law reported quite a lot of concerns since early days back in Winter 2023 here at UEA. 

00:29:54.000 --> 00:29:54.900
<v Fabio Aricò>
So....we can cite the source, but we cannot know about the "source of the source", very far away from academic publications.

00:30:20.000 --> 00:30:20.900
<v jane Pittaway>
very true Fabio

00:30:43.000 --> 00:30:43.900
<v Mike>
since a lot of the web after sept 2021 will have been created by generative AI and will be used to train new generative AI models the 'garbage out' will increase 

00:32:31.000 --> 00:32:31.900
<v Ashley Lawson>
I've had the same experience with Chat GPT apparently constructing law to fill gaps

00:32:57.000 --> 00:32:57.900
<v Mathew Hillier>
A lawyer in the US did this in front of a judge and got pinged for it.

00:33:06.000 --> 00:33:06.900
<v Mathew Hillier>
it faked up the cited cases.

00:33:33.000 --> 00:33:33.900
<v Lee (Leanne) Shibata [OVC Dept – TGE]>
Gosh Mathew.

00:33:34.000 --> 00:33:34.900
<v Mike>
how many human assessors will check validity?

00:34:29.000 --> 00:34:29.900
<v Carmela De Maio>
https://www.abc.net.au/news/2023-06-24/us-lawyer-uses-chatgpt-to-research-case-with-embarrassing-result/102490068

00:34:41.000 --> 00:34:41.900
<v Fabio Aricò>
A lot to be said about the human interaction taking place over training of AI. Who are the trainers, what do they know, how much of their bias transfers into "the knowledge"...

00:34:57.000 --> 00:34:57.900
<v Mathew Hillier>
Thanks Carmela 🙂

00:34:58.000 --> 00:34:58.900
<v Stephan>
It would be really interesting to repeat this exercise with GPT4.

00:35:06.000 --> 00:35:06.900
<v Mano>
In Computer Science, the rate of correct answers generated by ChatGPT is a lot higher.

00:35:18.000 --> 00:35:18.900
<v Laurine Hurley>
@stephan, yes it would

00:35:44.000 --> 00:35:44.900
<v Mathew Hillier>
the paid version of ChatGPT can search the web with a plugin. and the microsoft bot in creative mode is gpt4 and can search the web.

00:36:10.000 --> 00:36:10.900
<v Mike>
step by step / chain of thought prompting improves answers

00:36:14.000 --> 00:36:14.900
<v Stephan>
Not revealing the data on which ChapGPT is trained is a massive retrograde step - almost as though they want it to be used for cheating!

00:36:56.000 --> 00:36:56.900
<v Fabio Aricò>
Agreed Stephan, but of course they are a business, they want to preserve their margin. Too often we forget Google also is!

00:37:34.000 --> 00:37:34.900
<v jane Pittaway>
Yes as an Academic Integrity Adviser, it is very difficult to detect AI use

00:37:52.000 --> 00:37:52.900
<v Mathew Hillier>
lots of good advice on using chatgpt here (Ethan Mollick from Wharton US) https://www.oneusefulthing.org/p/how-to-use-ai-to-do-practical-stuff

00:38:05.000 --> 00:38:05.900
<v Stephan>
Another aspect in the legal sector is that students will be using these tools in practice, so we need to help them become familiar with them at the HE stage.

00:38:52.000 --> 00:38:52.900
<v Mathew Hillier>
Ethan Mollick's twitter steam has lots of interesting examples. https://twitter.com/emollick

00:38:59.000 --> 00:38:59.900
<v Mike>
can use generative AI as a personal tutor on a topic rather than to create assignment writing answer

00:39:08.000 --> 00:39:08.900
<v Lee (Leanne) Shibata [OVC Dept – TGE]>
"AI-proof" is mentioned across the sector - can we agree on a definition and can anything really be "AI-proof"?

00:39:13.000 --> 00:39:13.900
<v jane Pittaway>
I can understand how AI would assist in law, as my perception there is so much research involved in creating arguments.

00:39:56.000 --> 00:39:56.900
<v Mathew Hillier>
@Lee (Leanne) Shibata [OVC Dept – TGE] - i would say 'no' in both cases, but it would be nice dream 😉 

00:40:50.000 --> 00:40:50.900
<v Mike>
if paying for better AI gives better results how can we ensure equity?

00:41:06.000 --> 00:41:06.900
<v Stephan>
Great question, @Mike 

00:41:16.000 --> 00:41:16.900
<v Laurine Hurley>
key question, @Mike 

00:41:18.000 --> 00:41:18.900
<v Lotte Dyhrberg O'Neill>
please, define in-class assessment? Our students bring their PC's to class and use our wifi for internet access. Do you mean pen and invigilated written exams?

00:41:31.000 --> 00:41:31.900
<v Mano>
(2) is quite difficult in foundational questions (adding to numbers, for example)

00:41:38.000 --> 00:41:38.900
<v Ashley Lawson>
to an extent, students have always been able to improve their prospects in assessment by paying for resources / help (within the rules of academic misconduct)

00:41:38.000 --> 00:41:38.900
<v Nikki Donald, Product Innovation ~ On Gadigal Land ~>
@Mike One of the ethical issues presented by Iliada...

00:41:41.000 --> 00:41:41.900
<v Lucy Gabrielsen - BPPUniversity - UK>
@Mike - good question

00:41:53.000 --> 00:41:53.900
<v Matt Townsend (Cardiff University) #2>
Any institutions been looking at purchasing an OpenAI Enterprise licence?

00:42:10.000 --> 00:42:10.900
<v Fabio Aricò>
I'd be curious to hear that too. 

00:42:16.000 --> 00:42:16.900
<v Laurine Hurley>
@Ashley.. and☹   outside AI guidelines as well!

00:42:21.000 --> 00:42:21.900
<v Fabio Aricò>
(we don't have that)

00:42:27.000 --> 00:42:27.900
<v Lee (Leanne) Shibata [OVC Dept – TGE]>
@Mathew Hillier , I'm not in favour of the term "AI proof" - but thought worthy of posing the question here ;). Thanks for your response.

00:43:16.000 --> 00:43:16.900
<v Mathew Hillier>
Yep - horses for courses.  Gen Ai has many uses, we should be teaching the students how to use it, and it needs to be within the assessment frame.

00:43:17.000 --> 00:43:17.900
<v Lucy Gabrielsen - BPPUniversity - UK>
Definitely course specific usage.

00:43:27.000 --> 00:43:27.900
<v Paul Sijpkes>
Using ChatGPT with a dialogic could be an interesting use for assessment, it could make student thinking more explicit.  They could submit the full chain of thought prompting on a question.

00:44:38.000 --> 00:44:38.900
<v Tom Gislev>
In EdTech we use the term "Cognitive Partner" I think that fits proper to pedagogical use of AI

00:44:42.000 --> 00:44:42.900
<v Mike>
will AI do a better job in law courts than an 'ancient' judge?

00:44:46.000 --> 00:44:46.900
<v Laurine Hurley>
"acceptance of failed efforts" is often the stumbling block, sadly

00:44:59.000 --> 00:44:59.900
<v Paul Sijpkes>
That made no sense sorry, I meant to say, submitting a chain of though prompting session with an LLM could make a great assessment. Sorry long day!

00:46:06.000 --> 00:46:06.900
<v jane Pittaway>
Thanks Stuart

00:46:08.000 --> 00:46:08.900
<v Laurine Hurley>
thanks, Stuart!

00:46:12.000 --> 00:46:12.900
<v Dan>
Thanks Stuart 🙂

00:46:16.000 --> 00:46:16.900
<v Paul Sijpkes>
Thanks stuart

00:46:17.000 --> 00:46:17.900
<v Marion Palmer>
Thanks Stuart

00:46:17.000 --> 00:46:17.900
<v Eleonora Moura>
Thank you very much! 

00:46:19.000 --> 00:46:19.900
<v Carmela De Maio>
Thanks Stuart. 

00:46:22.000 --> 00:46:22.900
<v Ashley Lawson>
thanks both - really very useful!

00:46:24.000 --> 00:46:24.900
<v Tim Hunt>
Paywalls are nothing new in academia, sadly. I guess we know how to deal with them.

00:46:27.000 --> 00:46:27.900
<v Bronwyn Tarrant>
That was awesome - thank you.

00:46:28.000 --> 00:46:28.900
<v Nikki Donald, Product Innovation ~ On Gadigal Land ~>
Thank you Stuart!!

00:46:40.000 --> 00:46:40.900
<v Mano>
Thanks Stuart!

00:46:42.000 --> 00:46:42.900
<v Sue Muloin>
Thanks to both of you. Need to run now to teach!

00:46:44.000 --> 00:46:44.900
<v Tom Gislev>
Thanks for two solid presentations, really useful

00:47:00.000 --> 00:47:00.900
<v Lucy Gabrielsen - BPPUniversity - UK>
@STUART HARGREAVES very informative with useful reflective points. Thank you.

00:47:03.000 --> 00:47:03.900
<v Carmela De Maio>
And thanks Iliada. Both interesting presentations.

00:47:06.000 --> 00:47:06.900
<v Claudia>
Many thanks Stuart and Iliada.

00:47:20.000 --> 00:47:20.900
<v Lee (Leanne) Shibata [OVC Dept – TGE]>
Applaud your suggestion @Paul Sijpkes  (it made sense) - dialogic evaluative and critical thinking at play is a worthy idea.  

00:47:27.000 --> 00:47:27.900
<v Lotte Dyhrberg O'Neill>
Thanks, Stuart!

00:47:40.000 --> 00:47:40.900
<v Harvey Woolf>
Thanks got both of you.

00:48:01.000 --> 00:48:01.900
<v Rupal Lovell-Patel>
Thank you for the presentations.

00:48:01.000 --> 00:48:01.900
<v Sarah Goodjohn>
Thank you both

00:48:28.000 --> 00:48:28.900
<v Lucy Gabrielsen - BPPUniversity - UK>
Can we have the stats here in the Chat please and sent with the recording please.

00:49:00.000 --> 00:49:00.900
<v Nikki Donald, Product Innovation ~ On Gadigal Land ~>
Does 'efficient' equate to 'effective' learning tho?

00:49:15.000 --> 00:49:15.900
<v Lucy Gabrielsen - BPPUniversity - UK>
Much appreciated 👍

00:49:16.000 --> 00:49:16.900
<v Stephan>
Absolutely agree, @Nikki!

00:49:25.000 --> 00:49:25.900
<v Mathew Hillier>
we can add the survey results onto the recording page 🙂

00:49:43.000 --> 00:49:43.900
<v Lucy Gabrielsen - BPPUniversity - UK>
@Mathew Hillier Brilliant. Thanks

00:50:04.000 --> 00:50:04.900
<v Stephan>
When students say that it has made learning better - I suspect that many of them mean that they achieved the output they wanted more quickly. That does not necessarily involve learning!

00:50:13.000 --> 00:50:13.900
<v Mike>
AI tools will develop/improve faster than HE policies

00:50:39.000 --> 00:50:39.900
<v Paul Sijpkes>
Agreed @Nikki Donald, Product Innovation ~ On Gadigal Land ~ I'm not sure on whether 'efficiency' is something that is conducive to learning or critical thinking or deep analysis.

00:50:40.000 --> 00:50:40.900
<v Laurel Raethel>
Agree Stephan!

00:51:01.000 --> 00:51:01.900
<v Laurine Hurley>
@Mike, that, like death and taxes, is certain!

00:51:24.000 --> 00:51:24.900
<v Mike>
AI doesn't die or pay taxes !

00:51:25.000 --> 00:51:25.900
<v Lucy Gabrielsen - BPPUniversity - UK>
@Stephan good point regarding results rather than actual learning!

00:51:41.000 --> 00:51:41.900
<v Lee (Leanne) Shibata [OVC Dept – TGE]>
It is such an interesting concept "AI proofing", agreed Iliada and it is so complex and "contextualised" agreed Stuart. Thank you for sharing your thoughts. 

00:51:43.000 --> 00:51:43.900
<v Nikki Donald, Product Innovation ~ On Gadigal Land ~>
@Stephan and where has the critical thinking/reflection come in? Are they just generating answers without thinking about them??

00:51:52.000 --> 00:51:52.900
<v Tim Hunt>
The questions AI raises for the world is very much around "what does it mean to know stuff, to create, to be human".

One thing that is clear that in future (actually already true for decades) is that just remembering stuff is somewhere that computers are better. Therefore, going back to closed book exams in response to this would be very paradoxical!

00:52:05.000 --> 00:52:05.900
<v Ashley Lawson>
I think when we have an eye on preparing students for the workplace, efficiency is a value we should engender, but of course only one of many!

00:52:07.000 --> 00:52:07.900
<v Laurine Hurley>
@Tim, agree 100%

00:52:11.000 --> 00:52:11.900
<v Nikki Donald, Product Innovation ~ On Gadigal Land ~>
Yes totally @Lee (Leanne) Shibata [OVC Dept – TGE]  - context is critical!

00:52:49.000 --> 00:52:49.900
<v Mike>
it is dangerous to give powerful tools to novices ... would you give a chain saw to a child !

00:52:57.000 --> 00:52:57.900
<v Nikki Donald, Product Innovation ~ On Gadigal Land ~>
Yikes!!!

00:53:15.000 --> 00:53:15.900
<v jane Pittaway>
Thank you -for your answer

00:53:26.000 --> 00:53:26.900
<v Matt Townsend (Cardiff University) #2>
Thanks all. Got to rush off 🙂

00:53:58.000 --> 00:53:58.900
<v Jeanette Smit #2>
Thank you for the informative presentations.

00:54:39.000 --> 00:54:39.900
<v Mike>
USA have used AI to decide which prisoners get released

00:54:56.000 --> 00:54:56.900
<v Michael Kilmister (he/him) #2>
Thanks everyone, the talks and discussion have been thought provoking!

00:54:59.000 --> 00:54:59.900
<v Jenny James>
😳 

00:55:10.000 --> 00:55:10.900
<v Nikki Donald, Product Innovation ~ On Gadigal Land ~>
That's fascinating @Mike 

00:55:28.000 --> 00:55:28.900
<v Lee (Leanne) Shibata [OVC Dept – TGE]>
Thank you kindly to Mathew Hillier, Fabio Aricò, Iliada and Stuart. So much to gain from your shares and thank you for contextualising your experiences with AI.

00:55:38.000 --> 00:55:38.900
<v Andreas Kopf>
Stimulating discussions. Glad I joined

00:55:49.000 --> 00:55:49.900
<v Lucy Gabrielsen - BPPUniversity - UK>
We currently focus very much on students' use of AI in assessment writing.  Has anyone attempted using AI to make the marking giving effective feedback and be more efficient (faster) in marking assessments?

00:55:56.000 --> 00:55:56.900
<v Mathew Hillier>
The 'robodebt' scandal in AU bares out the importance of keeping humans in the loop. (this was an automated debt issuing system from the Au federal govt that issued debts on social security recipients and then asked the accused to prove the debt was not true).

00:56:41.000 --> 00:56:41.900
<v Bronwyn Tarrant>
well said Mathew!

00:56:46.000 --> 00:56:46.900
<v jane Pittaway>
@Lucy Gabrielsen - YES!

00:56:55.000 --> 00:56:55.900
<v Paul Sijpkes>
@Mathew Hillier and the A-Level 'mutant algorithm' lol@Mike

00:57:07.000 --> 00:57:07.900
<v Lee (Leanne) Shibata [OVC Dept – TGE]>
Thank you for also bringing the "human" element to this conversation and holistically exploring and experiencing AI from the learner and teacher's (but not limited to) perspective. 

00:57:10.000 --> 00:57:10.900
<v Laurine Hurley>
@Jane.. how well did it work?

00:57:20.000 --> 00:57:20.900
<v Mathew Hillier>
robodebt is an example of tech-neoliberalism gone mad

00:57:24.000 --> 00:57:24.900
<v Nikki Donald, Product Innovation ~ On Gadigal Land ~>
Thank you this has been fantastic.

00:57:26.000 --> 00:57:26.900
<v Lotte Dyhrberg O'Neill>
Thanks, Iliada, Stuart and Fabio.

00:57:29.000 --> 00:57:29.900
<v jane Pittaway>
Sorry Laurine, I meant yes what a great use for IA!

00:57:32.000 --> 00:57:32.900
<v Mike>
Scandal in UK post office prosecuted staff for fraud but in reality was fault with computer system

00:57:34.000 --> 00:57:34.900
<v jane Pittaway>
AI!

00:57:36.000 --> 00:57:36.900
<v Laurine Hurley>
thank you all!

00:57:37.000 --> 00:57:37.900
<v STUART HARGREAVES>
thanks everyone

00:57:39.000 --> 00:57:39.900
<v Tim Hunt>
I also think that, for anyone involved in maths education, these are not fundamentally new issues. Maths has been grappling with the question 'should students be allowed calculators in exams' for decades.

00:57:40.000 --> 00:57:40.900
<v Lucy Gabrielsen - BPPUniversity - UK>
@jane Pittaway is that a yes to having tried it?

00:57:40.000 --> 00:57:40.900
<v Ashley Lawson>
thank you so much all speakers and contributors in chat! Really useful.

00:57:49.000 --> 00:57:49.900
<v Laurine Hurley>
@Jane.. rats!

00:57:55.000 --> 00:57:55.900
<v Sarah Goodjohn>
Much food for thought, Thank you all.

00:57:55.000 --> 00:57:55.900
<v Annika Lindberg, Aarhus University, Denmark #2>
Thank you so much

00:57:56.000 --> 00:57:56.900
<v Lindesay>
Thanks everyone. Great discussion

00:58:00.000 --> 00:58:00.900
<v Iliada Eleftheriou>
Thanks everyone!! 

00:58:04.000 --> 00:58:04.900
<v Kelly Galvin (CLTS)>
Great session. Thank you! Very practical

00:58:06.000 --> 00:58:06.900
<v Tim Linsey>
Thank you to both presenters. Great session

00:58:06.000 --> 00:58:06.900
<v Fabio Aricò>
Thank you all for your energy this morning. It was great (as usual) 🙂

00:58:08.000 --> 00:58:08.900
<v Elin Weston>
Thanks everyone - this has been really interesting

00:58:10.000 --> 00:58:10.900
<v Dan>
Thanks all

00:58:10.000 --> 00:58:10.900
<v jane Pittaway>
THank you, great presentations & chat

00:58:11.000 --> 00:58:11.900
<v Bronwyn Tarrant>
thanks