Aug 11, 2023

In this Epsiode Dan talks to Travis Smith about many aspects of Generative AI and Data in Education.   Is AI the fuel that drives insight? Can we really personalise education?  We also look at examples of how AI is currently being used in education. 

________________________________________

TRANSCRIPT For this episode of The AI in Education Podcast
Series: 6
Episode: 2

This transcript was auto-generated. If you spot any important errors, do feel free to email the podcast hosts for corrections.

Welcome to the AI education podcast. Uh this is where we talk to great minds of the day around AI and its impact and in edu and and other contexts. And today we are super lucky because we've got the Australian Edu legend that is Travis Smith. He is a exchool executive K12 industry leader. Microsoft Australia, one of the best edu humans in the world and a an above average golfer, I think you'd say. Travel, do you reckon?
Is this about golf? Because if it is, I'm very excited.
It can be it can be about whatever you want.
Great.
How you doing, mate? You're right.
I'm very good. Thank you for having me, Dan. I look forward to a chat about AI in Edu.
In your role as the K12 lead, you've been doing a lot on this recently, haven't you? Can you explain a little bit more about what you've been doing?
Yeah, we have. We've been doing a lot on it for the last last um well since the start of the year really. But there's a lot of um a lot of hype about it, a lot of talk about it, and people trying to work out what's real, what's not, what they need to harness, what they don't, what they should ignore, and then let alone all the all the fear and the worry about AI more generally. And it it's kind of it's interesting because it's it's a topic that has never really been as mainstream as it is now. You know, everyone you talk to in the street
has an idea about what generative AI is and and has an opinion on it and that's not been you know that's not been the case before and we we know that AI has been a part of everything we've done for a long time you know whether it's education or otherwise but this seems to have really captured the imagination of of everyone so we're having lots of lots of good conversations and good good thinking about it
how does it actually lend itself in executives cuz I see a lot of stuff on LinkedIn with people going hey I've just found this new AI tool you know I've I've used this in this particular context, but I suppose you mentioned hype there. You know, I suppose some people are thinking whether this is hype or is it real like you mentioned and then also how do they actually implement that in a in a school or a school system. Uh I know you're on conversations around that. What are your thoughts on on that?
The first thing is it relates really clearly to their data strategy because we've been talking about data in schools for best part of 10 years deeply. you know about having a data platform and making sure your your ducks are in a row and you can get all information out of systems. Data is the fuel that's going to power AI basically. The the flip side of that though is that AI is going to fuel greater insights from their data than they've ever had before. So the ability for example for you know I mean cast our minds a couple of years forward the ability for for a teacher to write a natural language statement into a some kind of a data tool and have a dashboard built for them about the kids sitting in front of them is pretty powerful. So, so the data stuff is not going to go away and that's definitely one of the conversations that we've been having. The other main one is that you you can actually use these large language models like the GPT models in your own environment and and so because a lot of the fears around this stuff are actually because they're out in the wild. They're out on the internet, you know, they're, you know, and People are sort of worried about that because and as they probably should be because teachers might be, you know, accidentally putting personally identifiable information out there or you know there are certain use cases that in education are a little more um a little more worrying than others. And so you you know we talk to the executives and help them to understand that there is a way for them to bring those models inside their own environment which means that they can secure it and they can you know put some privacy and security around it and just like they do in everything in their network you know no one's just getting completely unfiltered access to everything on the internet in education. There's always some
some things in place and this is and this is no different. You know, there's definitely a few of those conversations happening around the country for sure.
Do you think it is there is a bit of hype around this like where are we? You know, you get a hype cycle where everybody's all frenzied up with it all and and I think I suppose I'm I'm trying I'm really trying not to be jaded about this, but like I mentioned in a in a previous podcast about COVID and how I thought that was going to help you know move things along but schools tick back into the norm they just wondering do you think personalization and stuff like that has really and changing assessment is it really going to make a difference you think this time
I think we've got the biggest opportunity we've ever had to do some real good in education
you know that data example I described before is one of them where you give the access to the data to the people who want to ask the questions without requiring any technical knowledge of them that could be profoundly impactful. I think there's a huge possibility for personalization. I mean, I I think that teachers now and kids now have the ability, provided it's done safely, securely, and everything else to have information personalized for them or to personalize information for their kids in a way that's never been possible before.
You know, I I was having a chat to a textbook manufacturer who makes sort of e textbooks and I was talking to them about the idea that the technology exists now that means that why does every one of your 25 kids have to see the same page? Like they might see the same sentiment, but
but if I'm reading three years lower than the other kid in the class, why why are we reading the same stuff? Like why aren't I getting a simplified version of it? The simple answer to that in the past is because it doesn't scale. You cannot do it at scale. But the technology can now do that. You know, imagine imagine You know, I went into my Bowen and Company e textbook. I've just invented a company for you.
Yay.
And yeah, and um I put in my Lexile reading level, which is my teachers told me, you know, you're a level 800, Trev.
And and then the the whole textbook language is simplified.
You know, that's the possibility of this. So, whilst I think there's some initial hype about it, I don't know whether we're um we should be thinking really big about the way that this could potentially change education forever. And I know that the, you know, there's work that Sal Khan and others are doing on creating personal AI tutors for kids. You know, the UAE Department of Education or Ministry of Education are doing a similar sort of thing. And I think provided it's done safely, securely, and it's designed by educators for the purpose of educating. I think it's a great idea, but you've just got to make sure that all the mechanisms for for controlling it are in are in place. And
yeah,
so I think there's Yeah, I I think that Whilst there is some initial hype, the reality the the way that we should be thinking about that this now in education is at you know especially at the system level which is where I sort of spend a lot of conversations with departments of education and Catholic dascese is what might we create that now we can do at scale that we could never do before and I think there's some huge possibilities in that space.
Oh there are yeah that's very true and there's so many I suppose it's an ecosystem of people that can that can help with this, isn't it? There's lots of I think schools are going to get an abundance of people doing whatever.AI that's a concern, but it's also a really good opportunity where people are looking at their products and going, how can I embed some of that AI stuff like the Nurture AI? I saw your webinars this week with Dylan William and and uh the nurture team where they put generative AI into the the feedback process uh you know in their in their work which is brilliant and it's about thinking outside the box. box really and thinking, okay, well, how can I add that to my tool to make that more effective? Um,
correct. And and I think there's I think there's ways that school systems could start thinking about how how to add this to their tool to make their tools more effective. Right? So, you think about internal platforms like student information systems, learning management systems, whatever. You know, one way they're going to innovate and build this stuff in is that the the the ma manufacturer, the designer of that proprietary product is going to build AI in at the front end, but there's nothing to stop them being able to build artificial intelligence in their local environment which is leveraging all of that stuff. You know, that is changing the way that their users interact with it through a power app or, you know, whatever other way
to start thinking about how to apply artificial intelligence around the systems that they've got in their environment. I think one of the biggest opportunities um with AI generally speaking and and people have a a kind of limited view which is completely understandable um if you're if you're not sort of thinking about it every day that what this generative AI stuff is that a human can write a sentence and the AI can respond in a humanlike way and bring bring to it all of this knowledge or you can ask it to create something in a graphical form and it can understand what that is and and create a graphic.
Yeah,
I think the real power of this is going to be less about that stuff and more about Fundamentally, what this means is that we forever change the ways that humans interact with computers because now we should assume that the computer understands what we're talking about. You know, I don't have to press 16 buttons in a certain sequence to make something happen. I'll just tell it what I want it to do and it knows how to do it. And so that's, you know, that's like the co-pilot stuff that Microsoft are are um you know, uh triing at the moment in Office where you just sort of tell it what you want it to create and let it create it. And I think there's if you extrapolate that idea out this this this fundamentally is the era where humans interact with machines in a very different way than before. And so if you put the large language model like a GPT model inside your cloud environment inside Azure let's say in in your tenant it's even telling that that when you go into the um you know the playground interface where you can actually start to train the model if you like. It's not quite, you know, it's that sounds more technical than it is, but there's an interface where it's got like what's the system message and the system message is basically where you type sentences about how you want this thing to behave and what you want it to do. So, you can say to it,
yeah,
you know, you are going to act as a an aid for teachers to help them uh design curriculum based around the Australian curriculum in whatever state. And you know, I mean, you can b if you do not know answer then say you do not know the answer. Uh never respond with blah. You're always going to put in hyperlinks to the point in the document you found. You know, you're you're actually and and all there's no coding in that. You're just telling it how
you want it to behave.
And then you can do stuff like upload sample responses. You know, here's here's an example of the way I want you to respond. And you can connect it to your own data sources in your environment. And
and that fundamentally, you know, is is the big shift I think which we haven't quite um I don't think at scale we've understood yet that this is this is these large language models are going to fundamentally mean that computers can understand our textbased sentence based or even voice-based inputs and do stuff for us whereas we had to do 55 clicks to get stuff happening in the past
yes it's so true isn't it that I think it's our user interface educational user it's a book for us educational user interface is 2.0 I suppose cuz it was um was it Sharon Oviet's future of educational interfaces I suppose it's kind of the next paradigm along from that you know the inking kind of stuff and it's the the natural input and conversations you can have with interfaces and it's not just edu is going to happen to our cars our mobile phones it'll be interesting to see what the next iterations of those operating systems will have in it as well
it will and the possibilities are only being sort of the surface is just being scratched at the moment. I think like I think the
you know once we work it out there'll be some really profound changes to the way we do it and and look I think back to your personalization question this is an opportunity to really level the playing field.
I mean we're never going to get around at the moment the the challenge you know the tyranny of distance the tyranny of access to internet or a mobile phone or a computer or a tablet or whatever that that is that is a lasting challenge. But if you think about the possibility of Every learner in this country, regardless of where they are, if they've got an internet connected device, having a personal tutor for their learning, that can be pretty profound. Not as good as having a human tutor, not as good as having another teacher or a onetoone support person sitting beside you, but we we know that that doesn't scale and can't scale. So, if these systems are designed by educators, so it's got the rigor behind it. And then to your point there, if the interface is right, Then we could we could come up with something pretty cool like you know imagine a student had a I don't know a tool a dashboard a thing that they then sort of said well what do I want to do here's here's the top 15 things that way that AI can help me as a learner you know rephrase this paragraph because I don't get it press that button paste it in done you know um simplify this language um give me um give me an alternative uh example of this, you know, whatever these inputs are, if we if we could find some way of making it easy enough for for kids to use that and we could put the protection around it, which we know we can do, then I think there's some real possibilities for for us using this stuff.
Yeah. And and you know what I was thinking as you were talking there, the the other interesting element to this as well in these in terms of these large language models is is that the the way you can connect with multiple disciplines. Whereas I I was just thinking as you're talking, you know, say 10 years ago or even probably even now today there's people making a living out there training um teachers in schools about apps on the iPad. You know, there's literacy apps, there's maths apps, there's this that and the other. Whereas with say a large language model like chat GBT, people are asking it to come up with maths lessons for them, geography lessons for them, science lessons. It's not so specializ So it's like one interface for everything. You know, I had a chat with a math teacher the other day who was trying to talk to their faculty uh about mathematics education and they were they were looking at the embedding inside the Bing chat um facility of maths and it's got connections into kind of um uh lex kind of graphics. So the the mathematical kind of um uh visualizations of things using those particular open formats and stuff and I I think oh chat GBT's got their connections into Wolf from Alpha and and all that kind of stuff. So it's almost like a
one interface is also the gold in here for teachers because they haven't got to go
well what app have I got to use for planning my maths lesson you know you can do everything in one area I suppose.
Yeah I think that's I think that's true and I think when we think about interface too it doesn't always have to be something that I type into you know I remember we were doing some work with um with good start early learning who run childcare centers around the country. And um you know we were we were sort of hacking if you like the ideas around you know what are the challenges for educators in their centers and and they use some great tools at the moment um you know to communicate with parents about their child's learning while they're at the at the at the um center for that day and sort of stuff they're doing. But the reality is that even though it's a it's a pretty um short process for a educator to gather evidence of something going on that's cool in little Dan's day that I want to share with his mom and dad. Um it's still when I've got 25 30 35 kids in a room and a few educators roaming, it's it's still an overhead, right?
And so if we think about the interface in, you know, imagine I could just say to my phone, capture a learning moment for Dan, you know, and then you know or record record a video of Dan about him playing with the frog next to the pond or whatever and then it just knows what that means and starts my camera and I record it and then it just does what it does with it like so that there's kind of educators can stay in the flow while give that kind of feedback to parents about not feedback that information to parents about what's going on in in learning and
you know maybe it's about learning four or five or six voice commands to document learning that somehow automatically automatically happen and sort of get stitched together to give a to provide that really important bit of communication back to the family. So yeah, I think the as I as I say we're kind of at the tip of the iceberg and thinking about it just with text and I think that the the possibilities are are much broader than that.
Yeah, I I agree. I I was speaking to a uh Catholic dascese the other day who you who then thinking about it in terms of they wanted to try to triage their call center. So it was about well like where can we how can we more effectively you know if a parent rings into the school system how do we triage that across so we're not putting people in a call queue or in a service ticket with something which is quite straightforward you know and bots have been around for a while around that but also what they were trying to do is analyze sentiment around conversations that people are having so that they can get sentiment analysis and actually prioritize things um around that. So it's like that's a different look the entire and I know we've talked about it for a while. There's entire business of schools that can utilize AI to smarten up those processes as well and be able to share those insights. You've got the two kind of elements I suppose that come together
I think. So and and there's other tools that are being created at the moment. Um you know, that are about um process automation within, you know, large systems like school systems or or schools or whatever. But there's actually artificial intelligence that's being designed to look across the data within your school system, let's say, and find the workflows that people are doing that are annoyingly slow. You know, if if a thousand times someone gets something from here, puts it over there, saves it as something else and puts it over there and then adds it to this system with a comment. Then the system can find that and say, you know, this is a process that can be, you know, expediated, made made easier for people um and then even um get to a point where eventually the artificial intelligence may be able to create a solution to the problem, create a workflow for it.
Yeah, that's so clever. And and then connection together, what I'm thinking is some of the stuff that we've got which you're using from a sales point of view, that would be gold. in a school point of view cuz if when I was teaching I remember you had so many kids the that came through when you did a parents evening and you know I had parents evening the other day and it's even it's even shorter these it feels to me that you know it's a 5 minutes in and out conversation whereas at least when you were going into the school not everybody got there though but it is more equitable cuz people are doing that via teams and things like that now but the um if I was a teacher and he was bringing that information to me you know sometimes you know I know this is bad practice. But sometimes you'll be you'll be halfway through the conversation before you even realize who the student is. Because you teach so many kids, you're kind of thinking it takes a while when you're having the conversation to say, "Oh, yeah. I I remember teaching David and you're looking at your notes and you're thinking where who's who's David again?" You know, which which David is it? I teach 50 Davids this year and like which one is it? So, it takes a while, but if you can have that personalized context before you even get to a teacher meeting to say, "Hey, this is David's background. This is David's family circumstance. You know, you might be speaking to his mom because his dad's passed away or whatever it might be. Uh and he's just dropped out of his sports soccer class and you know, this is his latest uh information. You know, the teachers in my recent um parent teacher interview, they obviously had a markbook cuz they, you know, like a traditional markbook. Some might have been using Excel, but they basically read to me three marks. based on what my kids had had, you know, like this is the this is how they performed in the last three assessments and they gave like an average kind of statement about them, you know, and and some had lots of really good insights, but
they I just imagine if they had more information, it would be even better.
Yeah. There's there's research that shows that there's something there's over 60 data places in a in your average school where data is stored about kids.
Wow.
And and so that's obviously teachers markbooks and it's LMS's and it's student information systems and it's all the traditional things but it's also the signup sheet in the in the school hall for the production and the you know the year 8 volleyball team list and you know all that stuff um the the instrumental music class lists all of all of that and it's it's basically just about um triangulating data the first part of it's about triangulating data to work out you know what is the full picture of of little Mary or whatever? What is the full picture? Um but then I think the second part as I alluded to before is shifting the ownership of that data to the people who actually want to ask questions of it. So one of the things we do see is and this is where the natural language input part of GPT models and the like are going to help. What we what we currently see is people creating data dashboards who are using data read people, you know, data scientists or switched on people with data
creating dashboards that don't quite land with the end user if the end user isn't a data user
or, you know, a data native because they, you know, I mean, they say take it on face value, but they can't do any other manipulation of it or drill down and read look at different variables or whatever.
So, you got to get it absolutely right, which is which is really hard in the complexity of a school. But flip that on its head and allow the teacher to write a statement about what they want to find out before the parent teacher interview so that they can just say show me show me um you know Dan's learning across the last six months across all subjects including mine include truency data blah blah blah blah blah and then bang up pops this rich report of information
but I suppose even taking it to the next level as a parent you know like if I got access to that I wouldn't need to have a parent teaching me I'm not trying to take the humanity out to teaching you because
I am okay well like there is more education system than the than the the context there. But to be fair, the reporting the reports that I've just had for my kids, you know, nap plan, we can pick those up and put those straight in the bin really because they're so old and outdated. They give you a bit of a litmus test of stuff, but well, that's my personal opinion anyway. I know they've got their own kind of value, but the general reports, you know, there was certain tick boxes on there, you know, does does my daughter do dance? Yes, she does. Does she, you know, what the extracurricular things? What is she doing in English and whatever? If I've got that information to hand all the time and I can ask that myself for that data and I could say, you know, how is Megan doing this week? You know, oh, by the way, you know, rather than the the um, you know, the the text message of your son's been late three times in the last month or whatever, he's now in a detention. You could correlate it all together and go this term, you know, he's tracking on this. You know, having that tracking, you you wouldn't even need those parent teacher meetings.
Yeah. I think all the conversation would be very different because it's not about information sharing, right? It's about a discussion of how they're going.
The I mean the other thing to say about that is too is that
you know even even the report con the content in the written part of a report is fairly contrived because it
has to be you have to be sensitive to everyone's needs. You know you
it's you know oh you can't say this or you can't say that or let's not be you know, let's try and find a positive way of saying everything. And, you know, sometimes you need to, you know, we need to have a a a good honest conversation, which happens in parent meetings all the time at schools. But I I do agree with the idea that what happened if you know, what happens if any parent in a classroom or any parent in a school had access to a set of data that they could also ask any questions of about their own child.
So, I think I think this this idea of, you know, data being the um fuel that that runs AI. I think I think it's true that AI could be the fuel that drives insight out of data because it it gives it gives different audiences who are not data literate per se. Like myself as a teacher, I I taught psychology, a bit of English, history, geography, you know, I wasn't a maths or or a science teacher who, you know, was big into data and stats and numbers and,
you know, so if I was presented something It was a bit confusing. I I couldn't sort of get my way through it. So, we yeah, we we we've got a possibility, I think, to think about, you know, not only the importance of data continuing, but the way that people digest data, um I think is a real possibility. And there's absolutely no reason in my my mind either while kids shouldn't have access to their own data.
Yeah, that's true.
Mention that. Yeah. Because if students can do that, then they can improve their own learning as they go in, you know. Um I know my bit of a when and I'm looking at my kids at the minute, you know, they're using GPT tools to kind of help them. So, you know, my my current example, I think I mentioned in the last podcast episode as well, is like my son was given the Handmaid's Tale to read for English and it's like it's pretty hard going the Handmaid's Tail even for an adult to go through it and and he got it to summarize and he wasn't going to read the book 100% he was not going to read the book. I could see in his eyes. So, he used Chat GB to summarize chapter at the time. So, could get a gist of the book and understand the context in the book. Now, that is negating from the fact that he wasn't reading and the point of him doing reading is to get used to reading and contextualizing and comprehension in text, but he did get an idea and a better handle on the book um and the context within it to do with the tensions between women and men and things like that much more effectively using summarization of chat GPT. But I think there's also for like we are in a bubble. We're in our own context bubble here and I think we we assume every teacher can use these things and they don't. There's going to be teachers out there that just type something into chat G GPT or Bing chat and go give me a lesson plan for your science volcanoes. It gives them some junk and then they go well that's rubbish then they move on right. Um so I think student agency is really good. I'm going to try to um speak to Nick Jackson because he's doing a lot u Dr. Nick Jackson down in um South Australia. is doing a lot with student agency and getting students involved in AI and how that works. But but when we step back from all of this stuff like as an executive team, you know, when you were speaking to these executives and departments of education and dascese and things, what what what should they be looking at? You know, if you had like a couple of simple things for them to do if they listening into this podcast, if there a couple of things that they could do now to kind of um help them manage AI and this age of AI going forwards.
Well, I think I think the first one is to is to keep that data journey going because we know that that's going to be even more important once tools are infused with AI.
The second the second one I think is to is to have um have discussions with all parts of the organization around what they could do with it to help them. So you know teacher meetings where I know that this happened in Melbourne uh a Catholic arch dascis where they pulled some teachers together and they had an amazing conversation about what teachers are doing, could be doing, might do with AI to save themselves time. We know we've got a massive crisis in the industry at the moment. People leaving leaving the industry. Teaching is hard work. It always has been. And um you know, everything is just piling up. There are real possibilities for us to fix this. And I think that we need to have good conversations with people about how to do how to use AI safely and securely to save themselves time to take a load off. you know, to sit get a first draft of something. Um, you know, I saw,
you know, a teacher who was had,
you know, a gazillion things on a to-do list, and one of them was to write an email to a to a student because they cheated in a year 12 assessment, and they had to come to a meeting with the assistant principal. It was, you know, a Victorian Curriculum Assessment Authority authority approved process. And, you know, they were just like, it's not that I can't do it. It's that I've got to sit down in front of a blank email and craft this thing. But with with Bing Chat, they were able to put in what they needed, obviously no personally identifiable information, and get a first draft of that email in about 2 minutes.
Yeah.
And then with a with another three or four minutes of re-editing it and changing it cuz it wasn't quite right, they were able to send it.
It's just we we've got to have conversations at every level about what are the things that we could do with this tool that are going to save time or what are the things we're worried about or what are the things we should be protecting or what are the non-negotiables we shouldn't be doing. All of those conversations need to occur because there's a gazillion use cases out there at every level of an organization like a department of education or a Catholic dascese whether it's the marketing team at the at the central office of a dascese or of a large private school or whether it's the you know the teachers in the classroom or whether it's kids or whether it's parents or whether it's anyone. There's so there's so many um conversations that need to be had. And the third thing I'd say is that you know it's possible right now to put these models in your own environment and protect it and try it. And so it might be just um you know worth send and we know that many of the departments and dasces are doing this now right they're getting the large language model they're putting it in their own environment where it's where it's protected and and only accessible by certain permissions and etc etc and then feeding it their own data so that um it can respond based on the knowledge base of that organization, not the knowledge base of the internet.
Yeah.
Some of which is rubbish.
So, so you know, get started with something small, think about what a use case is, but I think there's also um a conversation that needs to be had about not only having discussions and and meetings with different members of a uh a cross-section of a community, but also starting some basic training, you know, encouraging the you know, your IT teams in your organization to do some fundamental certifications or you know getting try and get ahead of the curve with this. There's teacher courses, there's you know there's stuff for for students like the Imagine Cup Junior stuff that we we run where kids can start thinking about artificial intelligence. There's courses for teachers, there's courses for IT people. There's
you know there's a lot of entry points into this and I think it's just about
you know knowledge is power in this space and and so you can have an informed discussion um about it. You need to kind of You need to get your head around what it is and what it isn't and what the threats are and what the opportunities are and and have a you know a conversation about it.
Yeah. So true. And when you're talking about the way they disseminate it, one good practice that I I found one of my ex-colagues back in the UK, Chris Goodall, he's posting a lot about AI at the minute in on his um on his LinkedIn feed. And what he does with his staff at that next level down from the executives, he does a he does a post every a session every week, but he posts on LinkedIn and he's he basically splits into three things. He does try this. So he'll he'll he'll put a prompt in an example. Um so it'll be like a something they can try, you know, today, you know, this week. Go and try this. Then there's something to watch and there's something to read and it's something short, you know, something different because there's so many tools out there um you know, for different contexts. So you you'll kind of say, "Okay, try this prompt, but put it for you your a lesson and then watch this video which is something the I don't know the Khan Academy is doing or something that Microsoft's released recently or whatever and then um you know and something to read as well which might be around ethics and AI or something. So you think about different modalities of teachers as well that that some will read something will rather go and try it some want to watch something. Um so there's have you seen any tools recently or got any examples of anything that that you've seen in edu that um that people have utilized?
There's lots there There's lots of them to your point before like everyone's coming up with a company called something.ai, right?
And there's so there's so many there's so many tools and and Microsoft's at a different end of the spectrum to that because we're sort of providing a platform for people to build
tools, right?
Tools are
um and and so I mean I if you look at you just join a Facebook group for example of educators globally or in Australia or whatever talking about how they're using AI, it's fascinating, you know, the the stuff that they're coming up with the way that they're discussing um you know quite quite um not sensitive topics but topics like um plagiarism, topics like you know is it a bit icky for teachers to be having autogenerated report comments? You know does that take the teacher out of the loop? You know they're having really good conversations and they're also sharing an awful lot of good tools.
Now obviously once all of this gets out in the wild then you've got you've got to know the privacy and the efficacy and the ethics behind what is happening with your data and you know all that stuff again. So
and that's why the you know the larger departments in Australia and New Zealand are thinking about bringing this in their own environment like South Australia have done you know they they've set up the open AI large language model in their own environment so that people can go crazy because they know it's safe they know it's protected they know there's not data leakage And so now they're exploring what are the use cases from that. So there's a there's a whole range of something.ai tools for summarizing large PDFs to whatever. And and I I should say too that you know the the the plagiarism challenge the um intellectual property is going to be a really interesting space and we're starting to see some challenges around that now.
Um you know there's a there's a whole lot of stuff that the human race has to work through here. Um cuz we we've got some pretty cool
cool tools and And like every other disruptive innovation, there's going to be some some things that we have to work through pretty seriously.
Yeah. No, that's really good point actually because it's going to affect everything, isn't it? I don't know how deep it's going to go into golf to help you out with golf, but probably um there'll probably be something that'll appeal at some stage. What you reckon?
No AI could help my golf mate. It's uh it's beyond beyond support.
Well, I said yes then. But no, of course AI is going to help everything. You something something will happen. Something something will definitely happen. So,
you know, thanks for joining us today on this uh podcast. Before we leave, I suppose, you know, is there anything that any one or two resources that you'd share that would be useful for these executive teams to kind of pick up on um to move forward in terms of AI in their in their schools?
Yeah, I think it I think it's well worth um getting involved in communities who are discing ing this stuff whether it's on LinkedIn whether it's on you know Facebook groups with educators in it you know there's lots and lots of conversations happening at the moment there's a course that we've run for educators called uh which is at aka.mai for educators and it's a training course where teachers who are starting on this journey can understand a little bit about what AI is a good a good resource but I mean there's a whole myriad of places that they could go if you're an IT um you know person who's working in an IT capacity then there's a huge range of training
yeah we got the fundamentals data I fundamentals um each cloud provider has got their own fundamentals training as well around that because obviously Google are doing stuff as are doing stuff as well and I I suppose the worry is you know without without ending on a bit of a downer here but my worry I was speaking to a diocese yesterday worry you know the old adage right and this isn't AI this happened years ago when I was when I was working in the school myself, there was a website called ratemyteers.co.uk and you could go in and you could rate every lesson that you go into and kids are going in and they were ranking uh teachers commenting about teachers. They could also do anonymous posts about teachers as well. Um and there was like the the entire UK education system and there was a global website as well. They're all, you know, came tumbling down. The same thing happened with the internet as well, you know. It was like What are we going to do about this? The same thing with calculators. Same thing with pens. Now we say AI. Um, you know, the conversation yesterday was about the worry that what happens if somebody puts the face of the CEO onto a naked picture, for example, or a picture of, you know, you know, like they've done with deep fakes of um Trump and and all of this kind of stuff and Barack Obama. Uh, and and there is a limit to what you can do, right? The cat's kind of out of the bag, so there is a limit to what you can do, but there are going to be cases where a company is going to do a brilliant tool like with GPT 4, they'll call it something, teachers will use it, they leak credentials into there because they need to log in and then, you know, these Russian hack groups or whatever or these, you know, state actors or whatever it might be, we'll just sort of mine credentials and stuff for teachers. So, you know, there's there's there's a bit this this isn't an old uh I suppose um conundrum and there are tools and security practices like you said about bring things in house to manage uh these applications when they're getting pushed out. But it's, you know, it is going to be a matter of time before something, you know, happens which is visible, you know, because you can't stop this because kids can go home and do this at home, right?
That's right. Yeah. And that's why it's so important to start thinking seriously about
how you can do this in an enterprise way, you know, which means you know, you know what that means. So, but you're right. I mean, there's it's it's no different to any of the myriad of tools that I have ever signed up for using a personal email address and a you know websites for this and shopping sites for that and you know it's it's all of that still that's not that's not a a new um that's not a new problem but it's it this is possibly an exacerbating factor for that
and it is it is a responsibility of everybody I remember a comment which somebody made to me in a school once was about you know they it people tend to get in the neck for any of the cyber stuff any the security things because of the policies and the technical implications but I remember an IT person saying to me if a student brought a knife into school you wouldn't take them to the woodwork class and and and say like this you know you know everybody's responsibility is security whether it's um you know and the use of these tools. So actually sometimes the responsibility does have to land on that teacher's door to say well look if you're going to utilize um these technologies that you can't close up because that's all we did with the rate my teacher site. We basically said, "Well, we can't block this cuz we can block people accessing in school, but they're going to go home. They're going home and, you know, um putting ratings to the teachers on there. We'll actually um try to embrace that and some teachers will put in the link to it. Did you enjoy my lesson? Give me feedback. Post on there." You know, you got to kind of embrace that. And I think some of the teachers are already embracing the tools like um Bing Chat and the like because they're actually sharing that information and saying, "Look, this is what I done. Or they might even say to the the kids. I saw an English teacher the other day was trying to reverse um reverse or utilize the tool for the um the im midjourney IM and Bing image creator. They were trying to utilize English. So using prompt engineering to develop kids English and descriptive writing. So who could come up with a best image and then you have to try to reverse engineer what prompt to try to get that image. And then she started showing images of like a a cityscape in the dark with a cat in it. And then the kids had to try to duplicate that with English narrative by going, you know, um, show me a picture of a cat in the dark with neon lights saying cafe and ra in the style of, you know, whatever painter. And, um, you know, it's really interesting the way that people can embrace these technologies.
Yeah. And I think that, you know, the creativity of of educators is just forever amazing and unlimited. Like there's very creative ways that teachers will think about using this stuff. Another example I saw was um this was months and months ago when when the plagiarism discussion was really on fire about this was when a student is asked to um you know do a write an essay or something at home or a response. If the teacher gets that electronically, what they would do is pop that into GPT and ask GP at to create seven comprehension questions based on that text. Then when the kids came into class, they'd sit down and be given the seven comprehension questions as half their marks.
Wow.
So if you didn't write the first half, like if you didn't write your essay,
that's clever.
You can't answer the questions. But if you did, you're fine.
And so, you know, there's really interesting ways, and I'm sure that's the tip of the iceberg in terms of creative ways teachers are thinking about
how assessment might change or some of the impacts of this. But anyway, I think we're at a, you know, although we're going through the Gartner hype cycle of this is going to be the biggest thing ever and then we're going to that trough of disillusionment it's called. I love the emotive language where where everyone's sort of thinking, oh, you know, oh, what about this and what about that? I I do think that we've got some profound opportunities if we invol if we if we involve um educators who are the specialists in learning in the way that these tools could be crafted to help kids understand more or the way they can be used to help teachers save more time and teach better.
I think we've got a profound opportunity at the moment
to change education for good.
Yeah, definitely. Well, on that good note, like my really sour note about the security element, thank you Tra for joining us today. Your insights have been amazing. I'll put some of the links in the show notes and things, but thanks Travy. You know, it's phenomenal. Keep up the good work that you're doing in Edu and supporting these systems because we certainly on a on on for a ride in the next uh couple of years.
Yeah. Thanks, Dan. Appreciate it.
No problem.