AI submissions are avoiding detection at a well-respected university in the UK

AI submissions are avoiding detection at a well-respected university in the UK
As exam results roll in for pupils across the Northern Hemisphere, a study conducted at the University of Reading showed that 94 percent of the AI-generated exam submissions went unnoticed. On average, these artificial responses received higher grades than those from real students. Dr Peter Scarfe who led the study, is on the show.

Wireless monitoring of audience reactions
How can a producer ensure a film’s success - apparently by researching how the audience members react to their production. A brand new facility at Bristol University in the UK will be able to monitor heart rate, blinking and brain activity to see when people become bored whilst watching a film. Professor Iain Gilchrist explains how the tech could also help to improve teaching.

The show is presented by Gareth Mitchell with expert commentary from Ghislaine Boddington. 

More on these stories:
A real-world test of artificial intelligence infiltration of a university examinations system: A “Turing Test” case study
Smart Cinema at the University of Bristol

Transcript

00:00:00 Gareth Mitchell 

Hello folks, it's Gareth. This is the somewhere on Earth podcast. And wouldn't you just know it's Tuesday the 20th of August 2024. Welcome to our little studio here in London and we have voices. Both our guests are remote today as in they're on Internet connections. They're not kind of remote in a psychological sense in any way. Right, that's enough waffle for what we call the little pre menu. Let's do it. 

00:00:31 Gareth Mitchell 

And with us today, giving us a little kind of pre chuckle before the full chuckling begins is Ghislaine Boddington again. Hello. Hello. Hello. 

00:00:38 Ghislaine Boddington 

Hello, Gareth. Yeah. Good looking forward to this show because there's two topics which are well within my very big interest area.  

00:00:45 Gareth Mitchell 

Within your purview. That's good to know. 

Yes. The use of ChatGPT alongside the human side, and especially the Smart Cinema One,  very much so, yeah. 

00:00:52 Gareth Mitchell 

Ohh, you're gonna love that. Oh, you've teased it so well, people that there now they want to listen on the edge of their seats. And I just have to either apologise or take praise for the fact I I have a slightly croaky voice the last couple of weeks. 

00:00:55 Ghislaine Boddington 

Hey, I'm teasing you. I'm teasing everyone. Yeah. Yeah. 

00:01:07 Gareth Mitchell 

Maybe you'll like my kind of quite low radio, gravelly voice. Or perhaps it's getting on your nerves. I'm sorry if I'm a bit croaky, but we'll get through it. It's OK. 

00:01:19 Gareth Mitchell 

And coming up. 

00:01:20 Gareth Mitchell 

Tonight. 

00:01:24 Gareth Mitchell 

The problem we're going to talk about today, this problem that's keeping academics up at night. They're incredibly worried about this. Students submitting coursework using the likes of ChatGPT. New research has investigated how well the chatbots foil plagiarism detection systems at universities. So in today's multiple choice exam, what is the outcome here from this research? Is it answer A? Yes academics should be worried about this? Or is it answer B? Academics, no need to worry. Well, the answer is very much A I'm afraid. Oh dear. We'll investigate. Also, as Ghislaine has teased ahead, prepare to take your seat in the cinema  that gauges in real time how you're feeling. 

00:02:12 Gareth Mitchell 

Yes, it's all right here on the Somewhere on Earth podcast.  

00:02:24 Gareth Mitchell 

So meet that perfect student, the model student completing assignments in an instant, yet achieving higher marks and outperforming most other students. The downside possibly is maybe they're not as fun to hang out with with the other scholars, and they're not very likely to join any of the sports teams. 

00:02:41 Gareth Mitchell 

But this student, as you probably already guessed. Of course I'm not talking about a real student, I'm talking about chatbots. In fact, this is, and I should put my serious voice on here, because this is genuinely worrying new research. And so what happens when the likes of ChatGPT are used by students to complete their university coursework and exams? 

00:03:00 Gareth Mitchell 

Now I work at university and I know very well, just for myself what a hot issue this, i.e. a student submits their work. How do we know whether it's really their work or if they had a little bit of help, shall we say, from a chatbot? Now, this paper that's just been published in the PLOS ONE journal suggests that such automated course work can slip through many of the detection systems that institutions kind of like to believe are quite effective.  

00:03:30 Peter Scarfe 

Let's hear from one of the authors of this new research, Peter Scarfe of the School of Psychology and Clinical Language Sciences at the University of Reading in England. Welcome to the podcast, Peter. How you doing? 

00:03:35 Peter Scarfe 

Yeah, I'm doing good. Well, thanks. Thanks for the invite. Nice to be here. 

00:03:39 Gareth Mitchell 

Good to have you. So let's set up some of the problem here, because I suppose the groundwork for your research is you're looking at the kind of work that isn't invigilated, so you're not necessarily under exam conditions with somebody sitting at the front looking at what you're doing. This would be what course work, or maybe some of those so-called take home exams that we're talking about here then is it? 

00:03:59 Peter Scarfe 

Yeah, yeah, that's exactly the sort of thing the kind of educational sectors trying to move away from traditional pen and paper exams for very good reasons. 

00:04:10 Peter Scarfe 

And the COVID pandemic obviously kind of accelerated that process, in that overnight we were kind of forced to move from invigilated exams to those which were taken at home. So yeah, you're completely right. The research was very much focused on the problem of AI for both take home exams and coursework. 

00:04:35 Gareth Mitchell 

Tell us about this study then. What were you looking at here, then. You effect, what trying to get something like ChatGPT to try and fool some of these university systems for detecting plagiarism. Just tell me about the research. 

00:04:48 Peter Scarfe 

ChatGPT had really kind of blown up onto the scene in quite a big way a couple of years ago, and there's been some very widely publicised anecdotal reports from educators, you know, across the world, who had entered their assessment questions into ChatGPT or another AI, PressGo, and then graded them and ChatGPT gained excellent answers. 

00:05:14 Peter Scarfe 

And so that is potentially a real problem for the educational sector because if a student can cheat using ChatGPT, go undetected and potentially gain a higher grade by cheating, then that really impacts the way in which we can assess our students. 

00:05:34 Peter Scarfe 

So with this kind of background of these anecdotal reports, you know, we're kind of scientists. We wanted to get some hard data. So we came up with the idea of a project which would be a real life test of our examination systems. So to do that we generated around 60 or so answers to exam questions across five modules of a psychology degree.  

00:06:14 Peter Scarfe 

We generated answers to those questions using ChatGPT and submitted those using fake student accounts which were created for the project and the key thing here was that to the markers, these submissions just look like any other basically. And our key questions were, number one, can we detect those AI written answers which are 100% AI generated and if not, what grades do they achieve? 

00:06:40 Gareth Mitchell 

And tell me how many of these submissions were successfully detected, you know, in other words, they were revealed as having been created by a chatbot. 

00:06:47 Peter Scarfe 

It was unfortunately a very small amount. I believe it was around kind of 94% undetectable basically. 

00:06:57 Gareth Mitchell 

That was the undetected ones. 

00:06:58 Peter Scarfe 

Yes, exactly. 

00:07:11 Ghislaine Boddington 

Right. 

00:07:00 Gareth Mitchell 

So the vast majority got through. And do you mean this would be like a human marker then, you were just relying on on the human being just to yeah, be persuaded that this was from a a real student. 

00:07:12 Peter Scarfe 

Yes, exactly. So we had plagiarism detection software, but at the time, and I think it's still the case actually, universities are not generally using kind of AI detection software and we could discuss why that's the case and the problems with that. So yes, it's very much up to the plagiarism detection software and our markers to be able to detect these AI written submissions. 

00:07:47 Gareth Mitchell 

Yeah. And not only were these submissions getting through, the vast majority undetected, but they actually did pretty well, didn't they? Just give me a sense of how the ChatGPT or the chatbot grades related and compared to real students’ grades. 

00:08:06 Peter Scarfe 

Across all of the modules, on average, the AI outperformed real student submissions by approximately half the classification boundary, so it's quite a substantial outperformance. It's particularly concerning because we generated the AI submissions in kind of the most naive way possible. 

00:08:40 Peter Scarfe 

What we did was we just cut and paste the question into ChatGPT, used a standardized prompt and PressGo, and if there were multiple submissions for a given exam for the same questions, we just used the ChatGPT regenerate button. So in some senses this was the easiest way in which AI could be detected or best, best case scenario. 

00:08:53 Gareth Mitchell 

Right. So the laziest. Yeah. The laziest student literally just typed the question into the chatbot, cut and paste the answer into your assignment and go to the pub. It wasn't even just try and tweak it a little bit, rewrite it, you know, just replace some of the verbs with your own to, you know, none of that. It literally this was the raw ChatGPT stuff got through the examiners. 

00:09:27 Gareth Mitchell 

So apart from sacking all the lecturers. No, not really. I'd love to know what Ghislaine makes of this, then you know because you're interested Ghislaine aren’t you in that issue that Peter raised of some of these software systems that universities are meant to have now to detect this kind of plagiarism. 

00:09:43 Ghislaine Boddington 

Yes, now I was interested when you said that about universities in the main not using AI detection softwares. Maybe you could tell us a little bit more about that and then and then I'd really like to talk to you also about the kind of more companion side of ChatGPT and the future and what your thoughts are on that too. 

00:10:00 Peter Scarfe 

AI detection’s a really difficult process. OpenAI who created ChatGPT released a detection software which they withdrew because it wasn't accurate enough. And there have been numerous companies, both in the educational sector and beyond, who have released AI detection software. But it is generally not recommended to be used in an educational setting to potentially penalize students. Why is that the case? Well, it's really difficult to detect a written text. 

00:10:37 Peter Scarfe 

You know, so for example, let's say I'm marking a piece of work and my magical AI detection software says oh this this piece of work is 53.5% AI written. How do I go about proving that at all? It's pretty impossible. The last thing on Earth we want to do is to be 

00:11:00 Peter Scarfe 

unfairly penalising students who are handing in their own work and it's been incorrectly flagged by this AI software. So it's very different to, say, plagiarism detectors, where the detector can kind of say, OK, this piece of work is copied from these sources and it can then list those sources and then you can go look at those sources and see the copy text. You know, that's very clear indication, whereas with AI that's just simply not possible. 

00:11:26 Ghislaine Boddington 

Right. Yes. Yeah, yeah. 

00:11:32 Ghislaine Boddington 

So in terms of the future, I'm also working at universities and this debate is, you're absolutely right on the table, and Gareth said it too, in every university in the world at the moment, we can see quite a lot happening,  

00:11:47 Ghislaine Boddington 

of debates in forums and in LinkedIn etc about ways to deal with this. But one of the areas that I'm interested in is the AI ethics discussion, is you know, if we look five years ahead in my work, we would consider now that actually, of course, students are going to be working with AI and with ChatGPT.  

00:12:05 Ghislaine Boddington 

And when you said that about the AI detection software say going, well I think 53% of this is AI written. How much does that really matter if we actually are starting to look at the students being companioned in a way by AI, into their, throughout their education, not necessarily just at university. 

00:12:23 Gareth Mitchell 

Yeah, exactly. Because on the course that I teach on, we we were actually very clear with the the students that you know within reason they can use. Now we just get it a bit like, but as an example we have is when calculators were first permitted in mathematics and you know and even in exams in the maths exam. And I'm old enough to remember that being a big discussion at our school about whether we could have calculators. 

00:12:25 Gareth Mitchell 

And it's almost like we're kind of going through that and sorry, Peter, I will bring you in because I'm, you're you're the guest and I want to know your opinion on this, but certainly on our course, we feel increasingly that, yes, our students should be able to demonstrate as good scholars that they can get the best out of these systems as assistance. But obviously still show academic excellence and then prove to their tutors that they've grasped the main concepts from the curriculum. But enough of me. Peter, you're the expert on this. What do you make of it all? 

00:13:10 Peter Scarfe 

I agree completely. I think the educational sector has had these disruptive moments previously, and I also agree with you that, you know, our students are going to be leaving university, going into the workplace, and they're going to be using these systems in their jobs. You know, you've already got AI being integrated into, you know, Microsoft Office. So as an educational establishment, both, you know, universities and schools. 

00:13:42 Peter Scarfe 

we need to be in a place to yes, we're teaching our students how to use these systems ethically, because that's what they'll be using in the workplace. So in in, in some sense, as you know, your example of the calculator,  

00:14:00 Peter Scarfe 

another example is Wolfram Alpha, is like a kind of mathematics solving system. So in some ways I see this just as a new tool. I think it's a very powerful tool and certainly much more powerful I think than a, you know, a pocket calculator. But I think the educational sector 

00:14:18 Peter Scarfe 

as a whole is just going to have to revisit our unsupervised assessments with the knowledge that even if we ask students not to use these systems in their completion, students probably will be using these systems. And that's not to say that, you know, all students are cheaters and stuff like that. That's just showing the reality of the situation. These are new tools. These are powerful new tools, and we need to equip our students for their use. You know, we've always had a mix of assessment types at universities. So some stuff is supervised, some is unsupervised. 

00:14:57 Peter Scarfe 

And I think that's going to continue. You know, certainly I I don't want the interpretation to a paper to be kind of, Oh God we need to move every single assessment to an in-person pen and paper exam. 

00:15:10 Ghislaine Boddington 

Yes. Yeah. We've seen some, I think some very, very interesting potentials for future, you know, human integrated AI, work within universities, etcetera where you know modules are set up where say you can use AI. First of all, we want you to do the AI very clever prompting, get out the best you can from the AI. Then we want you to go back to that and critique what's come out from the AI and add into it from your own sense, then we want you to put the two together and give us original critical thinking around it. So I believe new methodologies will emerge very quickly in this, around actually ways to extend and enhance our brain use of our brains. Yeah, yeah, yeah. 

00:15:53 Gareth Mitchell 

Yeah, it's actually, it's an exciting opportunity. Yeah, in higher education, yeah. And across the whole of education. Well, we're gonna talk more about this in the podcast subscription extra if you'll stay back for us, Peter Scarf, if that's all right. 

00:16:00 Peter Scarfe 

Cool. No, it's fine. 

00:16:02 Gareth Mitchell 

Much appreciated, but thank you in the meantime for your fascinating and important research there. I thoroughly enjoyed reading the paper and and this is such an important conversation. So Peter, thank you. All right then. So away from the lecture theatre. Now  you’re in a different kind of lecture theatre, you’re in a cinema or in an auditorium? You're at the show the action is gripping. It's all very immersive. 

00:16:27 Gareth Mitchell 

But hey, how are you feeling? Well, guess what? The venue knows, it can figure it out, and that's thanks to a new virtual production stage that uses computer generated imagery, CGI to help to create imagined worlds. Now this is all happening at the old Victorian gas works in Bristol in the South West of England. 

00:16:46 Gareth Mitchell 

And Professor Iain Gilchrist is a behavioral and cognitive neuroscientist. He's one of the researchers involved. So you have this virtual production stage, but then also this feedback idea that somehow the systems around the auditorium can work out what the audience are going through. So Iain, welcome to the podcast. 

00:17:05 Iain Gilchrist 

Thank you. Good to be here. 

00:17:06 Gareth Mitchell 

Well, lovely stuff. So it's a 36 seater cinema, isn't it? So just set the scene of the venue that we're talking about here. 

00:17:15 Iain Gilchrist 

We've got this amazing new facility we're building right by the main railway station, so walking distance from the railway station. It is based in this beautiful, historic Victorian building, as you say, which used to be the gas works, providing the gas supply for the whole of the city in the Victorian era. But now we've repurposed it as a new university-based facility, which includes a whole suite of really exciting innovative spaces to do science and engineering and development to support the creative industries in the West Country. 

00:17:47 Iain Gilchrist 

One of the things, potentially the most exciting thing is effectively what will look like a small cinema, so you know it's got an incredibly good projector. It's got a full sound system. You'll sit in a raked seat. It's a proper cinema seat. But actually, it's not just a cinema, it's a search laboratory as well, because we can monitor in detail in a whole variety of ways, I'm sure we'll talk about in a minute, exactly what's happening, moment by moment to those audience members as they watch that come [together]. 

00:18:14 Gareth Mitchell 

Yeah. So you can see how they're engaging with the material. I say see, you can sense you know, so you mentioned you're going to get into some of the things you are monitoring. So what are you monitoring and how are you figuring this out? 

00:18:24 Ghislaine Boddington 

And how, and how are you monitoring them? 

00:18:28 Iain Gilchrist 

How are we doing this? 

00:18:28 Ghislaine Boddington 

Wearable sensors? Is it in the seats? Is it do you? 

00:18:31 Gareth Mitchell 

How the heck? 

00:18:34 Ghislaine Boddington 

Do you put everybody, you know, gear everybody up before they go in. What? How is it happening, yeah. 

00:18:38 Iain Gilchrist 

So there are, I think there are three categories if you like, of ways that we're going to, we have been doing monitoring and we will be doing monitoring more in this space. The first category is wearable devices. So we've used very high spec smart  

00:18:50 Ghislaine Boddington 

Right. Yeah. 

00:18:53 Iain Gilchrist 

watches that people will wear, which will monitor their heart rate and their skin conductance. So when any human being is more roused or excited, you get slight changes in skin conductance because you sweat slightly on the surface of your skin, and we can pick up those changes. It's the same technology that underpins a lie detector test, but used in a slightly different way. 

00:19:13 Gareth Mitchell 

Skin galvanic response. Gosh, I'm clever, right yes? 

00:19:16 Ghislaine Boddington 

Oh yes, right up Gareth’s street. 

00:19:18 Iain Gilchrist 

You're ahead of me. You're ahead of me. 

00:19:22 Gareth Mitchell 

Got it off ChatGPT. It's fine. 

00:19:25 Gareth Mitchell 

It wrote my questions. It's brilliant. 

00:19:25 Iain Gilchrist 

So we can also use those kinds of sensors to monitor how much people fidget and move around in the seat. And. And we've also already  

00:19:34 Ghislaine Boddington 

Right, yes. 

00:19:05 Iain Gilchrist 

just published a paper where we've looked at how violently and for how long people clap at the end of a performance. So that was live theater, for instance. So that's our first class wearable devices. The second class is probably the more conventional way that we've been doing this for the last 50 years, which is to issue questionnaires at the end of a performance, at the end of an experience. 

00:19:56 Iain Gilchrist 

We give people questionnaires and we also run focus groups to ask them in depth about what they experienced and then the third class of ways is to use remote camera technology. So we can have cameras at the front of our cinema or in the auditorium or in a concert venue. 

00:20:11 Iain Gilchrist 

And we can record how the audience is responding, and we use different types of camera to get different kinds of signals. So we've got thermal cameras. We know that when people change their emotion, there are slight changes in the distribution of heat across their face. Believe it or not. And we can pick up those signals to see 

00:20:33 Gareth Mitchell 

My god I didn't know that. 

00:20:35 Iain Gilchrist 

moment by moment changes in people's response. We can also, of course, use those cameras to tell if people smile or frown, or grimace or yawn or turn away. We've got visible spectrum cameras, and we also got our camera array, which is in infrared, because of course, if you're in the cinema or in the theater, typically the lights go down and a visible spectrum camera isn’t much used to you at all at under those circumstances. 

00:20:54 Gareth Mitchell 

Right. So hence the. Yeah, sorry. Infrared. Yeah. Sorry. Are you listening, even more technology or is that about it? 

00:21:02 Iain Gilchrist 

Yeah, I was just gonna say that some of those, some of those, some of those technology allows us to do really nice and interesting things. So for instance, we can use the camera technologies to look at blink rate. So we know that when people are paying attention and focusing their energies on something, immersed as you described a minute ago, the blink rate changes, so we blink less. When we’re more distracted and less engaged we blink slightly more. We can also look to see whether people turn away because we can monitor  the pose of their head and we can see at what point do people eat their popcorn? Do people stop and and make that little side comment? You know, when you're in a film and you're really not enjoying, what you do quite often, when the immersive experience has been broken, you turn to the person beside you go, gosh this is rubbish, 

00:21:47 Iain Gilchrist 

I’m going for a pint instead. And we can capture those moments where people are no longer immersed in the content. So the film in front of them. But instead of beginning to become aware of the space around them, maybe the person they came in with, maybe the stranger who's being very noisy with their crisp packet. And so all that kind of fidgeting behavior we can monitor and get a handle on. 

00:22:07 Ghislaine Boddington 

Fascinating. 

00:22:07 Gareth Mitchell 

A huge number of data points. Yeah, in real time. 

00:22:13 Ghislaine Boddington 

 So most data points and and you're focusing on like in a way behavioral analytics around that, those multiple data points that are coming into you so. 

00:22:22 Gareth Mitchell 

Yeah. But like, why? Why? Why? Why do that? Why does it matter? 

00:21:49 Ghislaine Boddington 

It matters. 

00:22:25 Iain Gilchrist 

Why do that. So I think this new technology, particularly the wearable device and remote cameras, has allowed us really for the first time to, in a non-invasive way, measure moment by moment when people are responding. That's what we've unlocked this ability without tapping them on the shoulder and saying, you enjoying it? 

00:22:43 Iain Gilchrist 

And then waiting two minutes and tapping them on the shoulder again and saying are you still enjoying it. We can avoid that because we can do these passive measurements, which won't interrupt people's experience, but will give us a moment by moment record of how they're responding. But why is that important? Well, we know that 

00:23:01 Iain Gilchrist 

there are a very large number of films made each year which lose a huge amount of money and there are there are other films which make huge amounts of money and have a a deep and fundamental impact on what people think about, how they feel about a particular topic. They learn new things. And if we can give film makers a set of tools which allows them to say,  

00:23:29 Iain Gilchrist 

well, we might be losing the audience at this point. This is the point at which the audience is less interested, less engaged. It's the point where everybody turns to the person sitting beside them and is a bit distracted or this is the point when people realize that actually there was a wet day on the way into the cinema. My feet are wet. I'm starting to fidget because my feet are wet. Because I got wet on the way. If we can spot those moments, then the film makers can go back and use that as a tool to re-edit the content. 

00:23:55 Iain Gilchrist 

And maybe take out the points where the edit wasn't quite successful. So we're not going to dictate to film makers what they do. That's part of the creative process, but we're giving them an additional set of tools so that they can begin to think again about moments in the film or content which may be slightly less successful. 

00:24:14 Gareth Mitchell 

Yeah, but I mean, let me be the film critic here then. I'll be very cynical. And wouldn't I just say, but really audience members, they know they're coming into a lab. Fair enough. But do they really want to be surveilled just so it helps researchers help the film companies to make more money out of us, which is basically what this comes down to. That's a cynical way of asking the question. Or is the nice way of asking the question like, hey, this is a real opportunity to work out what really worked with audiences so that the people paying the good money going to cinema can really enjoy the experience even more thanks to this brilliant research, so I just wonder which side of that argument you might be on. 

00:24:52 Iain Gilchrist 

So so that what I've describe is a process that happens, you know, in that example, a process that happens before the release of the content. So the audience members who  come in and and become monitored, typically in our experiments are paid  

00:25:208 Iain Gilchrist 

to be part of those experiments. They come in, they know they're gonna be monitored, and they're signed. you know, it's been ethically approved that that happens and they've signed a whole set of waivers to explain exactly what's going to happen to their data, how we use it, how we're gonna anonymise it and all those kinds of things.  

00:25:22 Ghislaine Boddington 

Yeah, that's good. 

00:25:24 Iain Gilchrist 

What we're not talking about is the deployment of this technology in every Odeon Multiplex around the country because by then it's not very useful to you because you've already released the film. The content is out there. Now I I I said earlier that it's important that that we're not interested in interfering with the creative process and there are film makers who  

00:25:48 Iain Gilchrist 

actually don't really care about the audience. They're making something which is extremely personal to them. They've got something that they want to say, and they may be disinterested in some ways. And and we're not, we're not interested in in interfering with that creative process. But there are other film makers who genuinely want to make content where they change how people think and feel and behave beyond the content. 

00:26:13 Iain Gilchrist 

So if you're making a documentary about climate change, you really want people engaged, you want people engaged in that content, and you want that to be something they're left thinking about afterwards. And maybe, and that's the hardest thing as we know, as  behavioural scientists and psychologists, maybe changing their behavior in the long term. Now for those content creators, understanding the impact on the audience is absolutely critical. 

00:26:41 Ghislaine Boddington 

I'm really interested in this because I work in immersion experiences which I know you are also focusing on at this amazing new venue in in Bristol. In immersion, digital immersion experiences, whether it's AR, VR or just, you know, generative largescale installations,  

00:26:58 Ghislaine Boddington 

these behavioral, uhm, aspects are really important, because it's actually about the navigation of actually how you're moving people around, their reaction in various points of time and space, whether they actually do use that corner or get the fact they're meant to pull down that vertical switch or actually do come together in a way which is actually positive. 

00:27:21 Ghislaine Boddington 

So I can really understand it in that and I know that the r/AR does give you a lot of points back already data points back directly just from those particular technologies. So I can see film makers making some shifts. But I think in immersion work, this is absolutely imperatively part of, because it's an interaction between the audience, the user, really participant, and the immersion space. And we're really interested in what you said about clapping because having been in a live performance sector to all my life, 

00:27:50 Ghislaine Boddington 

I can listen to clapping and I know I tell you what it is. It's not just the clapping, it's actually the length of time, nanoseconds, between the end of a show and when the clap starts, that really matters whether you've got the audience like breathing in and then going yes. And clapping like mad or whatever they're just like, clap, clap, clap straight away at the beginning and you just get, you have this real intuitive sense starts to occur, which is linked to just a few tiny seconds at the end of a show. Fascinating. Yeah. 

00:28:17 Gareth Mitchell 

Yeah, and this is also, and of course you're able now to to have the empirical side of that as well. Ghislaine says she has that gut feeling, any performer does. You know, you just you read the audience definitely. 

00:27:25 Ghislaine Boddington 

Yes, yes, yes. The more you're in it, yeah. 

00:28:28 Gareth Mitchell 

And now you're coming in with your analytics and and you can bring those two together and actually do some, and yeah, final question I’m putting it to you, as well as all this stuff about, you know, helping film makers and, you know, create incredible experiences. But just do some darn good science here as well. It's fascinating. 

00:28:459 Ghislaine Boddington 

Yeah, exactly. Yeah, the quantifiable and the quantitative coming together. 

00:28:46 Iain Gilchrist 

For me, there is a really fundamental scientific question in here, and the interesting thing about human beings is they're easily distracted. If you're doing your tax return or you've been asked to write a report, or you've got to do your finances, 

00:29:00 Iain Gilchrist 

it's incredibly difficult to stay focused on that and you feel psychologically, you feel the effort of staying focused. However, what's amazing is we create as a society, culturally, we create these artifacts like film and television, and increasingly, VR immersive experiences,  AR immersive experiences 

00:29:22 Iain Gilchrist 

where we can remain focused for hours and that's incredible. It's incredible that we consistently, you know, for a very long time historically have made these things where they capture our attention, we become immersed in them and we stay focused in them for hours. 

00:29:43 Iain Gilchrist 

You know, and if you get caught up in a box set sometimes you forget to eat and drink. Certainly I forget to feed my kids and I think it's. I think it's that's really fascinating because because in the traditional, and Peter will know this,  

00:29:48 Gareth Mitchell 

Time has no meaning. Exactly, there's no meaning. 

00:29:56 Iain Gilchrist 

In the traditional psychology literature we think of sustained attention as being effortful. But sustained attention when you're in the theatre or you're in VR or watching a film it’s not effortful, it's it's poetic. It's, you know, it's a flow thing that goes on and at the end of it, the very end, 

00:30:31 Gareth Mitchell 

Yeah, in fact it’s hard not to pay attention. 

00:30:17 Iain Gilchrist 

particularly whether in the live performance you do have that moment where everybody's holding their breath. 

00:30:22 Ghislaine Boddington 

Yes. Yeah. 

00:30:22 Iain Gilchrist 

And then the the immersion, that sense of being there breaks and then you get the applause as you described. And I think that's really interesting and there's something fundamental scientifically to understand that. 

00:30:30 Ghislaine Boddington 

Yeah. 

00:30:32 Gareth Mitchell 

Incredible.  

00:30:34 Ghislaine Boddington 

And it's all about live-ness. And that's very, very important in today's world. The live response, the on the spot  

00:30:38 Gareth Mitchell 

It's, uh, can't beat it. 

00:30:41 Ghislaine Boddington 

real time live presence that there is yeah. Fantastic. 

00:30: 43 Gareth Mitchell 

Yeah, I know. And and I'm I'm thinking a discussion for another day, but I know you know Peter is still on the call from earlier, but coming back to universities, I think lecturers would love to, will probably hate in some ways to know, you know, how the students might be responding during a lecture for instance. So yeah, but I'm on that fascinating thought. I am going to hold next week we've, I'm getting told off in my headphones for overrunning. 

00:31:03 Gareth Mitchell 

But it's the producer’s fault Ania for booking such fascinating guests because we just go on and on. If only we had the podcast extra. Oh, we do. Lovely. So let's keep the conversation going in that. But you, dear listener, if you can't join us from the subscription, see you next time, and I'll just give a quick shout out for the socials because I think there's a lot in this programme that you will want to talk about. 

00:31:25 Gareth Mitchell 

So you know whether you are a student, for instance, at university and you have a view about using the likes of ChatGPT as an assistant or you're sick and tired of people casting aspersions on your excellent work by suspecting somehow that it was not fully your work. You may have a view. 

00:31:42 Gareth Mitchell 

And of course, if you're the kind of person who goes out and goes to the cinema and watches films and can identify with everything that we've just heard there from Iain, just what do you reckon, you know, what do you make of this research? So it sounds like I'm setting up a phone in here, but hardly. But I'm giving you an e-mail address. It's hello at somewhere on earth.co, hello somewhere on earth.co and on WhatsApp we are code 447486329484 and just search for Somewhere on Earth podcast S.O.E.P and things like that on the socials, but I'm sure you're all very good at just, you're listening to a tech podcast so you can figure out how to use social media and get in touch with us, I'm sure. 

00:32:21 Gareth Mitchell 

Thank you very much to the excellent Dylan for making us sound nice today and to our production manager, Liz, for making sure this basically works and stuff gets done, to producer and editor Ania Lichtarowicz for booking brilliant guests. That just flew by. And you Ghislaine as well. Marvelous. So many thank yous. Thank you.