Dr Leon Furze

Thanks to everyone who could attend the online launch for Practical AI Strategies – it was great to see so many familiar and new faces!

We covered a lot of ground, including a reading from the first couple of chapters, and discussions of assessment, academic integrity, privacy, data, and the use of GenAI in K-12 education.

Check out the full recording of the session below:

Transcript via Otter.ai [uncorrected]

Leon Furze 0:03
Okay, morning, everybody. And thanks for coming to the online version of the book launch, we had a face to face book launch last week in Melbourne, and at least is there in the crowd today. And thank you for organizing that. And it’s great that we could do an online version as well, for people who were unable to attend. So if you’re here this morning, or if you’re watching this on the recording, thank you very much. And I hope you enjoyed the session, I’m going to do a little bit of reading from the book and just talking about how the book was made and why it was made, and how it links to some of my PhD studies and the work that I do with schools. And then save some time at the end for q&a discussion. Any questions that you have around generative AI in education really.

And when we did the launch in Melbourne last week, we had some fantastic discussions around a really broad range of topics from things like copyrights, and how teachers can use these technologies without infringing copyright or breaching people’s intellectual property, right, the way through to how this is going to change assessment tasks, the impact it’s going to have on K to 12, and tertiary education and a whole range of different subjects. So if you have any questions do yell out. And if I don’t know the answer to the questions, I’ve just load up chat GPT on the side and get it to generate a totally made up response, which is pretty much what everybody does in any kind of webinar these days. Before I kick off, I’ll just introduce myself. And for the benefit of people I haven’t worked closely with, although there’s some familiar faces in the audience here, which is very nice. And I’ll begin by acknowledging the Gunditjmara people as the traditional custodians of the land where I’m presenting from today and extend that to any Aboriginal and Torres Strait Islanders who might be joining us this morning. So my name is Leon furze. I am the author of practical AI strategies, as well as practical reading strategies and practical writing strategies with Amber press. And I worked in secondary education for 15 years, predominantly as an English and literature teacher. I’ve also taught media, drama, stem, pretty much everything. And over the course of those years, I had a few different roles, predominantly working out here in southwest Victoria at the small Catholic school. And I’ve been a director of learning and teaching, I’ve been a head of English, my focus for most of the time has been on curriculum development. And that’s the background that I bring really to working with teachers and schools now when I do my professional development and my consulting work.

So most of the time when I’m working with schools, nowadays, it’s around generative artificial intelligence, obviously. But I’ve also got a bit of an English teacher hat still, I’m on the Council for the Victorian Association teachers for teaching of English, Vate. And the work that we do for those is obviously, really closely related to the reading and writing from my previous books. So you might see me around the place, you’ll see me with a few different hats on. But at the moment, most of what I talk about is generative AI, I talk endlessly on LinkedIn about generative AI. So if you did get here via LinkedIn, you’ve probably seen some of that already. Luckily, I don’t use any other social media. So I get to throw 100% of my time at that one platform. But there’s some great educators in K 12, in particular, who’ve moved across to LinkedIn recently. So if you don’t use that platform very much, I would really suggest that you get on there. And check out some of the teachers who are working with generative artificial intelligence, because there’s some fantastic resources being shared. I started my PhD at the end of 2022. So just a couple of weeks before chat GPT was launched. My PhD officially kicked off and it was really just good timing.

So back in May 2022, I had a conversation at my school I handed in my notice, effective the end of the year, nice long runway to find somebody to do Director of Teaching and Learning roll out here in the regional area. And over that period of time, I was planning on setting up some of my literacy consulting work and going sort of down that route. And then a couple of weeks after I handed in my notice and opportunity came up for a scholarship with Dr. Lucinda McKnight at Deakin University, who I know for the English teaching Association, and she was working on digital texts and a DECRA funded program exploring digital texts in the classroom. We had a conversation around generative artificial intelligence and she and I had been sort of playing around with these technologies for a little while they send us first article and the conversation about Gen AI was I think, in 2021. So she was a long time before chat GPT was launched. And we said look, yeah, the These technologies, they’re probably three to five years away from coming into the classroom. There’s, you know, there’s a lot of scope for us to play around, experiment, maybe get some teachers using early versions of the technology and find out what their opinions are. And then in a few years time, everybody will probably be using it. And then two weeks after the PhD started, ChatGPT launched and within about three weeks, everybody was using it.

So 100 million users in the space of just a couple of months, outstripping Instagram outstripping Snapchat, Tik Tok, and all of those social media platforms in terms of adoption worldwide. And it still continues to rock it on. So I finished up my teaching role in December, came out into the full time PhD and consulting work, and started to write on my blog, a lot around generative artificial intelligence, normally pushing out two or three posts a week, which was actually then really helpful for reasons you’ll see in a moment when it came to pulling this book together. So I’m going to read from a little bit of the book at the beginning, I’ve got a copy here. But Zoom’s blow will probably cut it out most of the time. And it’s about how the book was constructed. Some of the thinking that went into it, and then a little bit from the introduction around my PhD focus, and the ideas that kind of guide the structure of this book. I’m going to read the about the book part, but first.

This is Part disclaimer. Part explainer and part guilty confession. ChatGPT was used to create some of this book, shock, horror disgust, but not in the way that you might think. Since open AI kindly dropped chat GPT on its from a great height in November 2022. There’s been a lot of angst about cheating and plagiarism, which I’ll write about in part two. In the interest of transparency, I think that for now, it’s helpful if people explain how they’ve used generative artificial intelligence Gen AI. But in the future, the technology will be so ubiquitous that we probably won’t bother. If you can imagine a disclaimer like this for I’ve used spellcheck in Word. I haven’t relied on chat GPT to write the content of the book, all my blog posts, but that’s mostly because I enjoy writing. But I do use ChatGPT and other Gen AI tools daily. And that includes in the construction of the book. So here are some ways that Gen AI was used in the process.

First, I created a simple piece of software using ChatGPT that used a programming language called Python to scrape my blog posts for articles in the AI category. The code, took the contents of all of those posts, stripped out the website code and images and copied them into individual word documents. And then shuffled them around and edited those documents as the basis for many of the chapters in the book. So probably the biggest use of ChatGPT in creating the book was was the just the creation of little snippets of code to do admin tasks and tasks, which allow me to pull together all of that work from 18 months. 18 months worth of writing on the blog was around 45,000 words worth of draft material.

And once we decided on the, the table of contents and the structure of the book, I just dragged and dropped those around into those different folders, and then started to edit that myself and then handed it over to Rica, who’s also joining us today, for a fantastic job of human editing over the top of that, then I got a an in person proofread from one of my co authors on a previous book, Ben lighters, their sort of teacher’s perspective on this. And very, very rapidly, we pulled this book together working across just December into the end of January, which is an incredibly quick turnaround. And I think part of what that suggests is some of the changes that that some of these industries like publishing and education are going to face you know, I spent a lot of time writing those original blog posts. But then to pull them together and turn them into something like this final book that really did not take anywhere near as long as doing that in the in the past would have.

Leon Furze 9:14
I used another piece of code, also written in ChatGPT to compile the final chapters in the draft manuscript, which saved me about six and a half minutes of copying and pasting, very important for the original articles and some of the new and the updated chapters. I recorded the drafts in a voice memo and used otter.ai to transcribe them and GPT for to remove the errors and the repetitions. And I use chatGPT and occasionally Claude for working out the tangles in some of the clunky and more overwritten paragraphs because I tend to write how I speak with lots of punctuation and enthusiasm. So most of those prompts were phrases like take the following paragraph and make sure it actually makes sense. I’ve used image generation, obviously, in various places.

The there’s an AI iceberg analogy in in the introduction, which was a candidate graphic, pretty much everything else was aI generated. And even for those mostly administrative tasks, I still feel a little uneasy writing, I used ChatGPT while creating this book. But I think that’s something that we’re all going to have to get used to. So that’s that’s the about the book. And I put that in there, really, towards the end of editing the book together. Because I do think that it’s important to talk about how people are using generative AI, there’s a lot in the moment in the in the tertiary context that I work in. And in K to 12 education, of people sort of using AI is like a guilty pleasure. And you might have had this feeling yourself, if you’ve ever used chatGPT to write something or to help you draft an email, or you kind of feel a bit awkward about it afterwards, you don’t really want to talk to anybody about it. And we’re definitely getting that with students, you know, students are using generative AI. And then and just not telling anybody, because they think that it’s cheating, or they think that they’re going to get into trouble for it. And adults are using it in all kinds of ways.

You know, teachers have been using it to write comments for reports or items for annuals, newsletters and those kinds of things. And nobody’s admitting to it, because there’s still that kind of shifty feeling that you’re doing something a little bit illegitimate. And I think the only way we’re going to break away from that is people being really open and transparent about how they use the technologies. The Australian framework for generative AI in schools, which was just published at the end of last year has transparency as one of its core principles. And that actually means transparency for schools, teachers, parents, students, everybody all the way down. And so I think, in whatever industries we’re working in, and I know there’s a lot of authors in the in the group watching today, we need to be cogniscant of how we’re using the technology a little bit critical of how the technology works, but also open to to sharing how we’ve used the technology, which might not always mean I use cat GPT to write x, y. And Zed might be something like I’ve just demonstrated there.

But I think we can be open and honest about all of that. I’m going to read a little bit from the introduction, which also just talks through the structure of the book. And then we’ll start to open up for some questions and some discussion. So I applied for my PhD in 2022. interested in how large language models like opening eyes GPT, two and three might impact the way we write texts. And at the time, you could use open AI models as a developer by building an application on top of them. Or through the open AI playground which offered limited examples like short story creators, grammar correctors and sort of little toys basically. Just a couple of weeks after the official start date of my PhD in November, open AI released chat GPT. And that simple act of placing a chatbot interface on top of one of its most powerful models, and releasing it for free to the public, created a tremendous shift in how we interact with generative AI.

And since then, there’s been a lot of hype, but also some impressive developments in the field. For educators, the release of chatGPT also caused a fair amount of stress and anxiety. Media reports discuss the threat chatGPT posed to traditional education, academic integrity and even writing disciplines. I’m sure many of you will have seen those articles in The Atlantic saying things like the end of high school English and the death of the college essay. And articles like that is still being published. Right now. universities in Australia and worldwide moved to pen and paper examinations to cope with the implications of the technology departments of education in most Australian states and territories. Bend chat GPT, partly due to problematic terms and conditions, but also to address rising concerns over plagiarism and cheating.

So we saw that worldwide, you know, the group of eight universities in Australia came down hard initially and said we’re going to look for pen and paper alternatives to examinations, we’re gonna move to practical assessments. And they didn’t publish a more flexible policy on Gen AI until September 2023. So it was quite a long time. The Russell Group in the UK, which includes Oxford and Cambridge universities did a similar kind of trajectory. They started off with a very fearful approach and Oxford University in particular locked everything down to handwritten assessments. And then in July 2023, but earlier than over in Australia, they released policies which said look, we know it’s not going anywhere, we have to find ways to use it. And that’s pretty much been the case across the world. In the following months, K to 12, and tertiary education continue to grapple with the implications of the technologies for student use, particularly in generating texts for essays, topics and examinations.

But despite academic integrity concerns, the technologies have potential positive and creative uses, including as an assistive tech to support learners. Between November 22 and November 23, there were also significant advances in Gen AI tools for multimodal purposes, including text, audio, video, image and code. And the industry has moved obviously incredibly quickly outpacing education, government and the law. So rather than focusing on church CBT, for any or any other specific apps and services, this book takes a broader view of the technology and explores ways educators might benefit from AI in the day to day work. And the aim is to help educators learn to use the technology themselves, suggesting the best way to learn is through experimentation. I organized the book into six parts. As I said, a lot of that came from existing articles on how to use generative AI on the ethics and so on. And then some some of the new content, much of which as I said in the about the book part I wrote verbally rolling around on the farm, talking to myself into voice memos, and then use AI to tidy up those transcripts.

Part one is how the technologies work. Looking at the importance of understanding how generative AI models are actually constructed. Part two is the ethical concerns, so bias, intellectual property, privacy, data usage, and a whole range of complex ethical issues which surround this technology. Part three takes a look at guidelines and policies in schools in Australia and internationally with the UNESCO guidelines, and then suggests how schools can actually write their own guidelines and update existing policies. Part four is how to use prompts, how to use text based models. Part Five is How to Use Image generation. And Part Six is multimodal Gen AI and the near future of AI, so audio video code generation, and then some of the potential overlap that’s on the horizon with virtual reality and similar technologies, which will become huge in the next 12 months.

Now, with the release of Apple’s vision pro, we’re going to see the combination of AI and virtual reality. It’s going to be pretty wild. Tim Cook, the CEO of Apple only just started talking about AI, Apple have obviously been using AI and their products for many years. But it’s off brand, Microsoft and Google have been making a big song and dance about their AI. Apple have played it down. Tim Cook’s really only just announced that in iOS 18, the new Apple operating system for iPhones, there’s going to be stacks of artificial intelligence. So they’ve got a large language model, like, like GPT, sitting underneath a new version of Siri, Siri is actually going to work apparently, for the first time ever. We’re gonna see all kinds of incredible advances. And alongside that they’re obviously pumping a lot of energy into their spatial computing technologies. So that’s, you know, a little, a little preview of the book and a little bit of a window into how the book was created. And I’m sure you know, a number of you will have seen my blog posts over the last couple of years and will have seen

Leon Furze 18:18
various articles and some of my opinions around generative AI. For me, the key is that, you know, we know that there are huge ethical concerns. But there are ethical concerns with every technology that we use. You get into a car in the morning and drive to work. There’s a slew of ethical concerns, you pick up a phone and doom scroll through social media for nine hours. And there are serious ethical concerns there. You know, we know there are issues with students using digital technologies, students using social media platforms like tick tock and Snapchat, this is not new territory for us. The technology took us all by surprise. And to an extent, the way that we use generative AI might end up being different to traditional digital technologies. But the approach to dealing with these things ethically is nothing unusual. And that goes for education as well. For many years, you know, teachers in the classroom have been on the frontline of helping students to understand how to do things like digital cybersafety, digital consent, using technology appropriately. We don’t always get it right.

It took about 10 years before we realized the harms of social media, for example. And we started to really cottoned on to some of the bullying and the mental health impacts particularly on young people. I actually think we’ll do a better job than that this time around with generative AI because we we sort of know what we’re in for. So my approach is to keep all of the ethical things in mind, put it right at the front, you know, literally the second part of the book, in this case, and then go into the practical applications, the creative stuff, how we can actually use it. If you’re if you’re an educator, how are you going to you use it in your day to day work, just the mundane stuff to start with. And then increasingly some of the more critical and creative thinking that we can use. So what I’ll do is open up for any questions, any discussions, opportunities like that. And you know, feel free to unmute and talk, live or drop things into the chat. And I will, I’ll keep one eye on the chat whilst we’re answering any of those questions. So if somebody wants to kick off the discussion.

Speaker 1 20:33
Hi, Leon, I’m Boney, how are you? Yeah. I’m currently teaching year 12. English. So the concerns with all the meetings I I’m, I’m happy with the new technology, I’m happy about teaching students to act responsibly and to check about difference but what the concern that comes up during meetings is that if they use chat, GPT to do their submissions and all their you know, sex and stuff, what is going to happen when when is the exam time and they actually have to write and the the concern that teachers have is that they will lose that they will lose that originality or, you know, they lose that the all those sort of skills to write for the exams. And so that’s, that comes up at every meeting that we talk about, although where I work, I will get virtual school, Victoria is fully online. And they have very proactive about using AI and how to help students use them. But in the team meetings, that this is the pushback that comes all the time. So just wondering if you’ve got something that I can take back to one of those meetings?

Leon Furze 21:57
Yeah, I’ve had a few of these conversations with Martin, virtual schools. I’ve had quite a few conversations around these technologies, particularly in your context, as an online school, and I’ve had similar conversations in New South Wales with with distance education providers, you’ve probably got the hardest deal here. Because obviously, in traditional schools, my answer would be, look, students can still be doing a lot of the planning the idea generation, the notetaking annotation in the classroom, and you can authenticate work, to an extent using all of those just traditional Nethers measures with no device, you guys are a bit stuck. Because if you tell your students to not use the device, you have no control over that whatsoever. So one thing I would say, per virtual context straightaway is that there is no way to police or guarantee that students aren’t using generative AI. So your your policies, your approaches have to be really inclusive of the technology. There’s, there’s literally no way around it, you’ll spend far too much time wasting time on detection tools and things otherwise. And you know, having had conversation with Martin, and I’m sure he and a lot of virtual schools Victoria has to have agree with that. I think with with English, it’s really tricky, because it’s a compulsory subject. And, you know, a lot, a lot of students don’t want to do it, not to put too fine a point on it. And it pains me as an English teacher of 15 years to say that, but you’re always gonna get that kind of bell curve where the majority of them are hovering around the middle somewhere, and then you know, some of them slide off.

I don’t think that’s going to change too much with generative AI, the examination isn’t going to change. So will you know, we will still have the pen and paper examination, students who have relied too much on chat GPT, throughout the year or similar apps, will will find that they struggle in the exam, they will probably be the same students who would struggle in the examination anyway, for a number of reasons, either apathy or their skills in the subject area. So I think what we need to look out for is ways that maybe these technologies can be used in an assistive tech kind of way for students who have learning disabilities, or as a way to help students generate ideas, if they really struggle to get things down on paper. We still want them to put pen to paper and paper themselves and come up with original ideas. But the students who do really well in English, tend to do so because they’re in it for the love of the game anyway, you know, they will continue to do well, those students who, who right there, the eights, nines and 10s, in the examination will still be those students, they’ll still do a good job of that. I don’t think generative AI is going to change all of that I disagree with the articles and things that say he’s going to spell the end of high school English and so on. What I think a lot of those things come back to though is that maybe some of the writing that we do in the English curriculum is a bit too generic and a bit too formulaic. And it is just jumping through hoops. And so overtime, I would really love to see assessment practices and the system around English changing in acknowledgement of that. But in the short to mid term, I don’t think it’s going to significantly impact our students and how they fare in the exam. Good test start with an English significant question as well.

Speaker 2 25:23
I mean, maybe to talk a little bit. Yeah, I’d love if you could talk a little bit about the current state of AI detection, both for written work from a plagiarism point of view, but also thinking about like image detection, and some of the, you know, images we’re starting to see coming out, that really lead us to kind of, I guess, question truth? And is AI detection a reality now? Or what is the future of that? Look?

Leon Furze 25:52
Yeah, that’s I mean, there’s two sort of aspects to this one is detection from from an academic context, plagiarism tools, traditionally now becoming detection tools, and the other is detection from that kind of legal and mess or disinformation perspective. So I’ll talk about plagiarism first. I think plagiarism is a big industry. Plagiarism makes a lot of money for a lot of people, the collection and the use of student data, these models that are rolling out that have traditionally been tasked with catching plagiarism, which are now shifting to catching a I have a business model, which I’m not I don’t necessarily agree with I think it’s it’s a business model that’s contingent on taking students intellectual property and using it. So that’s a personal stance, in terms of how these technologies work, detection tools, for AI generated text don’t work. They don’t work well enough to hang an academic integrity claim on. And I think that’s evidence not just through the research which has come out which which shows that it’s, you know, maybe in some cases 80% accurate, but that it can be gamed the students with access to a better quality model can do a decent job of beating a detection tool, that they disenfranchise English as an additional language students, more false positives, and you know, all of those kinds of problems.

But it’s also evident in how companies like Turn It In have changed their marketing. If you have been following Turn It In initially, they were marketing their detection tools alongside their plagiarism detection software. Now, they’re marketing it as part of an AI writing an AI editing suite. And they’re actually marketing it as a tool for students to use to self assess and to see if they’ve been a bit too heavy handed with AI and then to make some suggestions, and they’re actually incorporating AI grammar checkers like, like co pilot and Grammarly use into their own platform. So I think if you know, if you look at the way the winds blowing, and you take turn it in marketing backflip as any indication, then certainly that suggests to me that, that we all know that detection software doesn’t work well enough for academic integrity purposes. I think that we need to change the paradigm around what is plagiarism? What is originality, and some of those conversations around assessment would be more much more healthy. The other side of things, the deep fakes. Yeah, this is obviously top of mind for governments and police forces and places like the AFP, we will end up I think, for a long while in a bit of an arms race between detection tools and generation tools. The way that image generation tools work is literally by taking a generative model and a model which tries to catch it and pitting them against each other, until they iteratively get better and better and better. So the technology itself is almost founded on the idea of beating deepfake detection. But there’s there’s a lot of really interesting research around image forensics and artificial intelligence.

And image forensics, you know, predates AI by decades. Even things like aI image generated images, the shadows and reflections in those images don’t line up accurately the the position of reflections, the vanishing points don’t line up the the shadows don’t point to correct light sources. So people are finding really interesting ways of doing sort of old school forensic analysis on digital images, which at the moment are proving just as good as as machine learning based methods to catch things. So right now image and video and audio is reasonably easy to detect. In 12 to 18 months, it will get harder and harder and harder and harder and and laws and regulations are going to have to catch up with that. But yeah, good really good questions to two sided question. I think one on the plagiarism and one on there kind of Deep fake and legal issues, but both really important things to talk about.

Speaker 3 30:08
Leon, it’s Mac Yeah. Matt, how you doing? Well, thank you. Thanks for the excellent contribution in the book and all the work that you’re doing. It’s so helpful guiding people. I wondered if you could talk about the challenge for schools that for the last 20 years have been working in building all sorts of protections around student information, or the school’s own intellectual property or teachers materials. And the use of that is if you just described exploding by happening in chat, GPT, or out in bar and other areas that don’t sit anywhere, at all within the managed schools, environment of content or student information, you talk a little bit in helping people build their policies in your work, but it just seems to me there’s sort of two systems running and aren’t speaking to each other yet. And have you thought, Oh, could you comment on the risks that that might create, or the ways in which people can bring them back into a kind of complete managed environment? Yeah.

Leon Furze 31:13
Calm conversation around sort of schools and education and intellectual property is always an interesting one. And different jurisdictions handle this differently. I know I’ve worked with colleagues in schools in the past, where their their schools have been really strict on, you know, whatever you create as a teacher in this school is the intellectual property of the school. And I mean, don’t down here where I’ve worked for the last few years, luckily, we’ve been fairly Cavalier with that. And I’m generally fairly Cavalier with my own intellectual property as well. I mean, I’ll post everything out there for free. And I haven’t put the robots dot txt, to stop my own website from being scraped. So the idea of schools kind of walling off their intellectual property, I find that problematic, actually, because I think that education, good education systems are predicated on the sharing of knowledge.

And one thing that I see with these, particularly the competition at the moment between the big foundation models, so Google, Gemini GPT, for a meta open sourcing their llama model is that each time the model is updated, we used to have open source research. And we could, we could sort of share and see how these models were built. Increasingly, there’s proprietary data being built into these models to make them more successful, and to fine tune them further and further and further. And I wouldn’t want to see that kind of logic, replicated across schools. So I wouldn’t want to see, for example, a school, deploying a language model based chatbot, that sits on top of its own data. And then kind of hoards that data like like a dragon sitting on top of a pile of gold and says, Well, we have the best data, we have the best intellectual property, we have the best teachers producing the best content. And so we’re going to build a chatbot on top of that, and we don’t want anybody else to use it. I think that’s really unhelpful.

And I also think that that will potentially worsen divides that we already have fairly well entrenched in the education system, it’s not hard to imagine that a wealthier school that can afford access to teachers who are coming from different kinds of universities, or they can pay their teachers more can then wall off their content. In terms of student data, it is important, obviously, that schools find ways to protect and guard their student data from leaking out of those circumstances. And I think that’s really a real responsibility of developers, particularly like Microsoft and Google who are going to be the biggest players in this space. Microsoft’s already putting copilot into schools, Google’s not far behind, rolling out things in its workspace for education packages, it’s a responsibility of those developers to make sure that school data is treated with the respect that it deserves. So again, sort of two, two answers to your question, I guess, one regarding intellectual property and knowledge sharing, and one regarding student data and privacy. But does that sort of address your point? If not answered?

Speaker 3 34:17
Well, it does. Well, no, it’s terrific. Thank you. I do think it is right to keep this separation between sort of knowledge creation being shared as widely as possible for the benefit of all. And yeah, this distinction about sensitive student information, particularly where, you know, the commercial market is very good at triangulating that information for all sorts of reasons. And it I do think there’s a question about the degree to which Microsoft or Google even if they’re building them inside their products are building additional protections for the students and creating the benefit of that for school. wasn’t educators. And maybe this sort of as a follow up question is about how well do you think you’re saying schools or systems engage with that degree of nuance? I think there’s, you know, people are either in a world of oh, well, it’s happening, let’s just let it run, which is dangerous without putting some sorts of constraints or policy positions around it, or banning it altogether and saying, Well, this just kind of happened inside the school guide, and then it’s happening outside, but I don’t, I’m not yet seeing from Google or Microsoft, the appreciation of different classes of student data or information, particularly as it relates to assessment or well being that, you know, really needs some special treatment. And that may not come without it being demanded from by education systems or schools.

Leon Furze 35:43
I think one thing that I’ve noticed is that, I mean, it’s really the business enterprise conversation that’s leading a lot of this and that the education context is very different. So we, we often see this with education, edtech, and digital technologies is that there’s a business or an enterprise product that just gets kind of ported into schools, and it’s not fit for purpose. So I would love to see in the next kind of six to 12 months or so a little bit more research and a bit more work happening around how to contextualize these technologies, for education for students for K to 12, or for tertiary, and not just copying and pasting, the Microsoft Enterprise copilot license into an education context, which is actually essentially what’s happening now with things like copilot for Microsoft 365. So yeah, just Yeah, really good point. I’m just gonna pick up on a couple in the chat, and then then see if anybody else has wants to throw into the conversation. So just from pushing us to think about how genuine an assessment, traditional exams are, yeah, I’m not a huge fan of traditional assessments. Examinations. I’ve written quite extensively in the past about examinations, I don’t think they do favors for a great number of students for all kinds of reasons, particularly nd students, students with physical disability as well. So many reasons why traditional exams could be thrown out of the window as far as I’m concerned. Michael, given the trends, you’re describing, how far away do you envisage a new measures for success for senior secondary school and our exams? They’ll fit for the exam question again? Well, look, no would be my answer. They’re not. But I know, so many educators and organizations across, certainly across this country and globally, who have been just pumping so much effort in the last five years, particularly into breaking down assessment forms and looking for alternative assessment forms. So you know, I worked with the University of Melbourne, in my last school on their new metrics program, which is now extended through Melbourne assessment sites in South Australia. They’re doing heaps of work with their learner profiles. There’s organizations across the country and across the world who were doing exactly that. Mark in the chat, Just following on from what Michael said, the rapid advancements, and the wider state of play with Gen AI might prompt a rethink on a greater focus on practical skills and education, like the practical use of the technologies, I think the best way to use any digital technology is to get in there and kind of muck around with it. And historically, what we’ve seen with digital tech, and education is a very surface level use of digital technologies. So using them for him for information retrieval, for knowledge handling, you know, database, sort of kind of work. So we’re just literally, using the bare minimum of digital technology, students use Google search, or they use Wikipedia, or they use a word processor, and really struggling to get much deeper through that. So I would say I would love to see more practical use of generative AI technologies beyond really simplistic uses.

And then Meg desperately also needing to teach educators about privacy, and encouraging more awareness of student data. Yeah, one thing that I often talk about with teachers in schools is, is the kind of data that you can and can’t put into an open platform like GPT 3.5, or even a closed off one like co-pilot, which may sort of only retain your data for 30 days and then delete it. We still have to be really conscious of what we’re putting in for an open platform like ChatGPT 3.5 The free version. As far as I’m concerned, if you put student names or email addresses into that, that would constitute a reportable data breach. Just similar to going on an excursion with a bunch of kids and leaving a clipboard with their names and email addresses in a toilet by mistake. You would have to report that in a school and I think you would really have to report it if a teacher put personal identifying information into one of these platforms as well. Okay, probably got time for another question or two and then we’ll then we’ll wrap up for this morning.

Speaker 3 40:11
Can I ask the question from a primary school principals point of view? Absolutely. What’s the place of AI for four students in a primary school? So you’re looking at five to 12 year olds roughly. Yeah.

Leon Furze 40:24
So this is a bit of a moving space at the moment. Some technologies have K to 12 licenses already, although they’re not really singing very loud about them at the moment. But if your school has got an Adobe Creative Cloud license, for example, all of the students can access tools like Adobe Firefly for image generation, they could use AI tools within platforms like Photoshop and Premiere Pro, although you probably not using something like Premiere Pro in the junior school too much. But certainly image generation things like Firefly are wide open for use by students. Other technologies like copilots, at the moment, for example, have that 13 up terms and conditions. But you can absolutely bet that Microsoft are working on a K to 12 license that sits within their normal education products. So I talk a lot with primary teachers and primary schools, my focus areas generally on getting the teachers to use the technology first and get really familiar with it. But then acknowledge that in the next six to 12 months, it probably will be accessible to students. And I also say that, you know, particularly in grades five, and six, you can talk about things like bias and equity and cyber safety and digital consent with students that age, they are using the technologies, you know, students or kids from about sort of nine years and up particularly a using AI and applications like Snapchat, even though they technically shouldn’t be. So it’s still an important conversation to have even with that age group.

Unknown Speaker 41:58
Thank you. Thanks.

Leon Furze 42:03
Any more questions? And I do one thing I love about a zoom session is we all get to sort of sit here in awkward silence. While we wait for any questions. It’s always my favorite part of zoom. Okay, Mike just popped into the chat there, they’ve had a lot of school, a lot of luck with school AI in that age, sort of providing sandboxes Yeah, and we will find increasingly options for students that are targeted towards particular age groups, and that have maybe a few more guardrails and safety features on there, including things in Australia, which are going to come out of the Department of Education. I’ve just dropped into the chat as well a link to a webinar next week, which is the teaching AI ethics. So the section of that book on AI ethics is a kind of a truncated version of of some longer writing and some longer work that I’ve done around AI ethics, and I’m running a webinar next week, which is an hour and a half on six of those main areas. We’ve a lot of case studies of what’s happening right now. We’re really fortunate last week, when we did the live launch that author from copyrights, the copyright agency turned up and was able to talk through some of what’s happening in Australian copyright. And I think some of these issues are really interesting to teachers, and I’ll be going into those in quite a bit of depth. So thank you very much for attending this morning, or if you’ve watching the recording, I hope you’ve enjoyed the recording as well. And if you’ve got any questions, yell out, come and find me on LinkedIn. And I’ll hopefully see some of you soon

Questions or comments? Get in touch: