WEBVTT 00:00:00.430 --> 00:00:02.890 In this episode, I'll be sharing an interview 00:00:02.890 --> 00:00:06.209 I recently conducted with Bailey Tudnam, an alumnus 00:00:06.209 --> 00:00:08.369 at the school where I teach, who is exploring 00:00:08.369 --> 00:00:11.810 ways in which he augments his workflow while 00:00:11.810 --> 00:00:14.330 designing products which will be deployed off 00:00:14.330 --> 00:00:17.730 -world. We chat about the reasons why he is being 00:00:17.730 --> 00:00:21.660 cautious in his exploration of AI. and why he 00:00:21.660 --> 00:00:24.539 thinks schools and universities must now adapt 00:00:24.539 --> 00:00:28.140 and incorporate AI literacy into their course 00:00:28.140 --> 00:00:31.260 design and delivery. It was great to catch up 00:00:31.260 --> 00:00:33.600 with Bailey, and I hope you find some useful 00:00:33.600 --> 00:00:37.560 takeaways from our conversation. The AI Academia 00:00:37.560 --> 00:00:41.000 podcast is a weekly show helping educators like 00:00:41.000 --> 00:00:44.780 you leverage AI in your everyday practice. I'm 00:00:44.780 --> 00:00:47.159 your host, Andy Fisher, and thanks for joining 00:00:47.159 --> 00:00:51.920 me. Let's dive straight into the interview. Bailey, 00:00:52.020 --> 00:00:55.299 welcome to the AIcademia podcast. Thank you. 00:00:55.340 --> 00:00:59.899 Thanks for having me. So it's the first guest 00:00:59.899 --> 00:01:02.039 interview of the podcast and it's something I've 00:01:02.039 --> 00:01:06.079 wanted to do for a while. But when we had a chance 00:01:06.079 --> 00:01:08.840 encounter on LinkedIn a couple of weeks ago and 00:01:08.840 --> 00:01:11.719 you told me what you're up to, I thought this 00:01:11.719 --> 00:01:14.640 is too good a conversation not to have live with 00:01:14.640 --> 00:01:18.700 everybody. So for the listeners, you attended 00:01:18.700 --> 00:01:22.060 the school where I teach. and have been on a 00:01:22.060 --> 00:01:25.140 bit of a wild journey since then. So why don't 00:01:25.140 --> 00:01:28.340 you tell us what happened once you left Norwich 00:01:28.340 --> 00:01:31.739 School, the close and the cathedral? Yeah, a 00:01:31.739 --> 00:01:33.939 lot has happened since then. So that was what 00:01:33.939 --> 00:01:36.180 must have been 2018, I think, when I left Norwich 00:01:36.180 --> 00:01:39.159 School. So after that year, I took a year out. 00:01:39.599 --> 00:01:41.760 So I had an injury for that year out. So I took 00:01:41.760 --> 00:01:45.540 a year out before going to university. So I spent 00:01:45.540 --> 00:01:49.739 that year working on a televised talent show 00:01:49.739 --> 00:01:53.040 series. which was very random. I didn't know 00:01:53.040 --> 00:01:55.819 that. Yeah, that was very strange. It was good 00:01:55.819 --> 00:01:58.640 fun. So I was the producer and the head judge, 00:01:58.879 --> 00:02:01.980 if you can believe it, on this Televotes talent 00:02:01.980 --> 00:02:06.859 show series in 2019. And then I came up to Glasgow. 00:02:07.099 --> 00:02:09.560 So I've been in Glasgow ever since. I started 00:02:09.560 --> 00:02:11.659 at University of Glasgow and Glasgow School of 00:02:11.659 --> 00:02:16.979 Art in 2019. And I studied a degree called product 00:02:16.979 --> 00:02:19.729 design engineering. so the way i kind of describe 00:02:19.729 --> 00:02:22.229 that to people is it's basically mechanical engineering 00:02:22.229 --> 00:02:24.889 with product design so we kind of do all the 00:02:24.889 --> 00:02:26.770 mechanical engineering at university of glasgow 00:02:26.770 --> 00:02:29.389 product design at glasgow school of art and it's 00:02:29.389 --> 00:02:32.349 kind of a fusion of those two and it's a fantastic 00:02:32.349 --> 00:02:35.930 degree and it really teaches you about not just 00:02:35.930 --> 00:02:39.569 how to design products but how they work and 00:02:39.569 --> 00:02:42.840 i'm very much of the opinion that you can't truly 00:02:42.840 --> 00:02:44.860 design a product unless you understand how it 00:02:44.860 --> 00:02:47.819 works i was about to ask that because often designers 00:02:47.819 --> 00:02:50.759 might not think i i don't know about either world 00:02:50.759 --> 00:02:53.340 but i would imagine if you design without understanding 00:02:53.340 --> 00:02:55.520 the engineering behind it you might create something 00:02:55.520 --> 00:02:58.740 that is a real problem a hundred percent and 00:02:58.740 --> 00:03:01.659 and you know it it's also almost cutting out 00:03:01.659 --> 00:03:03.900 the middleman a little bit you know and i'll 00:03:03.900 --> 00:03:07.439 get into the job that i'm on now um but you know 00:03:07.439 --> 00:03:10.150 in the past the sort of job that i'm in now they 00:03:10.150 --> 00:03:12.229 probably would have had a mechanical engineer 00:03:12.229 --> 00:03:14.590 and they would have had a product designer and 00:03:14.590 --> 00:03:16.090 they would have you know collaborated together 00:03:16.090 --> 00:03:19.210 and and the degree that i've done has allowed 00:03:19.210 --> 00:03:22.669 me to kind of do those two things at once and 00:03:22.669 --> 00:03:24.669 i think it's it's a lot more efficient in terms 00:03:24.669 --> 00:03:26.930 of the process when it comes to design and engineering 00:03:26.930 --> 00:03:32.750 um so yeah it's a fantastic degree um and yes 00:03:32.750 --> 00:03:36.729 i was there for five years um and now i'm working 00:03:36.729 --> 00:03:41.009 for a startup in Edinburgh, which is what I dropped 00:03:41.009 --> 00:03:43.129 you a message about on LinkedIn a few days ago. 00:03:43.710 --> 00:03:47.969 So that startup is designing autonomous kitchen 00:03:47.969 --> 00:03:51.689 modules, which is eventually, hopefully going 00:03:51.689 --> 00:03:55.969 to be used in lunar and Mars habitats, which 00:03:55.969 --> 00:03:59.370 is mad and not something that I expected. I did 00:03:59.370 --> 00:04:01.050 mention it to several of my colleagues at the 00:04:01.050 --> 00:04:04.129 school who remember you fondly, but primarily 00:04:04.129 --> 00:04:08.189 as a musician and as an entertainer. So when 00:04:08.189 --> 00:04:10.009 I said, oh, yeah, now he's building kitchens 00:04:10.009 --> 00:04:13.849 for the moon, they thought I was joking. I mean, 00:04:13.909 --> 00:04:18.290 this is a lovely conversation starter. You know, 00:04:18.310 --> 00:04:20.970 I go to different events and stuff every now 00:04:20.970 --> 00:04:22.089 and then, and people are always like, what do 00:04:22.089 --> 00:04:25.720 you do for work? kitchens for the moon and they 00:04:25.720 --> 00:04:29.060 think i'm joking and i'm not um and it's like 00:04:29.060 --> 00:04:30.860 i say it's not what i expected to be doing but 00:04:30.860 --> 00:04:35.399 it's a very interesting company um and it's fantastic 00:04:35.399 --> 00:04:38.319 you know it's a very small startup there's i 00:04:38.319 --> 00:04:42.180 think six of us in total um which is great for 00:04:42.180 --> 00:04:43.899 me coming out of uni because it's forcing me 00:04:43.899 --> 00:04:46.490 to learn so many different things and we're working 00:04:46.490 --> 00:04:48.589 with obviously major clients as you can imagine 00:04:48.589 --> 00:04:51.610 um so now i'm doing everything from the design 00:04:51.610 --> 00:04:53.829 side of things engineering business side funding 00:04:53.829 --> 00:04:57.509 applications um so yeah it's absolutely fantastic 00:04:57.509 --> 00:05:01.430 and it's a really interesting company and it's 00:05:01.430 --> 00:05:03.709 you know it's it's something that i never thought 00:05:03.709 --> 00:05:05.569 i'd be doing it's something until i joined this 00:05:05.569 --> 00:05:08.930 company that i thought was miles off You know, 00:05:08.949 --> 00:05:10.649 and I start and I'm meeting all these people 00:05:10.649 --> 00:05:12.490 and we're meeting these investors and these companies. 00:05:12.509 --> 00:05:14.790 And you realize that, you know, lunar habitats 00:05:14.790 --> 00:05:18.350 is expected to be around 10 years away. You know, 00:05:18.370 --> 00:05:19.850 and this is something that you think would be 00:05:19.850 --> 00:05:22.410 a lot further. Yeah, that was going to be one 00:05:22.410 --> 00:05:24.769 of my next questions was, what's the timeline? 00:05:24.870 --> 00:05:29.329 So 10 years to have a permanent or semi -permanent 00:05:29.329 --> 00:05:32.509 structure? So 10 years is expected to be permanent. 00:05:32.850 --> 00:05:35.730 And I've just been down, literally two days ago, 00:05:35.769 --> 00:05:39.110 I was down at the massive Spacecom Expo in London. 00:05:39.709 --> 00:05:42.250 And it's insane the amount of companies that 00:05:42.250 --> 00:05:44.370 are forming in this sector. You know, it's the 00:05:44.370 --> 00:05:45.990 fastest growing sector in the world now, the 00:05:45.990 --> 00:05:48.810 space industry, especially in Scotland is huge. 00:05:49.170 --> 00:05:50.850 And you're kind of going around and seeing all 00:05:50.850 --> 00:05:52.610 these startups and all the different things that 00:05:52.610 --> 00:05:54.370 you would never expect and things that you thought 00:05:54.370 --> 00:05:56.649 would be miles off. And then you've got, we had 00:05:56.649 --> 00:05:59.569 a keynote from Tim Peake there, who was... talking 00:05:59.569 --> 00:06:02.269 about you know the next sort of 10 to 20 to 30 00:06:02.269 --> 00:06:04.730 years and what we're expecting to see and and 00:06:04.730 --> 00:06:07.490 you know it's almost like space race 2 .0 you 00:06:07.490 --> 00:06:11.410 know it's this race to kind of habit the moon 00:06:11.410 --> 00:06:14.509 and and to um almost use that as a stepping stone 00:06:14.509 --> 00:06:16.569 to mars then so that's what i was going to ask 00:06:16.569 --> 00:06:19.629 is this a precursor to populating mars if it 00:06:19.629 --> 00:06:21.889 works on the moon then it should work i think 00:06:21.889 --> 00:06:26.019 that's the plan yeah i think companies or institutions 00:06:26.019 --> 00:06:29.500 just to say like like nasa for example are reluctant 00:06:29.500 --> 00:06:32.199 to call it that they're reluctant to treat it 00:06:32.199 --> 00:06:34.519 as a stepping stone because it's an it's an amazing 00:06:34.519 --> 00:06:37.120 thing in itself yeah i think to call it a stepping 00:06:37.120 --> 00:06:39.519 stone they are reluctant to say that but ultimately 00:06:39.519 --> 00:06:42.079 that is that is what it is the end goal is mars 00:06:42.079 --> 00:06:46.279 and you know if we can get to the moon the risk 00:06:46.279 --> 00:06:49.939 and the cost and the ease of then making that 00:06:49.939 --> 00:06:54.480 transition to mars greatly reduces um and so 00:06:54.480 --> 00:06:57.899 that that's the aim there so and obviously this 00:06:57.899 --> 00:07:01.079 the podcast is primarily focused upon ai tools 00:07:01.079 --> 00:07:04.560 so yes did you have to suddenly suddenly start 00:07:04.560 --> 00:07:08.379 upskilling in the use of ai as you came into 00:07:08.379 --> 00:07:10.779 working with the startup or was it in some ways 00:07:10.779 --> 00:07:14.529 integrated into your degree course Yeah, so it 00:07:14.529 --> 00:07:17.990 was never integrated into the degree. And I think 00:07:17.990 --> 00:07:20.110 it was one of those things that really kind of 00:07:20.110 --> 00:07:23.589 became a lot bigger right towards the end. And 00:07:23.589 --> 00:07:26.250 one of the other things I'm doing, which I'll 00:07:26.250 --> 00:07:28.949 get on to probably later on, is I'm also helping 00:07:28.949 --> 00:07:31.470 tutor second year students in the degree that 00:07:31.470 --> 00:07:33.209 I've just graduated from now. And it's something 00:07:33.209 --> 00:07:36.629 that I'm seeing a lot more now from the tutor 00:07:36.629 --> 00:07:38.689 side of things than I was from the student side 00:07:38.689 --> 00:07:40.269 of things. It's something that is becoming a 00:07:40.269 --> 00:07:45.220 part of. of every day. So I kind of learned a 00:07:45.220 --> 00:07:47.500 lot more of it when I started at this company. 00:07:47.660 --> 00:07:50.819 And my boss is very much under the impression 00:07:50.819 --> 00:07:55.819 of, you know, he's very pro -AI in what we do 00:07:55.819 --> 00:07:58.019 and he's very pro -efficiency. And he's like, 00:07:58.040 --> 00:08:00.759 if you can use AI to make your process more efficient, 00:08:00.860 --> 00:08:02.759 then use it. He's like, it's silly not to use 00:08:02.759 --> 00:08:06.379 it, but to use it as a tool. I would say in terms 00:08:06.379 --> 00:08:09.939 of what I do at work, I don't use it as much 00:08:09.939 --> 00:08:12.759 as other people in the company do. And I guess 00:08:12.759 --> 00:08:14.860 it's because what I'm doing tends to be a lot 00:08:14.860 --> 00:08:19.199 more about my sort of imagination and my innovation 00:08:19.199 --> 00:08:21.759 when it comes to product design. There's a couple 00:08:21.759 --> 00:08:23.319 of reasons why I don't use it, but that's one 00:08:23.319 --> 00:08:25.879 of the reasons. Whereas we have people like software 00:08:25.879 --> 00:08:29.139 developers, for example, who are using AI every 00:08:29.139 --> 00:08:31.779 single day. And they're using it to not necessarily 00:08:31.779 --> 00:08:34.379 write code, but to debug, to find issues and 00:08:34.379 --> 00:08:37.789 those sorts of things. I think the other... concern 00:08:37.789 --> 00:08:40.250 that i have with using it in certain aspects 00:08:40.250 --> 00:08:43.889 is the confidentiality of the work that we're 00:08:43.889 --> 00:08:45.929 doing and you know the contracts that we have 00:08:45.929 --> 00:08:49.470 and those sorts of things if i am using something 00:08:49.470 --> 00:08:52.789 like chat gpt to help me generate ideas and solutions 00:08:52.789 --> 00:08:55.929 to a problem that i've got i don't know where 00:08:55.929 --> 00:08:58.509 that information is then going at the moment 00:08:58.509 --> 00:09:00.750 you know it is a very sort of wishy -washy world 00:09:00.750 --> 00:09:03.169 of where that's going and who that's ending up 00:09:03.169 --> 00:09:07.419 with um So, yeah, I'm almost reluctant to use 00:09:07.419 --> 00:09:10.379 it in those kind of aspects of things. What I 00:09:10.379 --> 00:09:13.960 would say I do utilize it for at work is in research. 00:09:15.200 --> 00:09:17.379 And, you know, I'm very much under the impression 00:09:17.379 --> 00:09:21.320 in general of that AI should be there to aid, 00:09:21.399 --> 00:09:25.279 not to replace. And so even when I'm using it 00:09:25.279 --> 00:09:28.500 for research, that doesn't just replace my research. 00:09:28.559 --> 00:09:31.299 It's there to aid in what I do. And I am very... 00:09:31.549 --> 00:09:34.950 pro that i think it's no different using gpt 00:09:34.950 --> 00:09:37.269 to find information than it is to using google 00:09:37.269 --> 00:09:39.409 it just cuts out all of the rubbish that i'm 00:09:39.409 --> 00:09:41.870 not looking for you know when you search something 00:09:41.870 --> 00:09:43.830 on google you get all these tags of you know 00:09:43.830 --> 00:09:45.710 everything has tags of background stuff and it's 00:09:45.710 --> 00:09:48.250 nothing half of it's irrelevant whereas chat 00:09:48.250 --> 00:09:50.289 gpt will find me exactly what i'm looking for 00:09:50.289 --> 00:09:53.230 but i won't then just take that for gospel you 00:09:53.230 --> 00:09:56.330 know and is chat gpt the primary platform you're 00:09:56.330 --> 00:09:59.049 using or are you also using claude or perplexity 00:09:59.049 --> 00:10:03.129 or other so Me personally, I'm mainly using ChatGPT 00:10:03.129 --> 00:10:06.049 and the company has the advanced versions of 00:10:06.049 --> 00:10:08.730 ChatGPT and whatever. So that is what we would 00:10:08.730 --> 00:10:13.429 primarily use. Sometimes I'll use Copilot if 00:10:13.429 --> 00:10:15.850 I'm generating imagery. Obviously, it's still 00:10:15.850 --> 00:10:19.850 based on the same AI, but for some reason, Copilot 00:10:19.850 --> 00:10:23.269 tends to give me better imagery. It gives you 00:10:23.269 --> 00:10:28.059 more options. But primarily I'm using ChatGPT 00:10:28.059 --> 00:10:30.679 if I'm doing research. Yeah, and so you're using 00:10:30.679 --> 00:10:33.440 the, I think they refer to it as the deep search 00:10:33.440 --> 00:10:36.039 or deep research. There are so many names that 00:10:36.039 --> 00:10:37.639 are similar that I kind of lost track at this 00:10:37.639 --> 00:10:41.539 stage. So you're basically using it for people 00:10:41.539 --> 00:10:44.440 who aren't familiar with that tool. It allows 00:10:44.440 --> 00:10:48.019 you to agentically reach out, look at multiple 00:10:48.019 --> 00:10:51.570 online sources, and then curate them. and bring 00:10:51.570 --> 00:10:53.490 the information back, and you can mine down into 00:10:53.490 --> 00:10:56.190 the individual sources if you need to. Exactly, 00:10:56.350 --> 00:10:58.330 100%. Yeah, but it's very interesting what you 00:10:58.330 --> 00:11:00.809 were saying about the concern about proprietary 00:11:00.809 --> 00:11:03.509 information and the way in which that information 00:11:03.509 --> 00:11:06.750 is then used, particularly in research and cutting 00:11:06.750 --> 00:11:09.210 edge, because if, for example, someone had access 00:11:09.210 --> 00:11:12.529 to your search queries, they could, in theory, 00:11:12.690 --> 00:11:15.970 take that information, run with it. for the sake 00:11:15.970 --> 00:11:18.009 of argument's sake, use the deep seek model as 00:11:18.009 --> 00:11:21.169 a Chinese model, it could then end up in the 00:11:21.169 --> 00:11:23.509 hands of the CCP and they could use that information 00:11:23.509 --> 00:11:26.809 and then jump ahead of the work that you're doing. 00:11:27.029 --> 00:11:30.429 So that is an issue. And I can imagine that that's 00:11:30.429 --> 00:11:33.470 quite a roadblock for you, therefore, in terms 00:11:33.470 --> 00:11:36.690 of how far you can use AI. Has the company looked 00:11:36.690 --> 00:11:39.210 at the possibility of locally hosting a large 00:11:39.210 --> 00:11:42.370 language model so that the data never leaves? 00:11:44.299 --> 00:11:47.259 your offices i don't think so not that i know 00:11:47.259 --> 00:11:50.279 of but um obviously in terms of what we're doing 00:11:50.279 --> 00:11:52.240 like i say confidential contracts and things 00:11:52.240 --> 00:11:54.940 that is definitely an important step forward 00:11:54.940 --> 00:11:58.100 if if we're wanting to utilize ai for more things 00:11:58.100 --> 00:12:00.100 um but i don't think it's something that has 00:12:00.100 --> 00:12:02.419 been considered to this point yeah i think again 00:12:02.419 --> 00:12:04.759 it's still really early days um and i think that 00:12:04.759 --> 00:12:06.639 it's only at this point that the price point 00:12:06.639 --> 00:12:09.929 is coming down Because you would need an array 00:12:09.929 --> 00:12:12.169 of servers to be able to run one of the larger, 00:12:12.169 --> 00:12:14.730 large language models. So, again, for the benefit 00:12:14.730 --> 00:12:17.309 of the listeners, if you went with something 00:12:17.309 --> 00:12:22.350 like one of the Lama high -functioning systems, 00:12:22.529 --> 00:12:26.669 you're talking about 4 billion parameters, or 00:12:26.669 --> 00:12:30.570 400 billion parameters, rather. So even with 00:12:30.570 --> 00:12:32.090 the best -spec computer, you're not going to 00:12:32.090 --> 00:12:35.129 run that. But what they are now starting to release 00:12:35.129 --> 00:12:38.600 is... home -based supercomputers so nvidia's 00:12:38.600 --> 00:12:40.379 coming out with something in a couple of months 00:12:40.379 --> 00:12:42.639 that will be the equivalent of a thousand laptops 00:12:42.639 --> 00:12:45.879 in terms of its processing and it's tiny it's 00:12:45.879 --> 00:12:49.220 like bread box sized yeah and you can you you 00:12:49.220 --> 00:12:51.179 would be able to run any large language model 00:12:51.179 --> 00:12:53.320 as far as i'm aware on the market now locally 00:12:53.320 --> 00:12:56.159 and um there's there's no limitation on tokens 00:12:56.159 --> 00:12:57.720 as long as you've got the electricity to run 00:12:57.720 --> 00:13:02.330 it will work So that's quite exciting. I imagine 00:13:02.330 --> 00:13:04.970 that's the direction that companies with sensitive 00:13:04.970 --> 00:13:08.070 information will go. And schools as well, I think, 00:13:08.090 --> 00:13:10.490 because obviously the data we hold on students 00:13:10.490 --> 00:13:14.389 is confidential and important to keep safe. Yeah. 00:13:15.190 --> 00:13:18.049 So having said that, do you want to segue to 00:13:18.049 --> 00:13:20.129 talk a little bit about your work as a mentor 00:13:20.129 --> 00:13:25.110 to students running on your course and what you're 00:13:25.110 --> 00:13:27.090 running into, good and for bad, in terms of the 00:13:27.090 --> 00:13:31.519 way they're using AI? So I think, like I say, 00:13:31.600 --> 00:13:35.299 for me, when I was kind of, I'm kind of in a 00:13:35.299 --> 00:13:37.620 unique position where I've just finished education 00:13:37.620 --> 00:13:39.820 and now sort of working on it from the other 00:13:39.820 --> 00:13:42.419 side. So when I was finishing, it wasn't really 00:13:42.419 --> 00:13:46.000 something that... any of us were using um and 00:13:46.000 --> 00:13:48.340 you know this was only last year you know it 00:13:48.340 --> 00:13:51.559 just shows how rapidly this this area is changing 00:13:51.559 --> 00:13:53.960 it wasn't really something we were using um it 00:13:53.960 --> 00:13:57.899 kind of became really apparent towards the end 00:13:57.899 --> 00:13:59.980 of my degree when we were writing things like 00:13:59.980 --> 00:14:02.279 our final technical reports which is the equivalent 00:14:02.279 --> 00:14:04.860 of like a dissertation for our course and that 00:14:04.860 --> 00:14:08.000 sort of thing so i was kind of i did use it during 00:14:08.000 --> 00:14:10.159 that and i was again it was utilizing it for 00:14:10.159 --> 00:14:13.399 research so for like building my bibliography 00:14:13.399 --> 00:14:16.159 for my references for my report it was never 00:14:16.159 --> 00:14:19.759 for writing the report itself um you know in 00:14:19.759 --> 00:14:21.600 my opinion especially for it was a technical 00:14:21.600 --> 00:14:23.720 report but it was a creative discipline it was 00:14:23.720 --> 00:14:26.860 a creative report and i think you you lose something 00:14:26.860 --> 00:14:29.620 when you're using artificial intelligence to 00:14:29.620 --> 00:14:33.000 write a creative document for you you know this 00:14:33.559 --> 00:14:36.259 it's very unique the way that individuals in 00:14:36.259 --> 00:14:39.240 the creative world see different things and see 00:14:39.240 --> 00:14:41.159 different problems and tackle things and i think 00:14:41.159 --> 00:14:42.759 that really comes across in the way that you 00:14:42.759 --> 00:14:45.620 write stuff um and this is kind of the point 00:14:45.620 --> 00:14:47.299 that i'm going to get on to and i have discussions 00:14:47.299 --> 00:14:49.100 now and we actually had a discussion earlier 00:14:49.100 --> 00:14:51.019 in the week i mentioned that i was coming on 00:14:51.019 --> 00:14:52.960 this podcast with you so having a discussion 00:14:52.960 --> 00:14:54.919 with some of the other tutors earlier in the 00:14:54.919 --> 00:14:57.320 week about ai and education the problem is at 00:14:57.320 --> 00:15:00.419 the moment there is no line it you know it's 00:15:00.419 --> 00:15:03.799 so wishy -washy there's there's no boundary as 00:15:03.799 --> 00:15:06.460 to what is acceptable what isn't and because 00:15:06.460 --> 00:15:08.980 it's constantly changing so quickly it's hard 00:15:08.980 --> 00:15:12.440 to place that line and i think that's the issue 00:15:12.440 --> 00:15:16.000 but i think what concerns me is because it's 00:15:16.000 --> 00:15:18.899 so up in the air and no one knows what's going 00:15:18.899 --> 00:15:22.159 on the universities and the schools from what 00:15:22.159 --> 00:15:25.379 i can gather are scared of it and i think that's 00:15:25.379 --> 00:15:27.730 not the right approach I think, you know, to 00:15:27.730 --> 00:15:30.090 just be scared of it and to be being like, you 00:15:30.090 --> 00:15:31.830 know, I think I could easily see it getting to 00:15:31.830 --> 00:15:33.509 the point where it's like, you cannot use this 00:15:33.509 --> 00:15:35.149 because we don't know what's going on with it. 00:15:35.210 --> 00:15:37.750 You cannot use it. And I think that's a concerning 00:15:37.750 --> 00:15:39.850 approach because whether you like it or not, 00:15:39.950 --> 00:15:43.919 it's there. It's coming. And it needs to be integrated 00:15:43.919 --> 00:15:46.779 into the way that we teach and into what we teach. 00:15:47.139 --> 00:15:49.879 Yeah, absolutely. I'm saying the same thing in 00:15:49.879 --> 00:15:55.019 schools as well. The people who are fearful are 00:15:55.019 --> 00:15:57.500 insisting, for example, in English, you can imagine 00:15:57.500 --> 00:16:01.279 that coursework in GCSE is 40 % of the course. 00:16:01.759 --> 00:16:05.120 And it's hard to detect. AI detectors aren't 00:16:05.120 --> 00:16:07.320 really fit for purpose. They're more likely to 00:16:07.320 --> 00:16:09.840 pick up that someone's an EAL pupil or an SED 00:16:09.840 --> 00:16:15.799 pupil. And so what's the solution? There are 00:16:15.799 --> 00:16:17.799 solutions. In fact, a couple of episodes ago, 00:16:17.840 --> 00:16:20.759 I talked about ways in which you can adapt writing 00:16:20.759 --> 00:16:25.340 tasks in order to, to a degree, protect students 00:16:25.340 --> 00:16:29.500 from the temptation to use AI. But I think you're 00:16:29.500 --> 00:16:32.299 spot on in also saying that we want to engender 00:16:32.299 --> 00:16:34.960 a culture of academic integrity where people 00:16:34.960 --> 00:16:38.360 want to express themselves rather than outsource 00:16:38.360 --> 00:16:40.399 it to a machine because it's their thoughts, 00:16:40.419 --> 00:16:44.610 their ideas. I mean, I'm still navigating this 00:16:44.610 --> 00:16:47.470 myself, but if there's two kind of golden rules 00:16:47.470 --> 00:16:50.250 that I work from at the moment, it's one, what 00:16:50.250 --> 00:16:52.590 you've said, augment rather than replace your 00:16:52.590 --> 00:16:55.149 own thinking. Yes. And then the second one is 00:16:55.149 --> 00:16:58.679 don't. Don't ask AI to do things because they're 00:16:58.679 --> 00:17:01.620 hard. Ask AI to do something if it's mundane 00:17:01.620 --> 00:17:06.420 or routine or dull. Outsource that stuff. But 00:17:06.420 --> 00:17:08.220 the hard work, the critical thinking and the 00:17:08.220 --> 00:17:11.319 problem solving and the ideation, that's where 00:17:11.319 --> 00:17:13.859 we want to be making sure that the human is not 00:17:13.859 --> 00:17:17.299 just in the loop but driving the process. 100%. 00:17:17.299 --> 00:17:23.619 Yeah, completely agree. Yeah. And so do you foresee 00:17:23.619 --> 00:17:26.650 that... the course that you attended is going 00:17:26.650 --> 00:17:28.650 to look very different a couple of years from 00:17:28.650 --> 00:17:31.369 now because of the way in which ai is now finding 00:17:31.369 --> 00:17:34.750 its way into the process i think fundamentally 00:17:34.750 --> 00:17:37.710 it has to and that there has to be a big shift 00:17:37.710 --> 00:17:39.990 and i think especially because it's a creative 00:17:39.990 --> 00:17:43.970 degree you know there's so many ways that ai 00:17:43.970 --> 00:17:48.390 can affect this degree not you know it has elements 00:17:48.390 --> 00:17:52.980 of um you know reports and creative writing it 00:17:52.980 --> 00:17:55.279 has elements of image imagery and illustration 00:17:55.279 --> 00:17:58.880 it has you know your engineering your mechanical 00:17:58.880 --> 00:18:01.779 side of things you know one of the things that 00:18:01.779 --> 00:18:04.619 i've seen students doing in this course that 00:18:04.619 --> 00:18:06.700 i'm teaching at the moment so we're doing a project 00:18:06.700 --> 00:18:08.900 actually which is linked to the company that 00:18:08.900 --> 00:18:10.900 i'm working for so it's allowing them to do a 00:18:10.900 --> 00:18:13.259 project working with industry so the second year 00:18:13.259 --> 00:18:16.099 students are currently designing autonomous kitchen 00:18:16.099 --> 00:18:19.240 modules for mars which is really fun but one 00:18:19.240 --> 00:18:20.720 of the things that i've noticed them doing is 00:18:20.720 --> 00:18:24.160 they can actually take their circuit boards and 00:18:24.160 --> 00:18:27.279 their electronics take a picture of it put it 00:18:27.279 --> 00:18:30.299 into chat gpt and it will tell them where the 00:18:30.299 --> 00:18:34.460 wiring is wrong okay um you know and and that 00:18:34.460 --> 00:18:37.960 is that's one of the things where there's now 00:18:37.960 --> 00:18:40.680 a discussion of okay Is that stopping them from 00:18:40.680 --> 00:18:43.400 learning, you know, how to actually do it themselves 00:18:43.400 --> 00:18:45.539 and how to find the issues themselves? Or is 00:18:45.539 --> 00:18:48.039 it just teaching them how to do it and showing 00:18:48.039 --> 00:18:51.119 them, you know, it's those sorts of things need 00:18:51.119 --> 00:18:54.900 to be integrated into the course because just 00:18:54.900 --> 00:18:57.519 to be able to say, no, you can't do that. Well, 00:18:57.579 --> 00:18:59.039 actually, no, there's some good things that come 00:18:59.039 --> 00:19:01.220 from that. You know, it might be stopping them 00:19:01.220 --> 00:19:03.140 from doing certain things, but there's good things 00:19:03.140 --> 00:19:05.779 that come from that. So the course has to change 00:19:05.779 --> 00:19:10.460 fundamentally to. allow for ai in certain levels 00:19:10.460 --> 00:19:13.099 but again it's just the boundary of where does 00:19:13.099 --> 00:19:16.140 that end and is it going to change in two weeks 00:19:16.140 --> 00:19:18.759 is it going to change in two months and and this 00:19:18.759 --> 00:19:22.160 is as you say it's so fluid as well if we if 00:19:22.160 --> 00:19:25.539 we had a system whereby we knew that an ai model 00:19:25.539 --> 00:19:29.440 could 100 without hallucinations correctly identify 00:19:29.440 --> 00:19:32.380 electronic components and wiring to the point 00:19:32.380 --> 00:19:35.519 where the human's not required is it still a 00:19:35.519 --> 00:19:39.160 skill which it should be fundamental to the program 00:19:39.160 --> 00:19:41.420 in the same way that most people now can't do 00:19:41.420 --> 00:19:44.460 fire by friction because they don't need to so 00:19:44.460 --> 00:19:46.400 there was a time where that was an absolutely 00:19:46.400 --> 00:19:49.039 essential skill for anyone in a tribe it's not 00:19:49.039 --> 00:19:52.099 now and it may be that things that we now say 00:19:52.099 --> 00:19:55.519 i mean i i have great respect for coders and 00:19:55.519 --> 00:19:57.859 people who teach themselves coding but you know 00:19:57.859 --> 00:20:00.539 even um some of the people at the cutting edge 00:20:00.539 --> 00:20:03.400 of um ai are now saying you know six to nine 00:20:03.400 --> 00:20:07.059 months and Most coding, 90 % of coding will be 00:20:07.059 --> 00:20:09.460 done by AI platforms because it's faster, it's 00:20:09.460 --> 00:20:11.359 efficient. It doesn't mean that we don't need 00:20:11.359 --> 00:20:13.259 developers, but they're going to focus on the 00:20:13.259 --> 00:20:16.359 part that AI doesn't do well or where there's 00:20:16.359 --> 00:20:19.819 a more creative element. But maybe, I don't know, 00:20:19.980 --> 00:20:21.859 but maybe you do still need the fundamentals 00:20:21.859 --> 00:20:24.160 in order to then be able to do the more complex 00:20:24.160 --> 00:20:27.950 work afterwards. And it's how we do that, how 00:20:27.950 --> 00:20:31.509 we still engender a respect and investment of 00:20:31.509 --> 00:20:35.069 the time needed for the fundamentals. When students 00:20:35.069 --> 00:20:37.069 know all you need to do is take that picture. 00:20:37.529 --> 00:20:42.069 I mean, I became a victim of my own success with 00:20:42.069 --> 00:20:45.529 AI a few weeks back. I've been teaching ways 00:20:45.529 --> 00:20:48.950 in which AI can be leveraged. And I gave a very 00:20:48.950 --> 00:20:54.460 simple kind of 10 question task on... a text 00:20:54.460 --> 00:20:55.960 that we just studied. And one of the students 00:20:55.960 --> 00:20:58.259 pulled his phone out, photographed it, gave it 00:20:58.259 --> 00:21:00.859 to chat GPT and gave me the answers in a minute. 00:21:00.940 --> 00:21:03.799 And that was my 15 minute activity was just destroyed. 00:21:04.240 --> 00:21:06.480 And he smiled smugly. And then I thought, you 00:21:06.480 --> 00:21:08.400 know what? You're right. That's not on you. That's 00:21:08.400 --> 00:21:11.740 on me because I created a task that was mundane, 00:21:11.859 --> 00:21:15.019 routine, simple with a bit more creativity. I 00:21:15.019 --> 00:21:16.920 should have come up with something that was AI 00:21:16.920 --> 00:21:19.559 proofed and also more engaging so that that wasn't 00:21:19.559 --> 00:21:22.359 the first thing he wanted to do. Yeah. Yeah. 00:21:22.920 --> 00:21:25.720 Yeah, no, I completely agree. And I think you're 00:21:25.720 --> 00:21:28.259 completely right. And there was a quote that 00:21:28.259 --> 00:21:31.839 came up in one of your podcasts earlier on about, 00:21:31.920 --> 00:21:35.220 you know, AI should be doing the laundry, not 00:21:35.220 --> 00:21:38.599 the art. Yeah, I completely agree with that. 00:21:38.680 --> 00:21:41.599 And I think, you know, it should be used as a 00:21:41.599 --> 00:21:44.119 tool to do those mundane tasks, almost to the 00:21:44.119 --> 00:21:46.740 point that, you know, as a tool, it should almost 00:21:46.740 --> 00:21:49.160 disappear into the process. You know, it should 00:21:49.160 --> 00:21:51.660 become an extension of the thought, whether it's 00:21:52.119 --> 00:21:54.799 Whether it's ChatGPT helping with research, whether 00:21:54.799 --> 00:21:57.640 it's, I still do a lot of music stuff and I've 00:21:57.640 --> 00:22:00.400 explored things like Suno, whether it's Suno 00:22:00.400 --> 00:22:02.839 helping to shape musical compositions rather 00:22:02.839 --> 00:22:05.900 than to replace the music that you do, whether 00:22:05.900 --> 00:22:08.420 it's AI aiding in software development, I think 00:22:08.420 --> 00:22:12.160 the true value lies in how these tools amplify 00:22:12.160 --> 00:22:16.059 creativity rather than dilute it. absolutely 00:22:16.059 --> 00:22:18.319 um was andrew weeks at the school in the music 00:22:18.319 --> 00:22:20.859 department when you were he was he was indeed 00:22:20.859 --> 00:22:24.119 yeah so andrew and i are due to have a conversation 00:22:24.119 --> 00:22:28.579 just like ours on sunday um okay sort of a precursor 00:22:28.579 --> 00:22:32.599 to a future interview and he and I ran an AI 00:22:32.599 --> 00:22:35.640 music competition using Suno where my students 00:22:35.640 --> 00:22:38.900 in my AI course created musical tracks and then 00:22:38.900 --> 00:22:41.420 he adjudicated and I also threw in some human 00:22:41.420 --> 00:22:43.140 tracks and he had to figure out which were which 00:22:43.140 --> 00:22:48.180 and then to comment on the quality of the AI 00:22:48.180 --> 00:22:51.059 generated music and it led to some very rich 00:22:51.059 --> 00:22:54.099 conversations about exactly that. What are the 00:22:54.099 --> 00:22:57.680 concerns about AI music? system stealing work 00:22:57.680 --> 00:23:01.359 from human composers because he makes a good 00:23:01.359 --> 00:23:03.440 part of his living from original compositions 00:23:03.440 --> 00:23:07.599 um in what ways is it an empowering tool because 00:23:07.599 --> 00:23:10.500 it democratizes the process and allows people 00:23:10.500 --> 00:23:12.640 like me who have not got a musical bone in my 00:23:12.640 --> 00:23:15.000 body to be able to write a song and then put 00:23:15.000 --> 00:23:18.400 it to music um and then what's the shadow side 00:23:18.400 --> 00:23:20.619 what is it going to do more harm than good you 00:23:20.619 --> 00:23:23.579 know yeah yeah but i think you know it's it's 00:23:23.579 --> 00:23:27.039 a very interesting tool and once again it's something 00:23:27.039 --> 00:23:32.039 that i haven't personally used to replace anything 00:23:32.039 --> 00:23:34.700 and it you know i'm currently working on i've 00:23:34.700 --> 00:23:36.559 released a couple of singles last year i'm working 00:23:36.559 --> 00:23:39.019 on a an album at the moment and you know there's 00:23:39.019 --> 00:23:41.039 a couple of tracks in this album that i've written 00:23:41.039 --> 00:23:45.660 that are duets and in order to kind of show the 00:23:45.660 --> 00:23:48.000 person that is singing that track with me what 00:23:48.000 --> 00:23:49.900 it would sound like when those two voices are 00:23:49.900 --> 00:23:52.970 kind of together what i've used utilize suno 00:23:52.970 --> 00:23:57.490 for is i've been able to sing their part record 00:23:57.490 --> 00:24:00.710 it into suno get that back as a female vocal 00:24:00.710 --> 00:24:03.289 layer that with mine and send that to them to 00:24:03.289 --> 00:24:05.150 be able to be like you know this is this is kind 00:24:05.150 --> 00:24:07.470 of what i'm looking at but the interesting thing 00:24:07.470 --> 00:24:10.269 that's come from that is suno doesn't always 00:24:10.269 --> 00:24:13.650 replicate it as exactly as you've sung it and 00:24:13.650 --> 00:24:15.650 so sometimes i get things back and it does it 00:24:15.650 --> 00:24:18.009 slightly differently and i'm like oh i didn't 00:24:18.009 --> 00:24:19.970 think of doing it like that and that is actually 00:24:20.559 --> 00:24:22.900 encouraged me to change certain parts of how 00:24:22.900 --> 00:24:26.319 I've sung certain lines in these songs so again 00:24:26.319 --> 00:24:29.660 it's all about that aid not replace I think And 00:24:29.660 --> 00:24:32.000 so, yeah, and I can imagine in the future, assuming 00:24:32.000 --> 00:24:34.200 that AI continues to become more nuanced and 00:24:34.200 --> 00:24:36.799 refined, that we could genuinely have a collaborative 00:24:36.799 --> 00:24:38.960 creative process where it's actually hard to 00:24:38.960 --> 00:24:42.680 say where the AI's contribution ends and the 00:24:42.680 --> 00:24:45.319 human picks up from that and vice versa. And 00:24:45.319 --> 00:24:48.000 the end product is richer for the, you know, 00:24:48.000 --> 00:24:50.359 we could achieve the same thing, perhaps something 00:24:50.359 --> 00:24:52.799 better collaborating with other humans, but sometimes 00:24:52.799 --> 00:24:57.200 that's just not practical or possible. And I 00:24:57.200 --> 00:24:59.779 do hope that we find ourselves in a kind of a 00:24:59.779 --> 00:25:03.160 renaissance of the arts where rather than feeling 00:25:03.160 --> 00:25:05.819 that human beings are being displaced by AI, 00:25:06.079 --> 00:25:08.980 instead we find a kind of a rich new vein of 00:25:08.980 --> 00:25:14.559 creativity. 100%. So in your experience of using 00:25:14.559 --> 00:25:16.940 AI, whether it's Suno or ChatGPT or whatever, 00:25:17.099 --> 00:25:19.740 what are the things that you found empowering 00:25:19.740 --> 00:25:22.019 about it? And where's the frustration or the 00:25:22.019 --> 00:25:25.380 pain points for you at the moment? I think... 00:25:26.349 --> 00:25:29.390 In general, I would say the thing I would say 00:25:29.390 --> 00:25:31.710 that the same thing that makes it empowering 00:25:31.710 --> 00:25:34.269 is the thing that makes it frustrating at the 00:25:34.269 --> 00:25:36.029 moment because of the different roles that I'm 00:25:36.029 --> 00:25:40.529 doing. I think the fact that it's so easy to 00:25:40.529 --> 00:25:44.589 access and available makes it so powerful and 00:25:44.589 --> 00:25:48.549 efficient. But I think that is that ease of access 00:25:48.549 --> 00:25:51.829 is also what is unfortunately in some situations 00:25:51.829 --> 00:25:54.660 allowing some people to almost skip. the creative 00:25:54.660 --> 00:25:57.380 process entirely so i think it's it's a bit of 00:25:57.380 --> 00:25:58.980 both you know i think the thing that makes it 00:25:58.980 --> 00:26:00.839 empowering and frustrating is the same thing 00:26:00.839 --> 00:26:02.900 and it's you know the fact that it's it's there 00:26:02.900 --> 00:26:07.079 and it's so easily accessible you know it i think 00:26:07.079 --> 00:26:10.079 it completely depends on who's using it and what 00:26:10.079 --> 00:26:13.400 they're utilizing it for i think you know ultimately 00:26:13.400 --> 00:26:17.839 ai should elevate human ability I think that's 00:26:17.839 --> 00:26:20.339 what we've already discussed, and I'm sure you 00:26:20.339 --> 00:26:22.220 agree on that. It should elevate human ability 00:26:22.220 --> 00:26:24.740 rather than replace it, and I think its power 00:26:24.740 --> 00:26:28.140 does genuinely lie in its accessibility and its 00:26:28.140 --> 00:26:31.039 efficiency, but that's also its greatest flaw 00:26:31.039 --> 00:26:35.599 in how easily it enables bypassing process. Yeah, 00:26:35.660 --> 00:26:38.339 and have there been use cases where you've wanted 00:26:38.339 --> 00:26:40.039 to achieve something and it hasn't been able 00:26:40.039 --> 00:26:44.119 to do it for you? There has, yeah. There's been 00:26:44.119 --> 00:26:48.890 times when... I've been trying to generate imagery. 00:26:49.970 --> 00:26:54.029 So one of the things I have used it for is, for 00:26:54.029 --> 00:26:56.450 example, if I'm trying to storyboard for a client, 00:26:56.609 --> 00:26:59.470 if I'm, you know, this and this is for industry, 00:26:59.609 --> 00:27:02.410 I'm trying to storyboard, you know, things that 00:27:02.410 --> 00:27:04.170 we are doing and things that we're trying to 00:27:04.170 --> 00:27:06.630 present. And I've only got a certain period of 00:27:06.630 --> 00:27:09.150 time to do that. I would rather spend the majority 00:27:09.150 --> 00:27:13.029 of my time making sure the story is right than 00:27:13.029 --> 00:27:16.579 spending ages doing the illustration. Even though 00:27:16.579 --> 00:27:18.880 I enjoy the illustration and that's a creative 00:27:18.880 --> 00:27:20.619 part of the process, I would rather that it's 00:27:20.619 --> 00:27:22.299 more important that the story is right and all 00:27:22.299 --> 00:27:24.380 those sections are right. And so there have been 00:27:24.380 --> 00:27:26.900 times where I've then utilized AI to try and 00:27:26.900 --> 00:27:28.940 help me at least come up with ideas or generate 00:27:28.940 --> 00:27:32.559 imagery. And I would say with generating the 00:27:32.559 --> 00:27:35.660 imagery, that's been times where it hasn't quite 00:27:35.660 --> 00:27:38.460 got it and it hasn't shown what I've wanted it 00:27:38.460 --> 00:27:40.519 to show. And then you kind of go through this 00:27:40.519 --> 00:27:42.539 loop of iterating and being like, can you change 00:27:42.539 --> 00:27:44.539 this? Can you change that? And sometimes it just 00:27:44.539 --> 00:27:47.599 can't. Yeah, yeah, and I've run into the same 00:27:47.599 --> 00:27:49.700 thing. So do you use something like Midjourney 00:27:49.700 --> 00:27:53.769 as your art generator? Midjourney. Yeah, so Midjourney 00:27:53.769 --> 00:27:56.009 is both incredible and incredibly frustrating 00:27:56.009 --> 00:27:58.569 at the same time because it almost gets where 00:27:58.569 --> 00:28:01.109 you wanted it to be. And then I often find myself 00:28:01.109 --> 00:28:03.609 then taking it into Photoshop and using the warp 00:28:03.609 --> 00:28:08.630 tool and generative fill. In fact, if there's 00:28:08.630 --> 00:28:10.670 time after the call, I'll show you a piece of 00:28:10.670 --> 00:28:13.230 software that my students and I have been playing 00:28:13.230 --> 00:28:15.029 with called Newark. I don't know whether you've 00:28:15.029 --> 00:28:17.740 come across that. No, I haven't. No, no. I will 00:28:17.740 --> 00:28:20.319 at some point do a recording on it. The guys 00:28:20.319 --> 00:28:22.220 at Newark have been incredibly generous and they've 00:28:22.220 --> 00:28:24.779 kind of given us free credits for all the students 00:28:24.779 --> 00:28:28.980 I teach to play with it. But in short, it allows 00:28:28.980 --> 00:28:32.000 designers to be able to take a sketch and then, 00:28:32.019 --> 00:28:35.220 so you could sketch out the exact body posture 00:28:35.220 --> 00:28:37.700 you wanted for a character or the position of 00:28:37.700 --> 00:28:40.180 the... you know the kitchen on the moon or whatever 00:28:40.180 --> 00:28:42.740 it might be yeah and then but but as rough lines 00:28:42.740 --> 00:28:45.559 and then you bring that in as your um template 00:28:45.559 --> 00:28:49.000 image and then you can also drag in elements 00:28:49.000 --> 00:28:52.440 for textures and surfaces and then it will adhere 00:28:52.440 --> 00:28:55.799 exactly to the line you're outlined but populate 00:28:55.799 --> 00:28:58.579 it and shape it from the prompt you give it as 00:28:58.579 --> 00:29:02.039 well so for example i took a sketch that my son 00:29:02.039 --> 00:29:05.039 had done when he was four of a monster and imported 00:29:05.039 --> 00:29:09.500 that very kind of um childish line sketch and 00:29:09.500 --> 00:29:11.960 then turned it into a three -dimensional plushie 00:29:11.960 --> 00:29:16.500 that adhered exactly to the design um but i was 00:29:16.500 --> 00:29:18.400 able to import the kind of fluffiness and the 00:29:18.400 --> 00:29:21.539 the right colors and it's incredible um and i 00:29:21.539 --> 00:29:24.299 wonder whether that might be you know some something 00:29:24.299 --> 00:29:26.259 that you with your design skills might be able 00:29:26.259 --> 00:29:28.480 to leverage and make really good use of 100 yeah 00:29:28.480 --> 00:29:33.539 definitely so looking forward um to wait the 00:29:33.539 --> 00:29:36.299 way things are going to change Can you imagine 00:29:36.299 --> 00:29:39.059 how you might be augmenting your work moving 00:29:39.059 --> 00:29:42.400 forward if AI continues to advance and it becomes 00:29:42.400 --> 00:29:45.619 more powerful? Let's assume that we deal with 00:29:45.619 --> 00:29:47.980 the proprietary data and the protection of that 00:29:47.980 --> 00:29:49.940 data so that you're working with models that 00:29:49.940 --> 00:29:53.759 run in -house. In a kind of Star Trek world, 00:29:54.039 --> 00:29:56.339 what would you like AI to be able to do that 00:29:56.339 --> 00:29:58.859 it's not doing for you now that you would want 00:29:58.859 --> 00:30:01.500 to outsource without losing the creativity and 00:30:01.500 --> 00:30:04.619 your passion for what you do in the process? 00:30:05.119 --> 00:30:08.319 Yeah, it's a good question. I think ultimately 00:30:08.319 --> 00:30:14.140 the thing that I find not difficult, the thing 00:30:14.140 --> 00:30:16.720 that I find different at the moment in terms 00:30:16.720 --> 00:30:19.440 of especially the company that I'm at is it's 00:30:19.440 --> 00:30:24.640 a small team. And so in terms of what I do, I 00:30:24.640 --> 00:30:27.539 am the only product design engineer. OK, so it 00:30:27.539 --> 00:30:29.400 is me, you know, and I've got people to bounce 00:30:29.400 --> 00:30:31.700 ideas off of. We have a robotics engineer who's 00:30:31.700 --> 00:30:34.359 fantastic. And obviously the CEO and the CTO 00:30:34.359 --> 00:30:36.680 will kind of bounce ideas off each other. But 00:30:36.680 --> 00:30:39.019 I think the thing that I am that I miss from 00:30:39.019 --> 00:30:42.500 my time at university is being in a studio with, 00:30:42.599 --> 00:30:47.089 you know, 30 other people. When I get stuck on 00:30:47.089 --> 00:30:49.430 something and I'm trying to come up with ideas 00:30:49.430 --> 00:30:51.670 and generate different ways of doing something, 00:30:51.769 --> 00:30:53.289 I've got someone there that's like, oh, have 00:30:53.289 --> 00:30:54.829 you considered this? Have you considered that? 00:30:55.309 --> 00:30:59.089 That is the thing that I miss in what I'm doing 00:30:59.089 --> 00:31:02.589 at the moment. So I think the thing that AI can 00:31:02.589 --> 00:31:05.710 already do and that I could utilize that for 00:31:05.710 --> 00:31:09.680 is helping to come up with. you know tweaks of 00:31:09.680 --> 00:31:12.039 ways that i'm doing things and you know being 00:31:12.039 --> 00:31:14.160 able to be like this is what i've got this bit's 00:31:14.160 --> 00:31:16.579 not quite working i can't work out you know this 00:31:16.579 --> 00:31:20.160 that way to do it yeah um before i had people 00:31:20.160 --> 00:31:22.059 there that could kind of jump in on that and 00:31:22.059 --> 00:31:24.240 i think ai has the ability to do that but like 00:31:24.240 --> 00:31:27.299 you say at the moment with how things are i wouldn't 00:31:27.299 --> 00:31:30.000 trust it in terms of the confidentiality and 00:31:30.000 --> 00:31:36.299 the ip so genuine collaboration um on a on a 00:31:36.299 --> 00:31:39.019 level of creativity so that it's augmenting what 00:31:39.019 --> 00:31:41.579 you do and kind of bouncing ideas back or saying 00:31:41.579 --> 00:31:43.700 i really like this here's three alternative ideas 00:31:43.700 --> 00:31:45.700 and then you can choose what to do with it 100 00:31:45.700 --> 00:31:48.619 yeah i don't imagine we're very far away from 00:31:48.619 --> 00:31:53.230 that becoming a reality um yeah So, yeah, I definitely 00:31:53.230 --> 00:31:55.430 think we should be keeping an eye out for locally 00:31:55.430 --> 00:31:59.009 hosted models that can run big systems and see 00:31:59.009 --> 00:32:01.609 where it goes. So before we wrap things up, the 00:32:01.609 --> 00:32:03.210 last thing is I want you to imagine that I take 00:32:03.210 --> 00:32:05.349 you back to Norwich School now and you're restarting 00:32:05.349 --> 00:32:07.549 your sixth form. Terrifying a prospect there 00:32:07.549 --> 00:32:11.329 that might be. And you're doing your... Presumably, 00:32:11.349 --> 00:32:13.349 would you study the same subjects that you did, 00:32:13.490 --> 00:32:15.150 given that it's taken you to where you are now? 00:32:15.930 --> 00:32:20.619 Yeah, yeah. I mean, I was in a very... lucky 00:32:20.619 --> 00:32:23.039 position that i always knew what i wanted to 00:32:23.039 --> 00:32:26.579 do from a very a very young age i mean i remember 00:32:26.579 --> 00:32:29.900 i was picking my speaking my gcses when i was 00:32:29.900 --> 00:32:32.539 that must have been 13 is it when you pick yeah 00:32:32.539 --> 00:32:35.509 13 coming on to 14 yeah um And I remember picking 00:32:35.509 --> 00:32:38.450 my GCSEs and I was looking at companies that 00:32:38.450 --> 00:32:39.910 I wanted to work for and what I wanted to do. 00:32:39.950 --> 00:32:41.710 And I was looking at companies like Apple and 00:32:41.710 --> 00:32:44.329 their product design engineer roles. And then 00:32:44.329 --> 00:32:46.410 I looked at what degree I needed to do to get 00:32:46.410 --> 00:32:48.470 onto that. And I looked at what A -levels I needed 00:32:48.470 --> 00:32:50.210 to do to get onto that degree. And then I picked 00:32:50.210 --> 00:32:52.289 what GCSEs I needed to do to get onto that A 00:32:52.289 --> 00:32:54.990 -level. Yeah, so you were very focused. Yeah, 00:32:55.289 --> 00:32:57.230 I've always been very sort of tunnel vision on 00:32:57.230 --> 00:32:58.789 exactly what I wanted to do and where I wanted 00:32:58.789 --> 00:33:02.990 to get to. So I don't think I would change anything 00:33:02.990 --> 00:33:06.059 if I was to go back. Okay, so you're back there. 00:33:06.119 --> 00:33:08.220 You've got two years before you start your tertiary 00:33:08.220 --> 00:33:12.000 studies, but AI is where it is right now in terms 00:33:12.000 --> 00:33:16.420 of its ability. Do you think that schools like 00:33:16.420 --> 00:33:19.960 the one that you attended, it requires them at 00:33:19.960 --> 00:33:23.299 this stage to adapt their curriculum, their delivery, 00:33:23.440 --> 00:33:26.410 the assessment process or whatever? in order 00:33:26.410 --> 00:33:29.569 to give you the opportunity to learn about AI 00:33:29.569 --> 00:33:32.369 and use it? Because you've already said that 00:33:32.369 --> 00:33:34.970 a blanket ban on access is not an enlightened 00:33:34.970 --> 00:33:39.190 approach. Yeah, yeah. And do you think as a 13, 00:33:39.329 --> 00:33:41.349 14 -year -old version of you, you would have 00:33:41.349 --> 00:33:43.809 avoided the temptation to have ChatGPT doing 00:33:43.809 --> 00:33:46.670 all your homework? I'm just interested to see 00:33:46.670 --> 00:33:49.009 if you can imagine into that, what would that 00:33:49.009 --> 00:33:52.950 look like for you? Yeah, I think I definitely... 00:33:53.740 --> 00:33:56.480 I can say with some certainty that I would never 00:33:56.480 --> 00:34:01.779 have used AI to just replace doing work. And, 00:34:01.900 --> 00:34:03.880 you know, I've never been, you know, there's 00:34:03.880 --> 00:34:07.259 always been ways of cutting corners. Before AI, 00:34:07.400 --> 00:34:08.639 there's always been ways of cutting corners. 00:34:08.760 --> 00:34:10.500 And, you know, if someone's gonna cut corners, 00:34:10.619 --> 00:34:12.300 they're gonna cut corners, whether AI is there 00:34:12.300 --> 00:34:15.679 or not. And I've never been that sort of person 00:34:15.679 --> 00:34:19.070 because I genuinely value... learning how to 00:34:19.070 --> 00:34:21.829 do things. And I genuinely care about knowing 00:34:21.829 --> 00:34:23.929 how things work rather than just skipping to 00:34:23.929 --> 00:34:28.590 the end point. But I do think I definitely would 00:34:28.590 --> 00:34:32.650 have utilized AI tools. And it goes back to what 00:34:32.650 --> 00:34:35.489 we said, it's there. And I think it's naive to 00:34:35.489 --> 00:34:38.659 just pretend that it's not. yeah and so it definitely 00:34:38.659 --> 00:34:41.840 it has to be integrated into what we're doing 00:34:41.840 --> 00:34:44.099 and it's it's something that is becoming a part 00:34:44.099 --> 00:34:47.460 of everyday life and so to not teach people about 00:34:47.460 --> 00:34:50.000 that you know we teach people about you know 00:34:50.000 --> 00:34:51.860 almost everything else that is a part of everyday 00:34:51.860 --> 00:34:53.539 life because it's something that we have to live 00:34:53.539 --> 00:34:54.679 through and it's something that we have to learn 00:34:54.679 --> 00:34:57.000 and something we have to go through so i think 00:34:57.000 --> 00:35:00.599 it is only common sense for learning about ai 00:35:00.599 --> 00:35:04.980 and its benefits and its disadvantages to become 00:35:04.980 --> 00:35:08.099 a part of the curriculum i think it has to i 00:35:08.099 --> 00:35:11.079 i completely agree i actually had a tutorial 00:35:11.079 --> 00:35:13.699 period today with some sixth form students from 00:35:13.699 --> 00:35:16.219 the school and i was basically trying to kind 00:35:16.219 --> 00:35:19.179 of wake them up to the idea that most of them 00:35:19.179 --> 00:35:22.000 had used chat gpt at some point but none of them 00:35:22.000 --> 00:35:23.800 were really familiar with the range of tools 00:35:23.800 --> 00:35:25.280 that were available and what they could do in 00:35:25.280 --> 00:35:28.059 terms of coding for example yeah and i was kind 00:35:28.059 --> 00:35:30.480 of just inviting them to think on the fact that 00:35:30.480 --> 00:35:32.639 they're going into a profession where if you 00:35:32.639 --> 00:35:35.719 aren't able to augment your workflow with ai 00:35:35.719 --> 00:35:37.960 tools you're probably going to be an anachronism 00:35:37.960 --> 00:35:40.920 by the time you leave university so now's the 00:35:40.920 --> 00:35:43.460 time to start upskilling and thinking about ways 00:35:43.460 --> 00:35:45.960 in which you can work alongside these tools and 00:35:45.960 --> 00:35:48.659 and do it in an ethical way and i think it got 00:35:48.659 --> 00:35:51.059 a good reception i think they were both receptive 00:35:51.059 --> 00:35:53.260 and kind of quite excited by it but also a little 00:35:54.210 --> 00:35:56.829 intimidated by the learning curve that that involves 00:35:56.829 --> 00:35:58.949 because it can seem from the outside it does 00:35:58.949 --> 00:36:01.650 seem quite an intimidating thing but once you're 00:36:01.650 --> 00:36:03.550 starting to use the tools it's actually it's 00:36:03.550 --> 00:36:06.389 not that challenging is it really the user interface 00:36:06.389 --> 00:36:10.389 is a pretty straightforward 100 and i think you 00:36:10.389 --> 00:36:13.809 know the future of ai is only going to become 00:36:13.809 --> 00:36:17.289 more prominent and i think you know i think in 00:36:17.289 --> 00:36:19.809 terms of what's happening next with ai i think 00:36:19.809 --> 00:36:22.349 the point where it really has a real impact is 00:36:22.349 --> 00:36:26.280 when it becomes more physical and it moves beyond 00:36:26.280 --> 00:36:28.300 software and i think that's the point you know 00:36:28.300 --> 00:36:31.420 right now ai is mostly it's mostly digital digital 00:36:31.420 --> 00:36:35.440 but i think once it becomes integrated into robotics 00:36:35.440 --> 00:36:37.460 and physical products that's when it's going 00:36:37.460 --> 00:36:39.119 to change everyday life and you know that may 00:36:39.119 --> 00:36:41.739 be biased from a product designer's view but 00:36:41.739 --> 00:36:45.340 i think the future of ai really is not in digital 00:36:45.340 --> 00:36:48.519 interactions but it's in seamless integration 00:36:48.519 --> 00:36:52.139 of of products you know of hardware and software 00:36:52.670 --> 00:36:56.929 um and objects that will ultimately enrich and 00:36:56.929 --> 00:37:00.210 enhance sort of the fundamental rhythms of daily 00:37:00.210 --> 00:37:02.329 life you know i think it's the real shift is 00:37:02.329 --> 00:37:05.610 when it's going to help with everyday physical 00:37:05.610 --> 00:37:07.590 tasks whether that's cooking cleaning shopping 00:37:07.590 --> 00:37:11.650 that's when the real revolution is gonna gonna 00:37:11.650 --> 00:37:14.230 come i think yeah and i would imagine i can't 00:37:14.230 --> 00:37:17.099 imagine a kitchen on the moon that doesn't have 00:37:17.099 --> 00:37:19.900 humanoid robots operating within that kitchen. 00:37:20.519 --> 00:37:24.360 Most of the people at the cutting edge of robotics, 00:37:24.780 --> 00:37:27.219 when I'm watching or reading material from them, 00:37:27.340 --> 00:37:30.059 they're talking about the end of this year for 00:37:30.059 --> 00:37:33.679 the first AI -augmented humanoid robots to be 00:37:33.679 --> 00:37:35.920 in family homes. Initially, there's going to 00:37:35.920 --> 00:37:37.639 be a price point that will make it prohibitive 00:37:37.639 --> 00:37:39.460 for most people, but we know that that's going 00:37:39.460 --> 00:37:42.599 to come down. So probably three to five years, 00:37:42.780 --> 00:37:45.670 I would imagine. I mean, you've got obviously 00:37:45.670 --> 00:37:48.210 you've got companies like you've got Tesla and 00:37:48.210 --> 00:37:50.289 you've got companies like One X who are developing 00:37:50.289 --> 00:37:53.030 humanoid robots. And One X is a really interesting 00:37:53.030 --> 00:37:55.389 one. If people haven't looked into them, I highly 00:37:55.389 --> 00:37:57.510 recommend it because they've kind of been in 00:37:57.510 --> 00:37:59.610 the background and then they've suddenly revealed 00:37:59.610 --> 00:38:02.510 this humanoid robot. And it's way more advanced 00:38:02.510 --> 00:38:04.769 than anyone expected. I think Tesla are probably 00:38:04.769 --> 00:38:08.289 shaking in their boots. But companies like that 00:38:08.289 --> 00:38:10.690 who are developing these sort of humanoid robots 00:38:10.690 --> 00:38:13.900 with this AI, I think that is going to be. a 00:38:13.900 --> 00:38:18.420 game changer yeah absolutely and and also you 00:38:18.420 --> 00:38:21.840 can imagine the the utility of that for example 00:38:21.840 --> 00:38:24.480 working in environments that are high risk to 00:38:24.480 --> 00:38:27.460 human beings where you know toxic exposure and 00:38:27.460 --> 00:38:29.980 so forth yeah as well as you know care for the 00:38:29.980 --> 00:38:32.599 elderly and you can see just lots of practical 00:38:32.599 --> 00:38:34.960 ways in which it could improve the world yeah 00:38:34.960 --> 00:38:38.480 100 well we should have another conversation 00:38:38.480 --> 00:38:42.239 maybe in a year or so's time and see whether 00:38:42.239 --> 00:38:45.139 things have moved on and see what's changed yeah 00:38:45.139 --> 00:38:47.320 absolutely but i really appreciate the time you 00:38:47.320 --> 00:38:50.000 spent with us um yeah and you will however many 00:38:50.000 --> 00:38:51.800 interviews i do you will always be the first 00:38:51.800 --> 00:38:54.639 you were the first interviewer thank you for 00:38:54.639 --> 00:38:56.699 having me i appreciate it it's a pleasure and 00:38:56.699 --> 00:38:58.820 thanks for taking the time good luck in your 00:38:58.820 --> 00:39:01.320 projects and yes let's stay in touch and let 00:39:01.320 --> 00:39:03.380 people know once we get the first kitchen on 00:39:03.380 --> 00:39:05.889 the moon Hopefully you'll have your name stamped 00:39:05.889 --> 00:39:08.050 somewhere on the drawers. Hopefully I'll be up 00:39:08.050 --> 00:39:11.190 there installing it. Would you want to? I would 00:39:11.190 --> 00:39:15.949 love to. If NASA are listening, get me on a Falcon 00:39:15.949 --> 00:39:17.969 9, I would love it. Well, listen, you're young, 00:39:18.030 --> 00:39:19.630 you're in shape, and you've got the skillset 00:39:19.630 --> 00:39:21.010 and the knowledge. I can't see why they wouldn't. 00:39:22.309 --> 00:39:24.690 Thanks very much. Take care. Awesome. Thank you. 00:39:25.909 --> 00:39:28.050 I hope you enjoyed that interview with Bailey 00:39:28.050 --> 00:39:30.610 Tudman. Thanks for listening. Please do spread 00:39:30.610 --> 00:39:32.530 the word if you think others would like the show 00:39:32.530 --> 00:39:35.710 and do check out the AIcademia YouTube channel 00:39:35.710 --> 00:39:38.730 where you'll find practical tutorials that complement 00:39:38.730 --> 00:39:42.130 the topics covered in this podcast. Have a great 00:39:42.130 --> 00:39:44.289 week and I look forward to catching up again 00:39:44.289 --> 00:39:44.610 soon.