#284 - A Teacher's Perspective: It Has to Start with the Leadership Team (part 2)

Hello and welcome to Podcast. A multifaceted issue. While some institutions have embraced AI-powered tools for tasks like grading assignments or providing personalised learning experiences, Others remain cautious or have elected not to use AI in certain classes or departments. However, one thing is clear. There is no such thing as risk-free AI.

As these technologies continue to permeate our educational systems, it's crucial that we approach their adoption proactively rather than reactively. So, what better way to find out more about these issues than to talk to some people? who are leading AI in their schools. So today, as we explore AI's potential and its limitations for education, I'm joined by three wonderful guests in our Zoom studio. Adam Webster, Deputy Head of Innovation at Katerim School and CEO of Sphinx AI.

Scott Hayden, Head of Training, Learning and Digital at Basingstoke College of Technology, and Chris Goodall, Head of Digital Education at Bourne Education Trust. And so let's get going. I'm really keen to know what's happening in each of your organisations when it comes to artificial intelligence. So the first thing I'd love you to do.

is to say a little bit about yourself and your role, but then tell us what's happening with respect to AI in your organisation. How many people in your organisation are using it? How have you gone about it? Give us some insights for our listeners who might be quite new to thinking about using artificial intelligence. And Scott, I'm going to start with you. Thank you, Rose.

So I'm the head of teaching, learning and digital at Basingstoke College of Technology, a general further education college in North Hampshire. Got about five thousand students. We've got many learners who learn differently, got some apprentices, got some T level students. Very very vocational, engineering, hair and beauty, really hands-on. And with our two to three hundred members of staff, we've been focusing on using AI to free them up to do more human tasks.

Since May 2023, we turned on Google Gemini for all staff and have been using that to help with their schemes of work, with planning, resources, and we've started to see some really innovative uses of AI for helping enhance. assessment, feedback. And Ronnie in the last term started to see our students use it in controlled prototypes with the digital team helping them to create incredible evidence using generative AI.

That's brilliant. I love that you're seeing the AI as a route to create more opportunities for human interaction because that is what we need, isn't it? And I I love this idea of evidence. I'm interested in whether there were any reticent participants, so Google Gemini available to everybody. Did everybody engage with that enthusiastically or were there some people who perhaps were less keen? In May of last year.

Chris, one of our hospitality chef lecturers, had loads of planning to do and he was time poor because he had baby twins with his partner at the time. and Chris, by his own admission, is not the greatest at planning. Um so using his menus, his straggles of half written documents and notes on his desk.

We used Google Lens to zap them up, to use a particular prompt we'd worked on. And in two and a half hours he had his entire year planned. And he was challenged to think more creatively and innovatively about his sequence of learning, which freed him up to be more present. In the kitchen, in the classroom, and at home with his partner raising those kids. And Chris and stories like that become the way in.

In terms of people you wouldn't necessarily expect to use this really effectively. We focused on them in staff rooms and classrooms. Helping them to scratch their particular itch. And as we know with teachers, it's time. That's the thing, right? So we can have those stories with people like Chris.

And word of mouth spreads. And it isn't just, you know, Scott pushing the thing he thinks you should be doing. It's got to come from our wonderful teachers and staff wellbeing, giving you more time to do more human tasks, to develop those human skills that Now, World Economic Forum. and any number of different organizations have identified as being needed. UNESCO, Pearson, all these bodies have identified that there's compassion, critical thinking,

Emotional intelligence, empathy. We want our teachers in class focusing on developing human skills and using AI to try and free them up to have more time to focus on developing those skills in the live classroom, as it were. That's music to my ears, I love it. Um I also love the Chris example. Just one quick question.

You mentioned using lens and to some people listening they might not know what that is. Can you just give a quick Explanation of what Lens is and also maybe a little bit about what Gemini does, just for people who might not be as familiar as you are. Yeah, absolutely, Rose. So lens

And there is a Microsoft version as well. You can use lens to actually photograph some handwritten notes or some text from a magazine, for example, and to essentially suck up the text and turn it into a digital font, as it were, which you can then edit and adapt as you see fit. So we've been doing that quite a bit with a lot of our and we favor handwritten notes because the science is pretty robust about note taking handwritten

is pretty undeniable in terms of its um improved impact on retention. So our handwritten notes are taken and then adapted into text and then used to Be adapted in our tool like Gemini or ChatGPT or Copilot, whatever you're using. We use Gemini because we're a Google Reference College.

And Google Gemini is essentially a generative AI tool, a really good auto predict, a chat bot like Chat GPT, where you can speak to the large language model and to get it to do particular tasks for you. So that's what we've been doing and Our staff have been doing really cool stuff. Uh exemplars, case studies, content for slides, scenarios, starters, plenaries. They've got this thought partner now.

It's not going to do the job for them, it's going to help with some tasks and it's that way of framing it. that we've um approached this whole With. Great. That's great. Thank you. It's really good to get those explanations out so that people listening can think, Oh, maybe I could have a go at that. We're trying our best, you know, we've got a really innovative leader.

you know our principal and I can knock on his door and say, Anthony, I want to translate you into five languages to welcome our refugee learners. Fantastic. And he will let me do it, you know. So I don't take that for granted. And it's leader like that providing Opportunities for me with controlled, you know, mitigated risks always. I've got a three tick process.

Teaching and learning team, IT team and GDPR team. Anything we test always goes through those measures. But then Anthony, our principal, he's you know, he's he loves taking these chances'cause he understands. And is attuned to what's needed. And that's ultimately what wins out, right? What do the students need? And they need to understand how to bottle this lightning and uh use it well. So we're focusing now on a student AI module as well for September. Because we focused on staff first.

Next year is going to be helping the students to use this the right way, as our guest Chris Goodall today will no doubt touch on, helping students. um use AI well. In particular, his Snapchat example was wonderful and that really resonated for me, Chris, when I saw that on LinkedIn. Your use of teaching students how to use Snapchat AI, which is where many of our young people are. Um to use that well, I thought that was just wonderful uh teaching.

Thanks, Scott. That's fantastic. And so from Chef Chris to Head of Digital, Chris, Chris Goodall, you've just been mentioned and I'm going to come to you next. Thanks, Scott. Oh yeah, and thanks, Scott, as well, and thanks Rose. Um so I'm Chris Goodall, head of digital education for the Bourne Education Trust. To be honest, this is a fairly new role for me. So I I only started this role across the trust in September.

I spent the rest of my twenty five, I think, years uh in teaching and actually in classrooms, in and out, but um mainly in classrooms and when Chat GPT came around, I was using it in the classroom for most of the year, which is where I sort of gained my experience from. And then we rolled it out across the school. So the school I was in uh have now been using it and again in trying to experiment with it for about a year and a half now.

Um and now my role is to roll that out a similar model and what I've learned across the whole trust. So in terms of How many of our staff are using it? It's hugely varied across those 26 schools. So there are, just like in any school, you get possibly about a third of

the staff who are really pushing forward and innovating and experimenting. You might have another third that are are trialling various bits and pieces. And you might have some resistance and fearfuls fearful fearful staff. So we've got a real cross section. If I was to estimate

I would say pross possibly about sixty percent of staff. And and again, we focused on training staff rather than the students massively. We focused on let's get the teachers really understanding, let's get the support staff really understanding this. So we've done training for our teachers, we've done training for our head teachers and leadership teams, for our TAs, our office staff, for our parents. That's all part of the offer that we've we've put out.

For for every school, and we've got various uh levels of adoption. How staff are using it, it's again, it's hugely varied. So we've sold. Generative AI and we focused on large language models. So we started with Chat GPT. We use Copilot, but we don't r massively prescribe, although we we go for safety, so we recommend a safer route. And but we've sold it as a multi-purpose tool. So rather than say, right, here's AI, you can use it for this, this, and this, we've said, right, what's your problem?

Use AI to solve that problem. And again, this is where you get to resistant or fearful staff. This is where that approach is so good. So I as as well as giving a presentation around AI. We we I I do hands on training. And so where you get people in the classroom, I say, What's the next three things you were about to do? Let me help you and show you how AI can make that either better, quicker,

uh or support you in doing that. And again, it's not do it for you, it's support you in doing that. And yeah, I rarely get resistance with that sort of model because you're you're basically giving someone a hand to do stuff they're already doing. In terms of students, again, we're doing some very small scale pilots with students. In fact, I was doing some analysis of data this morning around various pilots. So we're we're at the early stage.

the students, but again, we focused on teachers because again when I was in the classroom using it, I was having loads of conversations with students about that use. And there was a lot, you know, ethical issues being brought up. There's so much that's brought up in a classroom when you're using it and experimenting with the teacher. We know the kids are using So uh it's it's it's about opening up those conversations. So that's sort of been been our approach.

to it. I I love the emphasis on teachers, but of course, you're right, the students are using this technology anyway and and they need to help as well. So it's great to hear that you're looking at some some students' um use cases too. I think it's really interesting this this model that you have. And I can see how well it works in an organisation like yours where you have somebody like you.

Who can lead the way and who is confident with the technology, can put people's minds at rest, can show them the practical ways to use it. But our benchmarking study is revealing that actually the level of training and continuum professional development available to the majority of teachers is not great. And therefore there will be a lot of people listening who probably don't have a Chris. How could they practically start to engage with this, do you think?

I think it's gotta start with the leadership team. Like any school initiative, this isn't I mean, forget AI, if you're launching anything in a school, you need to have the leadership team and the head teacher, the top of that organisation fully bought in. ac sydd wedi'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'i'

And I'm I'm generalising here. It's been a computer science teacher who's quite interested in a bit of tech, who generally launches it and there's real dangers to I mean, there's positives to that, but there's real dangers to it as well. So It's gotta start with a top top. It's gotta start with a planned approach to introduction of AI or any initiative. So I I would say in terms of starting that process.

You have also got to be organic as well. So there's a top-down element that needs to happen. But again, I I would say schools, schools in the past, certainly in the last five, ten years. have really grabbed hold of the evidence-based approach. And we started to see small hubs and teams in school. driving that where where staff had an interest and a skill in that. I think we're at a stage where you also need innovation hubs in school in a similar way where you

that you trial things on a small scale so that you can control that rollout a little bit, you can control the risks a little bit. And then when you lear what you learn, you can then share amongst the wider body of staff. So Uh y yes, I mean uh I I think the magic ingredient is having someone who's passionate, on the leadership team, and also can speak the language of human rather than tech. Those are your magical ingredients, but without that you're gonna need more structure in place.

I I love the magical ingredients and the talking human. That makes sense. So I guess it is a bit of a shout out to leaders, isn't it? Scott mentioned leadership, you've mentioned leadership, Chris, and it feels as if the leadership part is inevitably and as ever super important in this perspective. So I'm gonna come to Adam next. Thank you, Chris. That was great. I'm sure our listeners will really appreciate.

the kinds of tips that you've given them. Adam, over to you. Hi Ray's. Thanks very much. So yeah, I'm Adam Webster. I'm Deputy Head Innovation at Katerum School, also an English teacher. And I'm also uh the CEO of Sphinxai, which is a subsidiary company of Katerum School. So I originally started at Katerim a really long time ago with the mission of seeing if digital learning was going to be a thing or not.

And uh it turned out twelve years later, it definitely was. We were one of the first uh one-to-one iPad schools in the UK. And so we've been talking about teaching, learning and technology kind of integrating for a really, really long time. It has informed lots of the way we've developed the curriculum. We've in fact developed our own curriculum subject called Ed.

which speaks to so many of the values that uh Scott and Chris have already spoken about in terms of those sort of softer skills, the kind of outward facing, future looking, forward looking skills marrying together with the digital. That's what we do in Edge, which is very project-based, very collaborative. And so really when AI kind of popped up on everybody's radar, we were we felt quite ready for it. The first thing we did was was write a policy.

Which sounds terribly boring, but it's amazing when you speak to other schools how many have sort of stumbled at that very first hurdle, not really knowing what to what to do, what to write, what to say. And actually our our AI policy is less than the side of A4. And it basically says Students will tell us when they've used AI and we'll make sure we do the data protection bit really.

properly and carefully. And that's it. I think as soon as you start to dive into detail, you get stuck in the long grass and you never deliver anything. And so we did that. And then you know, very much as Chris described, we kind of uh looked at chat GPT. We made a really good decision, I think, in the the first people we brought into the room were the heads of academic departments. And we said, right, here is Chat GPT, here's the basics.

Now you go away and tell us what you think. Tell if it tell us if it's any good. Tell us what you can do with it. Tell us what you like, what you don't like. And we had that conversation over a series of weeks and getting lots of feedback. And that really helped us craft. uh our initial response. And we've seen teachers therefore using it in really interesting ways and sort of feeling quite empowered to do so.

Rydyn ni'n gwneud hynny'n gwneud hynny'n gwneud hynny. Rydyn ni'n gwneud hynny'n gwneud hynny'n gwneud hynny'n gwneud hynny. And that went re that that that went well. But I think there were kind of two major frustrations for me. One was a lack of being able to see what the kids were doing in Chat GPT as opposed to normal kind of web searches. And the second was really

Chat GPT and those other models are kind of the antithesis of good teaching and learning. They get the child straight to the answer without any learning struggle whatsoever. And that really bothered bothered us. And in a way that really changed the trajectory of the work I was doing with Sphinx AI, which was we were building a and continue to develop a very sophisticated pastor and academic track tracking platform which uses AI.

And we thought, well, we could we could solve that problem. And we actually built a chatbot which is designed for students, which takes care of all of those those concerns. So it speaks to the child using age-appropriate language. but it also has a teacher persona. So when the child says, write me an essay.

Our chatbot, which is called RileyBot, says, no, I'm not going to write you an essay, but I'll help you plan an essay and you can upload a picture of your essay and I'll give you some feedback on it.

And that I think really encourages that struggle of learning, which is so fundamentally important and is is ultimately at the heart of a lot of the negativity about AI in in schools, is it's just going to take away all of the the integrity of learning and we've really tried to address that whilst leveraging all of the good things about it.

That point about integrity is really important, it integrity of learning, integ integrity of teaching. I I'm going to come back to that, but I just want to roll back a bit because I was really interested in that sort of three step process, policy first. Second, you get the heads of departments in, get them trying things out. And then third, you look at the students. Going back to that second step, and it also, my question is also influenced by something Chris said too.

One thing I'm observing is that for very understandable reasons, most educators don't really understand AI. And so whilst in an ideal world, of course, you want to say, right, you've got this problem, this challenge, here's a technology that can help you. Let's map the challenge to the tech, because we don't want to do it the other way around. But because people have a limited understanding of AI, some of the biggest challenges that could actually be helped by AI are not.

Because it's not clear how you do that. So, how do you help those heads of departments to really see the potential? of the AI. Do you understand my concern, Adam? Oh 100%, Ray's. I think I think you're absolutely right. And I think in that bigger picture, Chris alluded to it already that

And most schools are not in a position to be able to have this conversation. In fact, this conversation probably seems quite ridiculous to them because they don't even have good enough Wi-Fi, they don't have devices, they don't have the infrastructure. So that absolutely feel this feels a million miles away from where they are.

But I think if you are in a school that is ready to have the conversation and that is the critical thing, you've got to be ready, gotta have as as we've taught, we've got to have the leadership in place that believes in it, that has the competency to make those good and challenging decisions. Then you're right, yeah. Skilling up people to be able to know what is possible is kind of your first big step. And I think you can do that in two ways. One is

you give them the freedom and the time to experiment. Very much like I remember my time right at the beginning of my journey with with the iPads, I was an English teacher in a classroom with a trolley full of iPads. And I taught some terrible lessons in order to figure out what you could and couldn't do with iPads. But I was super lucky to work in a school that says it's okay to make mistakes.

As long as we learn from them. And the beauty of it was I made the mistakes and then could pass on that learning to the rest of the team. And that's really important. But the second thing that we did. was we actually did training for for teachers, but also to the professional services teams in school to say, right, this is what AI can do. Here's some basic terminology. Here's some sort of things that are in place right now. Now you know that. Go away and look at the problems you've got.

And come and talk to us. uh about how you think AI might be able to apply to those problems. And if it is a good fit, we'll look at it. And if it isn't, that's okay too. But I think just giving people that baseline of This is, this is what it is and this is what it could do, helps open up the conversation about what we as an organization want it to do.

That makes so much sense and I think that's extremely useful for people listening. The very practical advice there. Just coming back to that point about integrity of learning and the way that these large language models They can do anything, can't they? And yet you've just reflected on your own experience of being an English teacher, being in charge of a trolley of iPads. And you said terrible teaching. I doubt it was that. But you made mistakes. And hey, isn't that how we all learn?

And these large language models don't really encourage you to make mistakes because they're always there to help you, always there to say, I can do, I can help you, I can help you with anything. And so I think this learning integrity piece. is really important. And it marries very much to what Chris and Scott have already said about the importance of future-facing human skills.

So last word for you from you, Adam, for the moment on that kind of that tension between the large language model generative AI and what we really want our learners to be doing. Particularly for people who don't have somebody like you who can build them a Riley bot. Well, I think

I think that that tension is is absolutely at the core of of the anxiety that people have. Like, is AI gonna replace teachers? Is a thing that comes up again and again, right? And the answer I think pretty categorically is no, because the human relationship

that are fundamental to the school experience are what matter more than anything else. Like I was speaking recently and we had this conversation as a group, be a primary and secondary, all sorts of different backgrounds, that the job of a teacher Is to help humans, little humans become better, bigger humans. That's and everything else is kind of a sideshow, really. And that remains fundamental. So what we want AI to do is be a partner for the teacher, not a threat, not

something that undermines what we know works and is effective for building those relationships, building com uh confidence, building resilience. If AI doesn't do those things, then it's a it's a terrible use of a really interesting technology. Absolutely and I'm sure a lot of our listeners. pleased to hear you say that.

Moving on now because we're always short of time on the podcast and and I do want to look a little more deeply the kind of risks and the safe garning issues because that's something we hear people are concerned about quite a lot. And Scott, I'm going to come back to you because you had this great acronym, well it wasn't an acronym, but the three-tick approach.

And I'd love you to say a little bit more about that, the three-tick approach, because it sounds like something other people could possibly try and implement. Yeah, thanks, Rose. Uh uh the three tick pros I've been at this college for ages, right? I've been here fifteen years as a teacher and then digital person and now I'm I observe and coach our teachers as well. And

uh we've got it wrong for many years, you know, and um we come to this through iterating and learning, you know,'cause that's as Adam touches on, that's how we learn, right? We we go through the mud, we we make mistakes, we dust ourselves off, we go again. And when it comes to our implementation of educational technologies. It's much the same. So I'll find stuff and I'll jump straight into it and give it a go, you know. I love exploring and being curious and playing with things and that's

The only thing I do really, I just play. I'm lucky to have a place a sandbox to play in, as it were. That's one step. I might like it, play with it, test it, much like Adam and the iPad. And you be the guinea pig, the prototype, as it were, and then you'll feed it back to the rest of the tribe once you've foraged and curated the thing that might be good for the rest of your folk. That's that's one tick though, that's just one thing. That's an echo chamber, that's a filter bubble, you know?

Um, we then in order to scale it up, have to go to the IT team. I raise an IT ticket and say, all right, you lot, have a look at this. Is it compatible with the system? Is it all right to use this? Because sometimes I've used things and it's not been good and I've and I've been told off. And rightly so. Wasn't aware of what I was doing. I was naive, you know, I was just playing. IT team will now tick it or not tick it. They might go no and therefore it stops at the second tick.

But if he gets the second tick, I then go into a meeting, which sounds more serious than what it is, but it's me opposite Greg or Claire, one of our data protection officers. I fill in my GDPR. data protection impact assessment form with my with my team. And my team was made up of students who've become apprentices and then members of our digital team. We'll fill in the form and we sit and we have it out. We talk about why this benefits teaching and learning.

And they will challenge me. I've literally got a meeting after uh this podcast today talking about uh some of the Gemini Pro licenses and whether we should proceed with that. Well, I've got to go in armed and prepared and to explain and justify everything from a safeguarding and ethical point of view. We've also got an AI ethics meeting every term. No, I'll take that back. It's every month now.

Where IT, Teaching and Learning and and the data protection team meet and we talk and discuss. Just because we could doesn't mean we should. We talk about these things at length. We have C P D hour every single week where we upskill on these different tools as well and help people to be more mindful and conscious of all these different things that have been thrown in our lap, you know? It's it's overwhelming.

And we try to make it as easy for people to understand as possible by continuously making it small. And um so schemes of work, planning in May of last year. That was the scratch to itch, that was the problem, to borrow the phrase of my colleagues on the call. So therefore it become pretty undeniable when you're a tired teacher and you want a bit of summer back, maybe an evening or weekend off from planning, to see it do that, you go, oh, okay, fair enough. There might be something in this.

So using it in that way is something we're really keen on doing, but all the while being mindful of the safeguarding side of it, the ethical side of it as well. Um, and just to touch on something that Adam mentioned in his response earlier on, I think this is something that I know Chris um will have thoughts on as well in terms of A by product of all of this generity of AI and this boom is that we we need to rethink assessment ultimately.

To rethink assessment and to coach and train our teachers in a patient, calm way to rethink assessment. So we've now got students using prompts to help them revise for their T level engineering exam. and relating it, like the one, this young lady in T level engineering, relating it to, you know, the the Airbus A three eight oh plane, which is her favourite plane. So she can understand and connect it, this complex theory about

aerodynamic airflow, um, in a way that she understands. And young Pender, who's one of our future pathway students and a somewhat non-traditional learner, is thriving by recording a podcast at his assessment. And he's researching preparing and rehearsing, you know, and developing understanding about the cause consequence reasoning and forward planning nature of this politics discussion he's been asked to do for his assessment.

So transcending the essay in the exam for more authentic assessment, vlogs, podcasts, professional discussions, witness statements. The rise of authentic assessment. That's exciting. It's it's to lean into that and to get us rethinking the essay is something I'm really.

Well, we're doing a lot of at the college and we're only going to do more of it moving forward. Again, that's great. And I think this business of rethinking the essay is fascinating. There was a piece in the Times Higher educational supplement last week saying how many academics were horrified. in universities by the number of very poor quality chat GPT generated essays they were having to mark.

And it's kind of, yeah, we've got to move on from the essay. It's not about policing the students, it's about moving on from the essay and rethinking the assessment. So I think these are great examples.

And it's lovely to hear about that AI ethics meeting. There's now monthly, that was termly, which demonstrates that the challenges that we face here, but that teaching and learning, IT, GDPR, you've got to get those three ticks before you can go.

Chris, I'm gonna come to you now and and you've already been kind of given a a call in by Scott around the assessment, but before we get to that, and and do please pick up on that because I'm really keen to hear because I know it's something you're very concerned about. But I thought you posted something on LinkedIn recently, which was a kind of a infographic about the work you've been doing looking at different use cases. And I thought that was incredibly useful.

for a lot of people out there who might be concerned about this. And I'm not really sure because most people don't understand much about AI and so they hear it's risky, but but what what are the risks and and how do we know? So could you start us off on that, but do please pick up that. Rethinking assessment piece at the end if you'd like to as well. Yeah, yeah. So I yeah, I put, I mean, basically because uh I'm seeing around a range of schools.

how people are using this and it's so varied. I what I try to do is go, how do we categorize this? And I was I categorized it in five use cases really. And again, the way I've categorized these are totally arguable. You know, there's so much nuance. to all of this. So what I I'm never saying I've got this right. I'm saying that these are these are some of my my you know, my my feelings, my thoughts around it. So uh when people st first started using it, I I considered

what people were using, what I was using AI for was very low risk. Creating curriculum resources I think is a very low risk use case. And the reason for that is the human is all always in that process. So when I'm categorizing the r these risks, I'm thinking of Whether AI is actually making decisions or not, I'm thinking of what's the impact on the number and range of stakeholders.

And I'm thinking about where the human is and how far the human is stepping back out the process. So those are the three things really I've looked at when I've categorised these rights. So creating curriculum resources, it was a creative activity. When I was using it, I was using it to create stuff.

Yes, facts are in there, but the humans right front and centre probing that, the same as you would any resource in your classroom, you'd look at it and go, Is this appropriate? Is this safe? Is this Is this correct? So for me, that's quite a low risk use case. I guess the second level I started to see was people using it for communication.

Particularly with parents, so report writing or or responding to email. And I think this is where there's an unseen risk here, because people see that as fairly low, and it is a fairly low risk, again, as long as the humans front and centre in that process.

I think when you're new to AI, you see it so amazing That you don't really anticipate that uh the response it's gonna get from per a person receiving that So, where I'm using it for communication, I'm knowing that there's someone at the other end of the AI actually digesting that information. And I think we're in a a realm at the moment with more and more people getting to know AI content and how it writes, that if people literally just fire off an AI generated email back to someone

If they perceive that as AI written, it's gonna feel I I I received one the other day. It felt like I've been dismissed. And I think we need much more emotional intelligence over the output of AI. And so I think there's a there's a real risk there. And I know teachers are using this to write reports. And if they're just using it to press a button and write a report.

People are gonna spot that. That you're the soul won't be there, and people will feel that. So I think that's a higher risk use case. Data analysis and again it's depending on the data that you you're doing. Again, it's m it that's a much higher risk because depending on the data, y y data protection is everything. So I I probably don't need to explain that one so much.

Then when we get into the realms of assessment and again I'll pick up on this this thing here is When you're assessing work, you know, we don't know if you're using AI to assess and mark. w where the bias is gonna come out. And I I think again, uh we could talk all day about this, but it's probably a use case that I would say

I'm a I'm a I'm not too sure about at this stage. I think it's gonna come. It definitely will come at some point. But at this stage, I think we need to be a lot more intelligent about our use of AI and and and looking back to assessment and risk. There are some risks I want to mitigate. But actually, for me, there are some risks I want to expose. I want the cracks to appear in the education system in some ways.

And we're gonna have to because of the way the future is moving. And so again we've got to be a bit nuanced and intelligent about what are we trying to nail down. And if we're trying to nail down the cheating. I think we're fighting a losing battle. We have got to adapt.

And so we want to expose those risks as much as possible because otherwise we're going to be in a position where we're trying to hold back the system and it's just going to creak and creak and creak and it's going to crack without being proactive. So I think there's a case where

We have to be a bit more proactive. I guess the higher risk highest risk use case that I've identified is moving towards Tutor chat box type situation where the again the human is really stepping back out of that process. Um and again, Adam's product, uh you know, I've I've I've had a we've had a meeting, we talked about it. I think it's really good the way it is, it works like a teacher. I think that's a real, again, intelligent, nuanced way of implementing.

But it's the highest risk because potentially we're going to see the teacher step right back out of that process.

That's sort of how I've I've categorized the risk. And again, it changes all the time. So these are all going to change all the time. So I'm always key to say, no matter what's put out. You can't stick with that for very long. And again, Adam's right with his policy. Is you try and put a lengthy policy in place, you're gonna be re rewriting that every day. So Absolutely. And and you know, you said

before you went through that, which was incredibly useful, Chris. Thank you. You're never saying I've got this right. I think we're all in that space. I don't think any of us feel a hundred percent confident. about what's happening at the moment. Because let's face it, the people who are building these technologies don't really know how they work. We're seeing lots of research papers coming out from the people behind

from OpenAI, from Google, from Anthropic, that reveal that they're still exploring how these tools work. And so what chance do we have? So but we have to do something. And so the kind of thing you've created there is practically very useful. And it helps people ask questions, doesn't it? So that they themselves can say, okay, is this a risk I think I can mitigate?

Or is it not? But I love this idea of exposing risks in order to bring about change. I think that's a a really good idea. So more of that. Thanks, Chris. Adam, over to you for thoughts about How we deal with the risks? How do we cope with making sure that what we do with AI is not going to endanger people? Because if we get it wrong, then the tremendous benefits that can come from AI are going to be lost.

So yeah, I've got a couple a few different angles to take on this, I guess. That firstly with the with the company with Sphinx AI, we have an AI ethics advisory board and we have some really outstanding people on that, like industry leaders in their field. But we also have five people on that board and two of them are current catering school students. One who will likely go on to study ethics and philosophy at university, but one who is going to study computer science.

And he was a really interesting case because he came to this meeting. They're obviously they held their own brilliantly against these sort of professionals, but he came away from that and sort of said, I'd never, I'd never thought about it from that point of view. You know, I just build this stuff and it's made me think twice. And I thought, That's the biggest win I could possibly have had in that scenario. You know, we've taken a guy who is going to be at the cutting edge. He's super talented.

of this in this field. And I have helped just open his mind to, I think Scott said it, just because you can doesn't mean you should. such an important mindset for for schools, but also for everybody in that industry. I think we have very much taken the approach at school that we need to educate the students.

you know, everything from deep fakes to bias and you know, all of that stuff. And but we But I think it's also important to say that from a from a people point of view, AI is not something you just talk about in isolation. Like there's a whole the tech industry as a whole is built to draw you into certain certain pathways. So we do so for as an example, we do a module uh where we talk about wireframing and UI.

I ask the kids in class to get their phone out, go to their favourite app and see how much of that app they can use using one hand. And ask them why they think that's so important. Like why can you do everything on TikTok with your thumb? Because they want to keep you scrolling for as long as possible, make that as frictionless as possible. And like you'll be and sort of explaining to them, there's a whole infrastructure designed. to make them stop paying attention.

And then we're saying no, stop and pay attention and look at and you and that goes all the way through to, you know, I I was sh I maybe very naively shocked when I I found out that there are lots of AI avatar accounts on social media that have kind of

As an example a couple of years ago, I think it's one of the early ones, of this looks like this, I think, Brazilian girl who has millions of pounds worth of sponsorship, but she's essentially built and controlled by some guy in a room in Somewhere in the state. And it was really shocking to think like these are the these are the new protocols that they're having to deal with every day. Like not just Is this information good information? Is this is this person even a real person?

Like what and like that's a a hugely complex world that the kids that we are all uh helping educate are growing up in. I think we have a massive responsibility. to get that right and make open their eyes as widely as possible without scaring them, because that's also really important. Because like you said, Ray's lots of good positive thing could happen and should happen. But I think only if development happens responsibly and users are skilled up enough to be discerning and make good decisions.

Yes, it's such a worry, isn't it? And when I saw Apple Intelligence being launched with the slogan that it was going to allow you to do more things effortlessly, it makes me just sigh and think, oh, but You don't learn effortlessly. You have to make some strenuous mental effort. So I think There is a real worry on many fronts about the way young people are being encouraged to use these technologies, sucked in as you say, by the way that they're designed.

You know, the consumerization of artificial intelligence is not going to do education any favours whatsoever and and probably not society. And we do really need to find ways to help avoid what's happened with social media, which has obviously been incredibly disruptive for many young people. So it is a huge responsibility, I agree with you.

Unfortunately, we're running out of time and I don't want to end on a com too much of a down note because there are some positive things as well. So I'm just going to come back to each of you and say What would your last word to people listening who maybe are new to AI, wondering where to get started? What would you say to them as a kind of encouragement, tip?

Also, if you have any thoughts on other issues that you want to raise on this podcast, please, now is your moment. I'm come back to you first, Adam, and then I'm going to come to you, Chris. Oh, it's a great question, Rose. I guess I would say don't believe the hype. but also don't don't believe the the sort of doom and naysayers either. I think that the the battle lines haven't been drawn yet.

I think what we've done at Caterum is a thing that I really hope happens elsewhere where teachers who know what it feels like to be in a school are co-designing solutions. We know that that is effective. We know that, you know, I can go and talk to schools and I can I show them Riley Bar. And I s and I show them like we've we've got built in safeguarding. So all the teacher that if you teach a kid, you can see all of their conversation history.

If you're a parent of a younger student, you can see that conversation history. It like inappropriate content gets flagged. And you show that to a school and they say, yeah, okay, you can tell you're a teacher because you, you know, you've understood what the concerns are. And you've made it transparent, you've made it easy. And I think that is how you

That is how you change the the narrative, I think, is show them that there's good there is good practice. I think your example, what you just said about social media is so true. Like schools. Schools got to the social media narrative a little bit late. By the time we got there, habits had been formed and ultimately we were then kind of reacting to it.

As a school, you can't point to a social media and go, that's what a what good social media looks like. And part of what sort of inspired Some of the thinking around the Rileybot project was if we can at least show because you know Chris is totally right, like Snapchat AI, that's where the kids are at, of course. But if I can build something that in a school, you can point to and go, this is what safe AI looks like.

Rydyn ni'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd i'n mynd. We don't have that with social media. So I think if we can grab that opportunity now, that could just sort of change the narrative enough. to mean that we get some really, really great stuff out of AI in the future.

I like that. That's a really nice practical tip. Look at what good AI looks like and then compare the other tool. so that you can see what the differences are. So you might choose to use one of those other tools, but you know what's going on. Of course, knowing is part of it, but the way some of these technologies are built that you were

talking about with Snapchat and oh TikTok and you know they're built to hook you in. That's what they're doing. So it's still a challenge, but that's very helpful I think. I also like what you're saying there about the battle lines are not drawn yet. And I think we have to try and make sure education gets a louder voice in what's happening. Thank you, Adam.

Very thoughtful, very helpful. Chris, you are gonna have the final word on this. What do you want to say to us? I I I guess I guess what I'd say is. If we're gonna get to the other side of this is in if indeed there is another side of this. We're gonna need everybody on board. Whether whether people believe this is the best thing to happen to humanity or the worst thing.

Actually, we need those skeptics fully on board and playing with this stuff because they're the voices that are gonna I I'm very aware my natural propensity with tech is to play with it and enjoy it. I I I hopefully have uh maintained some balance, but what we need is some of those people who aren't uh who are saying, No, this is wrong. W we I'm not gonna touch this. We need them to touch it because they're gonna be the best people.

to start putting on brakes. They're going to be the people with the natural skill sets to go, are you really thinking about this? So definitely I'd encourage everybody to get. I would I would also say this to get on board, the the way this is designed is to be simple. It's you know it's already allowing us to do things where the coding, for example, if you haven't got those coding skills.

someone who can't code can now do some coding. I'm not saying it's going to be the best coding, but they can do some coding. And so you know when when training people tank telling people to get on board, it's really keep it simple. Those key safety messages of data protection, bias, source safety and security, they're really important.

start with that. But the technical bit is can you write in a box and press go? If you can do that, you can get something else. So the technical bit is the easiest bit possible. And then the really the bit actually that requires you to get really good with AI is all your human skills that you've been developing. Curiosity, communication, you're communicating with AI, resilience when it doesn't give you what you want, creativity to try and think well how can it

How can it help me? How can it do different things? Building a rapport with the AI so that you know what it's good at, what it's not so good at, critical thinking and empathy, all of these human skills. Developing those with AI is going to help you become better with AI and learn about what the risks are, what the benefits are, and hopefully enable us to get to that other side. But we need everybody on board, not just enthusiasts. That's yeah, absolutely.

Clary and Claude, we need everybody on board. I agree. We do need everybody on board because Yeah. As you said, Adam, the battle lines haven't been drawn yet and we need to get in there. I also think that point you just made, Chris, about the simplicity is important for anybody who's listening who hasn't tried one of the really available.

So free is always a loaded term but won't cost you money. Um large language models out there. They are incredibly it is just a little box that you type in. It's incredibly easy to use. And I think sometimes people who are, you know, don't feel comfortable with technology can imagine that it's something much more complicated than it really is. So that's an important point to make.

Thank you everybody for joining me. Thank you to Adam. Thank you to Scott. Thank you to Chris. Thank you for joining me on the EdTech podcast. It's been great to have you on the pod today. Thank you. I hugely appreciate having Adam, Chris and Scott in the pod studio with me today. It's been great to hear all your thoughts.

and to have you contributing to this episode of the podcast. I know our listeners are going to get a lot out of the discussion that we've had today and in particular all those practical pieces of advice that you've offered. If you want more information on the series and our wonderful guests, visit the EdTech Podcast website at theedtechpodcast.com and connect with us via social media.

To see how the team here at Educate Ventures uses AI ethically to make education and training better for schools, organisations, and individuals, go to Educate Ventures. Or join the conversation in LinkedIn. You've been listening to the EdTech podcast presented by me.