1 00:00:00,040 --> 00:00:02,320 What does it mean to be a software engineer today? 2 00:00:02,400 --> 00:00:04,800 I think it's a very different landscape to what it was even 3 00:00:04,800 --> 00:00:05,520 two years ago. You. 4 00:00:05,720 --> 00:00:08,680 Can't even hire other people unless you can justify how AI 5 00:00:08,680 --> 00:00:11,560 couldn't actually help you with that extra resource, right? 6 00:00:11,800 --> 00:00:13,880 But what do we do about our junior engineers? 7 00:00:13,880 --> 00:00:17,200 It's going to be harder to break into the industry harder than it 8 00:00:17,200 --> 00:00:20,160 was say three years ago. I think we're trading off that 9 00:00:20,400 --> 00:00:23,320 short term speed increase and what we perceive to be 10 00:00:23,320 --> 00:00:26,120 productivity gains, which they are, but at what cost. 11 00:00:26,400 --> 00:00:29,960 That speed really is very seductive and fun, but fragility 12 00:00:30,280 --> 00:00:32,360 and complexity is very expensive. 13 00:00:32,520 --> 00:00:35,520 One of the main questions that I'm asking as part of my 14 00:00:35,680 --> 00:00:38,440 research is how does the introduction of AI coding 15 00:00:38,440 --> 00:00:43,320 assistance shift the perceived focus across various development 16 00:00:43,320 --> 00:00:46,760 tasks such as like designing and writing, refactoring, testing, 17 00:00:46,760 --> 00:00:49,600 debugging, reviewing? If we've just taken a technology 18 00:00:49,600 --> 00:00:52,520 that can speed up one of those phases by a lot, well, where's 19 00:00:52,520 --> 00:00:54,200 the next bottleneck? We'll come up with the 20 00:00:54,200 --> 00:00:57,400 requirements and then the other end as well, who will do the 21 00:00:57,400 --> 00:01:01,120 review and testing and all that. One might ask themselves, well, 22 00:01:01,120 --> 00:01:03,480 where is that extra time going then because we're saving all 23 00:01:03,480 --> 00:01:06,080 this time? Or are we just writing more code 24 00:01:06,080 --> 00:01:08,520 faster? I would suggest to start 25 00:01:08,520 --> 00:01:12,800 thinking of these AI tools as not just mere tools, but as team 26 00:01:12,800 --> 00:01:15,160 members. I do wonder if we might head to 27 00:01:15,160 --> 00:01:18,960 a place where we each have our own AI that knows us intimate. 28 00:01:19,160 --> 00:01:22,560 What if it could also connect to a model that is yours? 29 00:01:22,560 --> 00:01:24,880 Is your personal one that knows the way that you think? 30 00:01:41,760 --> 00:01:44,280 Hello everyone, welcome back to the new episode of the 31 00:01:44,280 --> 00:01:47,320 Technician podcast. Today I have with me a repeat 32 00:01:47,320 --> 00:01:50,400 guest, Annie Vella. So I think the previous episode 33 00:01:50,400 --> 00:01:53,600 that we did is episode 30 like 4 years ago. 34 00:01:53,720 --> 00:01:57,000 It's been quite a long while. The reason I reach out to Annie 35 00:01:57,000 --> 00:02:00,240 is because I saw her quite active on social media, on 36 00:02:00,240 --> 00:02:04,120 LinkedIn and also on her research, right, talking about 37 00:02:04,120 --> 00:02:07,920 AI assisted development or AI in general, right, generative AI in 38 00:02:07,920 --> 00:02:11,039 software engineering. So today I hope we have a good 39 00:02:11,039 --> 00:02:14,520 discussion to talk about, you know, Dehype about AI, maybe 40 00:02:14,520 --> 00:02:18,680 some fears that some people have and try to maybe demystify and 41 00:02:18,680 --> 00:02:21,920 give us some better guidance. So any welcome to the show. 42 00:02:22,600 --> 00:02:25,320 Thank you, Henry. It's so good to be back after, 43 00:02:25,320 --> 00:02:30,720 yeah, just over 4 years I think. And boy, has a bit changed in 44 00:02:30,720 --> 00:02:32,360 the world in that time, hasn't it? 45 00:02:33,120 --> 00:02:36,680 Yeah. So any, I think we know about AI 46 00:02:36,680 --> 00:02:39,000 usage these days, right? Everyone is talking about it 47 00:02:39,000 --> 00:02:42,080 over social media. There are new technologies even 48 00:02:42,080 --> 00:02:46,040 being invented recently, right. And the term vibe coding is kind 49 00:02:46,040 --> 00:02:48,040 of like very, very popular these days. 50 00:02:48,400 --> 00:02:50,840 So maybe tell us a little bit more four years apart, right. 51 00:02:50,840 --> 00:02:54,040 What do you see have changed, maybe in your day-to-day work 52 00:02:54,040 --> 00:02:56,400 with software engineering, Or what do you see other people 53 00:02:56,400 --> 00:03:00,240 doing in software engineering? Well, four years in a time when 54 00:03:00,240 --> 00:03:03,920 things have changed as much. So personally I've changed roles 55 00:03:03,920 --> 00:03:06,960 since the last time we spoke. So I think when we spoke last 56 00:03:06,960 --> 00:03:10,960 time I was about to start a senior engineering manager role 57 00:03:11,400 --> 00:03:16,200 and since then I've actually moved into a on the staff plus 58 00:03:16,200 --> 00:03:19,880 engineering career track. I'm a distinguished engineer and 59 00:03:19,880 --> 00:03:22,400 so my own role has changed in that time. 60 00:03:22,400 --> 00:03:25,760 And so the things that I spend time doing have also changed. 61 00:03:25,760 --> 00:03:30,440 But more importantly, I would say end of 2022 or November 62 00:03:30,440 --> 00:03:33,360 2022, I think it was when ChatGPT was released. 63 00:03:33,800 --> 00:03:36,600 I think that caught the world by surprise. 64 00:03:36,600 --> 00:03:41,840 And what's been so interesting is to see how quickly it's been 65 00:03:41,840 --> 00:03:43,400 adopted. I was watching a video just 66 00:03:43,400 --> 00:03:49,280 yesterday, Sam Altman said that there's 500 million active users 67 00:03:49,280 --> 00:03:53,320 using ChatGPT. That's just a huge number that 68 00:03:53,400 --> 00:03:54,880 it's hard to wrap your head around, right? 69 00:03:55,360 --> 00:03:57,960 But then starting to see where it's being applied. 70 00:03:58,200 --> 00:04:01,240 Not just ChatGPT, but the underlying technology, LLMS, 71 00:04:01,760 --> 00:04:04,440 obviously very, very useful in software engineering. 72 00:04:04,440 --> 00:04:08,320 And hence, as you said, you know, vibe coding, all of these 73 00:04:08,320 --> 00:04:10,280 sorts of terms are starting to come out. 74 00:04:10,560 --> 00:04:13,880 And I was lucky enough to be in a position where I could 75 00:04:13,880 --> 00:04:17,480 entertain the idea of going back to university to do the masters 76 00:04:17,480 --> 00:04:20,640 that I always told myself I would do when I graduated over 77 00:04:20,640 --> 00:04:22,800 20 years ago. I thought, oh, I'll go back to 78 00:04:22,800 --> 00:04:26,480 university one day and do that further study, but I had to find 79 00:04:26,480 --> 00:04:30,480 the thing that was going to pull me in and, you know, encourage 80 00:04:30,480 --> 00:04:33,720 me to spend my free time thinking about it. 81 00:04:33,720 --> 00:04:39,160 So yeah, early 2024, I decided to do a masters of Engineering, 82 00:04:39,400 --> 00:04:45,320 focusing on the impact of AI on software engineering because, I 83 00:04:45,320 --> 00:04:48,200 mean, it just feels like the most natural topic for me to 84 00:04:48,200 --> 00:04:50,400 focus on. I love software engineering so 85 00:04:50,400 --> 00:04:52,800 much. I've done it since I was such a 86 00:04:52,800 --> 00:04:54,640 young. I think we covered this probably 87 00:04:54,640 --> 00:04:57,040 when we spoke last time. But you know, since the age of 88 00:04:57,040 --> 00:05:00,800 6, I have been fascinated at what you can accomplish by 89 00:05:00,800 --> 00:05:02,680 instructing a computer to do something. 90 00:05:03,160 --> 00:05:04,800 But there's always been limitations, right? 91 00:05:04,800 --> 00:05:08,840 It can't think for itself. You have to be so specific in 92 00:05:08,840 --> 00:05:11,400 your instructions. But this technology starts to 93 00:05:11,400 --> 00:05:15,760 allow us to be a little more fuzzy with our instructions that 94 00:05:15,760 --> 00:05:19,800 non determinism that we didn't really have before or was very 95 00:05:19,800 --> 00:05:22,760 hard to achieve. And so, yeah, how does that 96 00:05:22,760 --> 00:05:25,880 impact the way that we write software, the type of software 97 00:05:25,880 --> 00:05:28,040 that we build, but also how we build software? 98 00:05:28,440 --> 00:05:31,000 And that's what I focused my part time masters on. 99 00:05:31,000 --> 00:05:33,400 So I'm about halfway through. And that's why you will have 100 00:05:33,400 --> 00:05:37,440 seen that I'm one of the reasons I'm so active on social media 101 00:05:37,440 --> 00:05:39,640 now about this is because I find it fascinating. 102 00:05:39,640 --> 00:05:43,400 I think a lot of people do, but particularly the impact on 103 00:05:43,400 --> 00:05:46,000 software engineering and the software engineer themselves. 104 00:05:46,000 --> 00:05:49,200 You know, what does it mean to be a software engineer today? 105 00:05:49,680 --> 00:05:52,360 I think it's a very different landscape to what it was even 106 00:05:52,360 --> 00:05:54,920 two years ago or four years ago when we spoke. 107 00:05:55,320 --> 00:05:57,480 The sorts of skills that you need to build up, you know, 108 00:05:57,480 --> 00:06:00,640 that's all changing. So, OK, I wish I had a crystal 109 00:06:00,640 --> 00:06:03,280 ball. But as I do not, the best I can 110 00:06:03,280 --> 00:06:06,760 do is try and research what's happening today so that we might 111 00:06:06,760 --> 00:06:10,600 project and predict what happens tomorrow, and so we can prepare 112 00:06:10,600 --> 00:06:14,160 ourselves for a brave new world. Yeah, So I think it's really 113 00:06:14,160 --> 00:06:17,080 interesting the, you know, you just saw a video, you know, the 114 00:06:17,520 --> 00:06:19,480 active users using ChatGPT, right? 115 00:06:19,480 --> 00:06:21,640 I guess these days it's not just ChatGPT. 116 00:06:21,640 --> 00:06:25,320 There are even more, right? You have like perplexity Gemini, 117 00:06:25,320 --> 00:06:28,680 you know Claude and I don't know God, what, you know, like other 118 00:06:29,080 --> 00:06:31,800 AI software. So I think also interestingly, 119 00:06:31,840 --> 00:06:35,000 just a few days ago as well, there was a announcement by 120 00:06:35,000 --> 00:06:38,960 Shopify CEO, right, Toby Lucas saying that everyone now is 121 00:06:38,960 --> 00:06:41,760 expected to use AI, the term reflexive AI. 122 00:06:41,760 --> 00:06:45,360 I'm not sure what that means, but basically saying that you 123 00:06:45,360 --> 00:06:48,600 can't even hire other people unless and justify how AI 124 00:06:48,600 --> 00:06:52,040 couldn't actually help you with that extra resource, right? 125 00:06:52,280 --> 00:06:56,280 So I think that's a very, very new thing in my opinion, right? 126 00:06:56,280 --> 00:06:59,600 So asking people to try to use AI as much as possible. 127 00:06:59,960 --> 00:07:03,120 So I think in your paper you mentioned about this future of 128 00:07:03,160 --> 00:07:05,680 AI driven software engineering or I think let's just focus on 129 00:07:05,680 --> 00:07:08,680 this topic for today. What do you see changing? 130 00:07:08,680 --> 00:07:11,840 You know, software development is still kind of like, you know, 131 00:07:11,840 --> 00:07:14,360 a large kind of a work body, right? 132 00:07:14,360 --> 00:07:17,800 So that there's so many activities, people try to use it 133 00:07:17,800 --> 00:07:20,480 so much, you know, building applications, building programs 134 00:07:20,800 --> 00:07:22,400 and things like that. But what do you see actually 135 00:07:22,400 --> 00:07:24,920 changing in terms of software engineering in general? 136 00:07:25,640 --> 00:07:28,640 Yeah. So that paper was an attempt to 137 00:07:28,720 --> 00:07:32,680 look into the future and see what of all of the different 138 00:07:32,680 --> 00:07:35,480 types of activities that are involved in the software 139 00:07:35,480 --> 00:07:38,680 development life cycle, which of those would be impacted and in 140 00:07:38,680 --> 00:07:42,360 what ways. My own research for my masters 141 00:07:42,360 --> 00:07:45,640 is focusing more on like the things that software engineers 142 00:07:46,480 --> 00:07:48,760 do themselves today. And this is something that I've 143 00:07:48,760 --> 00:07:51,680 been reflecting on so much as a part of all of this. 144 00:07:52,320 --> 00:07:55,760 You know, when I went to university over 20 years ago, I 145 00:07:55,760 --> 00:07:58,680 was taught the full software development life cycle. 146 00:07:58,680 --> 00:08:02,080 You know, I, of course we learnt how to code, but we also learnt 147 00:08:02,080 --> 00:08:07,120 how to analyse the requirements, you know, and discuss with the 148 00:08:07,120 --> 00:08:09,600 customer what the problem they were trying to solve and 149 00:08:09,600 --> 00:08:12,200 encourage them to not give us the solution, but instead just 150 00:08:12,200 --> 00:08:16,360 really focus on the problem and let us as of the team who are 151 00:08:16,440 --> 00:08:18,920 going to solve all their problems, come up with the right 152 00:08:18,920 --> 00:08:21,520 solutions for them. You know, right through to 153 00:08:21,840 --> 00:08:25,360 modelling use cases, UML, you know, back then when we used to 154 00:08:25,360 --> 00:08:28,680 do a lot more of that kind of stuff upfront on the design of 155 00:08:28,680 --> 00:08:31,200 the software and then the build and thinking through all the 156 00:08:31,360 --> 00:08:34,840 test scenarios and then obviously the deployment and the 157 00:08:34,840 --> 00:08:38,360 maintenance of it. And what I've noticed throughout 158 00:08:38,360 --> 00:08:43,039 the years is that probably due to the fact that there's such a 159 00:08:43,039 --> 00:08:46,000 massive demand for software engineers, but not enough 160 00:08:46,000 --> 00:08:50,120 supply, we have created specializations, you know, and, 161 00:08:50,120 --> 00:08:55,960 and try to kind of reduce or narrow down what a software 162 00:08:55,960 --> 00:08:59,160 engineer should focus on to the things that we perceive only 163 00:08:59,160 --> 00:09:00,680 they can do with the skills that they have. 164 00:09:00,680 --> 00:09:02,040 And that is the writing of the code. 165 00:09:02,480 --> 00:09:06,040 What I find really fascinating about this shift is if you you 166 00:09:06,040 --> 00:09:11,160 read what people like the CEO of GitHub have to say, you know, 167 00:09:11,160 --> 00:09:14,200 they proudly announced that now with these tools, software 168 00:09:14,200 --> 00:09:16,600 engineers will be able to do a lot more of that higher level 169 00:09:16,600 --> 00:09:20,640 thinking, critical thinking, you know, all the other tasks that 170 00:09:20,640 --> 00:09:23,640 were always meant to be a part of software engineering. 171 00:09:23,880 --> 00:09:26,440 But the truth of the matter is that many software engineers 172 00:09:26,440 --> 00:09:29,160 have specialized on the coding aspect of it. 173 00:09:29,640 --> 00:09:32,840 I used to work many years ago when I was in the Netherlands. 174 00:09:33,000 --> 00:09:36,880 I worked with a team who were based in the UK and I remember 175 00:09:37,320 --> 00:09:42,000 one of their engineers there was just really, really smart, but 176 00:09:42,000 --> 00:09:45,920 focused all of his energy and his skill building and becoming 177 00:09:45,960 --> 00:09:49,400 like the best React developer. He knew React inside out. 178 00:09:49,400 --> 00:09:54,320 That was his area of interest and his domain, you know, and we 179 00:09:54,320 --> 00:09:57,080 had frequent discussions because I was trying to encourage him to 180 00:09:57,080 --> 00:09:59,600 think more about the product that he was building as well. 181 00:09:59,600 --> 00:10:02,440 Like what is the problem you're trying to solve with this 182 00:10:02,440 --> 00:10:04,360 technology that you've become very good at? 183 00:10:04,960 --> 00:10:07,080 And he just didn't have much interest in understanding the 184 00:10:07,080 --> 00:10:10,280 domain we were working in. You know, he felt his area of 185 00:10:10,280 --> 00:10:12,560 expertise should be the technology and others could 186 00:10:12,560 --> 00:10:16,120 think about the product. Well, I fear for people who have 187 00:10:16,560 --> 00:10:19,200 gone down that path because I think that the tools that we 188 00:10:19,200 --> 00:10:22,280 have today and, and of course, it's only going to get better. 189 00:10:22,840 --> 00:10:24,400 They're very good at writing the code. 190 00:10:24,600 --> 00:10:29,000 They can generate code way faster than any of us can hope 191 00:10:29,000 --> 00:10:32,280 to ever type. So that shifts things, right? 192 00:10:32,360 --> 00:10:35,760 But where does it shift it to? So one of the main questions 193 00:10:35,760 --> 00:10:39,640 that I'm asking as part of my research is how does the 194 00:10:39,640 --> 00:10:44,200 introduction of AI coding assistance shift the perceived 195 00:10:44,200 --> 00:10:49,400 focus across various development tasks such as like designing and 196 00:10:49,400 --> 00:10:52,160 writing, refactoring, testing, debugging, reviewing. 197 00:10:52,600 --> 00:10:55,520 You know, like I have a hypothesis that developers may 198 00:10:55,520 --> 00:10:58,320 spend less time writing code and perhaps more time reviewing. 199 00:10:58,920 --> 00:11:03,040 Well, so far the data is showing that yes, developers feel that 200 00:11:03,040 --> 00:11:06,240 they will spend or that they already are spending less time 201 00:11:06,240 --> 00:11:09,280 writing code, this time refactoring code as well, and 202 00:11:09,280 --> 00:11:11,520 this time testing because they can generate tests. 203 00:11:11,520 --> 00:11:14,240 That all makes sense. But what we're not seeing is 204 00:11:14,240 --> 00:11:17,880 really like an equal and opposite increase in the 205 00:11:17,880 --> 00:11:20,240 reviewing. The time spent reviewing. 206 00:11:20,680 --> 00:11:23,720 There is a slight increase there, but not equivalent, 207 00:11:23,720 --> 00:11:26,200 right? So one might ask themselves, 208 00:11:26,200 --> 00:11:28,560 well, where is that extra time going then because you know, 209 00:11:28,560 --> 00:11:32,000 we're saving all this time or are we just writing more code 210 00:11:32,000 --> 00:11:34,680 faster? And what does that mean into the 211 00:11:34,680 --> 00:11:36,680 future, right? What are the 2nd order or 212 00:11:36,680 --> 00:11:39,320 lagging indicators that we should start looking for? 213 00:11:39,840 --> 00:11:44,080 Well, thankfully we don't need to look too far because Dora and 214 00:11:44,080 --> 00:11:48,800 the Get Clear report, I came across that one a couple of 215 00:11:48,800 --> 00:11:53,400 weeks ago and you know, they're analyzing the impacts of this 216 00:11:53,520 --> 00:11:57,120 technological shift and we are already seeing some of those 217 00:11:57,120 --> 00:11:59,680 second order or lagging indicators. 218 00:12:00,200 --> 00:12:04,120 And it's not really a great picture to be honest, Henry. 219 00:12:04,120 --> 00:12:07,960 It's like I think we're trading off that short term speed 220 00:12:07,960 --> 00:12:11,440 increase and what we perceive to be productivity gains, which 221 00:12:11,440 --> 00:12:16,520 they are, but at what cost, you know, and I think that's going 222 00:12:16,520 --> 00:12:19,840 to become quite interesting. So software developers right 223 00:12:19,840 --> 00:12:21,960 now, they're generating code faster. 224 00:12:21,960 --> 00:12:25,160 They're also feeling like some of the code that gets generated 225 00:12:25,160 --> 00:12:27,160 isn't great. You know, sometimes it 226 00:12:27,200 --> 00:12:30,680 hallucinates for all the tools that are available, some are 227 00:12:30,680 --> 00:12:34,080 better than others, but you still need to be vigilant and 228 00:12:34,320 --> 00:12:37,840 manually adjust and carefully verify that the code that it's 229 00:12:37,840 --> 00:12:41,360 generated is doing what you hope it is, which is where vibe 230 00:12:41,360 --> 00:12:44,000 coding comes in and tells you actually, maybe you don't need 231 00:12:44,000 --> 00:12:46,520 to worry too much about that. So that's another topic. 232 00:12:47,040 --> 00:12:50,160 And another thing that's quite interesting that I'm seeing in 233 00:12:50,160 --> 00:12:54,920 my research is software engineers are trading the more 234 00:12:54,920 --> 00:12:58,920 traditional sort of search tools for AI coding assistance. 235 00:12:58,920 --> 00:13:01,560 So previously you would have gone to Stack Overflow, you 236 00:13:01,560 --> 00:13:05,160 would have gone to Google, maybe even opened up the documentation 237 00:13:05,640 --> 00:13:07,840 for a particular library or framework that you're working 238 00:13:07,840 --> 00:13:10,040 with. That actually seems to be 239 00:13:10,040 --> 00:13:12,840 shifting as well. So we can already see a decline 240 00:13:13,160 --> 00:13:16,760 in the use of Stack Overflow. You know, it had its day and it 241 00:13:16,760 --> 00:13:20,040 seems like now it's just easier to stay in the IDE. 242 00:13:20,040 --> 00:13:23,520 Why would you pop out of it into a browser if you can just ask 243 00:13:23,840 --> 00:13:25,960 the AI coding assistant to help you with that? 244 00:13:26,720 --> 00:13:30,960 So are we moving towards being more like orchestrators of AI 245 00:13:30,960 --> 00:13:33,680 coding assistants, Like just being able to describe what it 246 00:13:33,680 --> 00:13:36,600 is that you want or getting really good at prompting? 247 00:13:37,000 --> 00:13:40,960 Is prompt engineering becoming a new core competency, and does it 248 00:13:40,960 --> 00:13:46,160 potentially replace the ability to recall all of the syntax that 249 00:13:46,160 --> 00:13:49,360 we used to have to learn? So these are the big questions 250 00:13:49,360 --> 00:13:52,800 I'm asking, and I'm, yeah, still super fascinated about. 251 00:13:52,840 --> 00:13:55,160 It's an evolving picture, absolutely. 252 00:13:55,760 --> 00:13:57,520 Yeah. So as you were mentioning all 253 00:13:57,520 --> 00:14:00,880 those, I kind of like reflect back to what my recent 254 00:14:00,880 --> 00:14:03,080 experience as well. So I could definitely relate 255 00:14:03,080 --> 00:14:05,480 with some of these things. These are changing especially 256 00:14:05,480 --> 00:14:08,640 for me, right. So uniquely enough, right, I was 257 00:14:08,640 --> 00:14:11,800 also into management now back kind of like semi management, 258 00:14:11,800 --> 00:14:14,360 semi hands on. So I think I could see that 259 00:14:14,360 --> 00:14:16,800 so-called the dichotomy, you know, being a hands on developer 260 00:14:16,800 --> 00:14:20,800 and also as a manager, right? So trying to think of how we can 261 00:14:20,800 --> 00:14:24,480 use AI more effectively and being more productive and also 262 00:14:24,480 --> 00:14:28,400 not forgetting about the second level or maybe third level order 263 00:14:28,400 --> 00:14:30,960 of thinking, right? So I think 1 aspect that you 264 00:14:30,960 --> 00:14:33,440 mentioned very interesting is about the role of software 265 00:14:33,440 --> 00:14:35,360 engineering. You know, traditionally we were 266 00:14:35,360 --> 00:14:39,240 trained with so many different things like requirements and 267 00:14:39,240 --> 00:14:42,120 things like that. But over the time as maybe like 268 00:14:42,120 --> 00:14:45,440 what you mentioned, like so many solver developers specializing 269 00:14:45,440 --> 00:14:48,440 into different roles, some even specializing in different 270 00:14:48,440 --> 00:14:50,760 technologies. Maybe it's a React, maybe it's a 271 00:14:50,760 --> 00:14:54,960 cloud thing, maybe it's DevOps, SRE, whatever that is, right? 272 00:14:55,360 --> 00:14:59,680 And now it seems like these kind of specialization is blurring a 273 00:14:59,680 --> 00:15:02,520 little bit, I would say. So for example, if I'm not a if 274 00:15:02,520 --> 00:15:05,200 I'm not a good front end engineer, but I do understand a 275 00:15:05,200 --> 00:15:07,360 little bit about front end technology, right? 276 00:15:07,520 --> 00:15:10,800 I could now come up with some kind of maybe decent good 277 00:15:10,960 --> 00:15:14,920 looking UI and also complimenting myself with other 278 00:15:14,920 --> 00:15:16,920 skills like back end engineer or maybe cloud. 279 00:15:17,160 --> 00:15:21,280 I can build much better application with more coverage 280 00:15:21,280 --> 00:15:23,400 in terms of, you know, technology thing, right? 281 00:15:23,720 --> 00:15:27,360 So do you, do you think these days software engineers have to 282 00:15:27,360 --> 00:15:31,480 think of becoming a generalist more and more, especially now 283 00:15:31,480 --> 00:15:35,000 that AI can help actually to maybe provide shortcuts, you 284 00:15:35,000 --> 00:15:38,200 know, finding the answers quickly, but also upscaling 285 00:15:38,200 --> 00:15:40,480 themselves in those skills? So what do you think about this? 286 00:15:41,120 --> 00:15:43,440 Yes, I very much agree with that. 287 00:15:44,040 --> 00:15:49,640 And you know, I think as time goes on, different phases will 288 00:15:49,640 --> 00:15:52,240 come and go. You know, maybe we had to go 289 00:15:52,240 --> 00:15:56,160 through a phase where there were so many new technologies coming 290 00:15:56,160 --> 00:15:58,080 up. I lost count of the number of 291 00:15:58,320 --> 00:16:01,760 front end frameworks that, you know, went through the hype 292 00:16:02,040 --> 00:16:04,400 phase. You know, I think I stopped 293 00:16:05,320 --> 00:16:07,960 playing around with and I say playing because I was never 294 00:16:07,960 --> 00:16:10,440 officially a front end developer, but as a sort of full 295 00:16:10,440 --> 00:16:13,240 stack developer, I did have to work in front ends and I was 296 00:16:13,240 --> 00:16:15,680 never very good at it. But I think the last time I did 297 00:16:15,680 --> 00:16:18,120 anything decent in front end was angular days. 298 00:16:18,120 --> 00:16:21,240 And you imagine how far, like, how long ago that feels now. 299 00:16:21,720 --> 00:16:25,360 But, yeah, we've gone through a phase where we've introduced so 300 00:16:25,360 --> 00:16:28,880 many new frameworks and technologies, cloud and mobile 301 00:16:28,880 --> 00:16:32,560 itself, you know, And people chose to specialize because 302 00:16:32,680 --> 00:16:36,160 there was a demand for it. But now, as you say, like, these 303 00:16:36,160 --> 00:16:40,560 tools do in many ways democratize the access to 304 00:16:41,160 --> 00:16:44,720 becoming a developer, somebody who can contribute to 305 00:16:44,720 --> 00:16:47,520 development. In fact, I read something by 306 00:16:47,600 --> 00:16:50,000 Andre Carpathy. I hope I'm saying his name 307 00:16:50,000 --> 00:16:52,440 right. Apparently Twitter has articles 308 00:16:52,440 --> 00:16:54,200 on it now. I, I didn't and it's not called 309 00:16:54,200 --> 00:16:56,440 Twitter. It takes, but he wrote an X 310 00:16:56,440 --> 00:17:01,040 article on the diffusion of this technology, which previously 311 00:17:01,080 --> 00:17:05,359 like big technological changes like this might have started in 312 00:17:05,359 --> 00:17:10,440 government or big companies 1st and then trickled down to the 313 00:17:10,440 --> 00:17:15,000 average individual at home. You know, but this technology 314 00:17:15,000 --> 00:17:18,520 seems to have been released out to the general public 1st. 315 00:17:18,520 --> 00:17:21,880 And although I'm sure that there are, you know, more advanced 316 00:17:21,880 --> 00:17:25,079 models that the companies are using internally and maybe even 317 00:17:25,079 --> 00:17:28,200 the government, the perception is that it's out in the public 318 00:17:28,200 --> 00:17:30,680 much earlier than it would normally have been with other 319 00:17:30,680 --> 00:17:33,120 technologies. And that means that, you know, 320 00:17:33,120 --> 00:17:36,360 the general public, you and I and everybody else can use these 321 00:17:36,360 --> 00:17:39,760 technologies to sort of supercharge themselves. 322 00:17:39,760 --> 00:17:43,360 So as you say, like I too have found myself, I've had to write 323 00:17:43,360 --> 00:17:47,400 a bunch of R scripts to analyse the data that I've collected as 324 00:17:47,400 --> 00:17:50,000 part of my research. And I'm not an art developer, 325 00:17:50,000 --> 00:17:54,040 but with the help of, I'm using Windsurf connecting to clawed 326 00:17:54,040 --> 00:17:59,240 Sonnet 3.5, it's a huge help. It just shortcuts the amount of 327 00:17:59,240 --> 00:18:02,320 time I would have had to spend. And now I feel like I could 328 00:18:02,320 --> 00:18:05,760 probably write something in any language with just a little bit 329 00:18:05,760 --> 00:18:08,000 of help. What I love about that, Henry, 330 00:18:08,000 --> 00:18:11,680 is, you know, last time we spoke, I was right in the middle 331 00:18:11,680 --> 00:18:13,840 of trying to be the best engineering manager and 332 00:18:13,840 --> 00:18:15,080 technical leader that I could be. 333 00:18:15,080 --> 00:18:20,760 I really focused so much of my energy on how do I give other 334 00:18:20,760 --> 00:18:26,520 engineers enough confidence to try things and fail sometimes, 335 00:18:26,520 --> 00:18:29,320 but safely, you know, because you learn through your failures 336 00:18:29,320 --> 00:18:32,480 and your mistakes, probably better lessons than you'll ever 337 00:18:32,480 --> 00:18:35,560 learn any other way. How do you create an environment 338 00:18:35,560 --> 00:18:39,760 in which people lean into that and then are able to discover 339 00:18:39,760 --> 00:18:44,000 for themselves some new path, some new way of some creative 340 00:18:44,000 --> 00:18:47,760 way of solving a problem and be able to feel really proud about 341 00:18:47,760 --> 00:18:49,960 that? And I think like we spend so 342 00:18:49,960 --> 00:18:53,400 much time working, so much of our lives are spent at work, 343 00:18:53,720 --> 00:18:57,680 that if you can take joy in your work and be proud of the work 344 00:18:57,680 --> 00:19:00,040 that you do, it makes a time fly by. 345 00:19:00,040 --> 00:19:03,160 And that's, I think part of my passion for software engineering 346 00:19:03,160 --> 00:19:04,840 is I've never really seen it as work. 347 00:19:04,840 --> 00:19:06,560 It's always been an enjoyable thing for me. 348 00:19:07,120 --> 00:19:10,800 So I wonder if these tools can help give perhaps the more 349 00:19:10,800 --> 00:19:16,640 introverted, quieter members of our teams like a quiet assistant 350 00:19:16,640 --> 00:19:19,520 just sitting there privately on their computer. 351 00:19:19,720 --> 00:19:24,000 You know, now, those who might previously have not had the 352 00:19:24,000 --> 00:19:27,240 courage to put their hand up and admit to the rest of their team 353 00:19:27,680 --> 00:19:30,120 that they don't know how to do something, that they need a bit 354 00:19:30,120 --> 00:19:32,280 of help. Because, you know, it's can be 355 00:19:32,280 --> 00:19:35,360 hard to admit that, you know, not everybody has the courage to 356 00:19:35,360 --> 00:19:37,480 do that. And some, you know, over time 357 00:19:37,480 --> 00:19:40,120 might find that courage, but some might never. 358 00:19:40,640 --> 00:19:43,520 But this tool, if you can just chat with it about some idea 359 00:19:43,520 --> 00:19:46,800 that you have before you get to the point of, OK, I think I 360 00:19:46,800 --> 00:19:49,200 really need to ask for help, You might be able to solve a lot 361 00:19:49,200 --> 00:19:53,960 more of your own insecurities about trying some technical 362 00:19:53,960 --> 00:19:56,600 solution for something. And I love the idea of that. 363 00:19:56,600 --> 00:20:00,160 I think that can really help allow developers. 364 00:20:00,320 --> 00:20:01,960 Yeah, more freedom in that sense. 365 00:20:02,200 --> 00:20:05,240 And to that end, as you said, like now previously you thought 366 00:20:05,240 --> 00:20:07,520 of yourself as not a very good front end developer, so you 367 00:20:07,520 --> 00:20:09,960 probably never. Going to put your hand up to do 368 00:20:09,960 --> 00:20:11,440 that piece of work or for that job. 369 00:20:11,440 --> 00:20:15,040 Well, now you might actually really consider actually going 370 00:20:15,040 --> 00:20:17,000 for that. And that's got to be a good 371 00:20:17,000 --> 00:20:21,800 thing, you know, but beneath it, it still needs to be backed by a 372 00:20:21,800 --> 00:20:24,840 deeper understanding of what it is that you're doing because you 373 00:20:24,840 --> 00:20:27,200 do need to think about how long is this going to last? 374 00:20:27,200 --> 00:20:28,880 If it's just a prototype, then go for it. 375 00:20:28,880 --> 00:20:31,440 You know, if it's just going to be thrown away and just proving 376 00:20:31,440 --> 00:20:33,600 something. These tools are amazing and I 377 00:20:33,600 --> 00:20:36,920 see that shine through and some of the open-ended answers that I 378 00:20:36,920 --> 00:20:40,680 got in my research. But if it's a production code 379 00:20:40,680 --> 00:20:44,200 that needs to be maintained over time, you still need to probably 380 00:20:44,520 --> 00:20:47,680 be a little careful about the complexity of the code, the 381 00:20:47,680 --> 00:20:50,960 duplication that you're that you might be stumbling into. 382 00:20:51,480 --> 00:20:54,320 But that's maybe where we see introduction of new tools. 383 00:20:54,320 --> 00:20:57,960 Well, you know, layering of agents that can help you 384 00:20:57,960 --> 00:21:02,480 identify those mistakes as well. Gosh, Pandora's box has been 385 00:21:02,480 --> 00:21:05,520 opened. And, you know, who knows where 386 00:21:06,040 --> 00:21:09,760 this evolution will take us, But it's very interesting. 387 00:21:10,600 --> 00:21:13,400 Yeah, myself, I'm afraid I couldn't keep up with so many 388 00:21:13,400 --> 00:21:15,920 different changes, new technologies, new inventions, 389 00:21:15,920 --> 00:21:19,240 new tools being popped up. And sometimes it's not just one 390 00:21:19,240 --> 00:21:21,640 aspect of software engineering, it could be other things like 391 00:21:21,640 --> 00:21:26,120 testing, generating UI, generating app, things like 392 00:21:26,160 --> 00:21:28,240 bold, lovable and all that, right? 393 00:21:28,640 --> 00:21:32,120 So I think definitely this is something that everyone at least 394 00:21:32,120 --> 00:21:34,200 needs to follow a little bit, I guess. 395 00:21:34,240 --> 00:21:37,480 Otherwise you will be fallen behind like by a lot, right? 396 00:21:37,520 --> 00:21:41,080 Especially from people who leverage these tools everyday. 397 00:21:41,440 --> 00:21:44,200 I think you make a great points about, you know, people who are 398 00:21:44,200 --> 00:21:47,640 maybe a little bit more introverted or shy or maybe some 399 00:21:47,640 --> 00:21:49,520 imposter syndrome as well, right? 400 00:21:49,640 --> 00:21:52,560 They are scared, you know, asking questions with others. 401 00:21:52,560 --> 00:21:56,200 So now I think everyone has some kind of assistant, although you 402 00:21:56,200 --> 00:21:59,160 still need to Fact Check, you still need to try to understand 403 00:21:59,520 --> 00:22:02,040 why it suggests certain things that way, right? 404 00:22:02,280 --> 00:22:05,360 So I think at least now we have some partners that we can just 405 00:22:05,600 --> 00:22:09,880 ask, you know, the partners that never get tired and always seem 406 00:22:09,880 --> 00:22:13,400 to be resourceful, always trying to help us the best they can. 407 00:22:14,040 --> 00:22:15,880 Yeah. You mentioned the time spent for 408 00:22:15,880 --> 00:22:20,400 engineers now has becoming less about writing code because, you 409 00:22:20,400 --> 00:22:23,560 know, writing code is not the bottleneck anymore that everyone 410 00:22:23,560 --> 00:22:27,320 seems to be able to churn out maybe features fast, right? 411 00:22:27,720 --> 00:22:31,560 But all other aspects of software development maybe gets 412 00:22:31,560 --> 00:22:33,120 impacted, right? So to speak, right. 413 00:22:33,320 --> 00:22:35,920 So now if let's say you can write a lot of more features, 414 00:22:35,920 --> 00:22:39,360 that means who will come up with the requirements and then the 415 00:22:39,440 --> 00:22:42,960 other end as well, who will do the review and testing and all 416 00:22:42,960 --> 00:22:45,000 that. So do you see some kind of 417 00:22:45,080 --> 00:22:47,680 problems that could happen because like what you mentioned 418 00:22:47,680 --> 00:22:51,160 the Dora research base saying like there's an increased 419 00:22:51,160 --> 00:22:53,080 productivity there is, right? Definitely. 420 00:22:53,080 --> 00:22:56,840 But the impact of software delivery performance is actually 421 00:22:56,840 --> 00:22:59,440 going down. And maybe the Get Clear report 422 00:22:59,440 --> 00:23:01,720 that you mentioned as well, mentioning that there are a lot 423 00:23:01,720 --> 00:23:04,640 of churns in terms of code they are being produced from, meaning 424 00:23:04,640 --> 00:23:07,520 that you write this code today and you keep changing them over 425 00:23:07,520 --> 00:23:09,400 time, right? And the churn is pretty high. 426 00:23:09,600 --> 00:23:12,640 So what do you see the impact now if let's say code writing is 427 00:23:12,640 --> 00:23:15,680 not the bottleneck anymore? What would other aspects that we 428 00:23:15,680 --> 00:23:19,040 need to be concerned about? That's a really interesting 429 00:23:19,040 --> 00:23:22,880 question. And honestly I think that's if I 430 00:23:22,880 --> 00:23:27,120 could work that out, my research would be complete. 431 00:23:27,760 --> 00:23:32,400 I think you're right to mention like software development life 432 00:23:32,400 --> 00:23:36,160 cycle is a system, right? Over the years we have 433 00:23:36,160 --> 00:23:40,360 identified different phases things work from sort of left to 434 00:23:40,360 --> 00:23:42,760 right. And then we decided maybe we 435 00:23:42,760 --> 00:23:46,080 should have iterations in there because if Waterfall didn't work 436 00:23:46,080 --> 00:23:51,600 out super well for everyone, and if we've just taken a technology 437 00:23:51,600 --> 00:23:55,320 that can speed up one of those phases by a lot, well, where's 438 00:23:55,320 --> 00:23:58,040 the next bottleneck? Like now you don't have enough 439 00:23:58,120 --> 00:24:01,120 requirements coming in or there's not enough UX work 440 00:24:01,120 --> 00:24:03,920 happening up front. And so the developers are sort 441 00:24:03,920 --> 00:24:06,960 of waiting around for where's my next piece of work coming from 442 00:24:06,960 --> 00:24:10,400 because I've done my bit. Now you, it's like every team's 443 00:24:10,880 --> 00:24:13,400 dream, I suppose. You know, like, oh wow, how 444 00:24:13,400 --> 00:24:16,760 developers have nothing to do. It's been a long time since I've 445 00:24:16,760 --> 00:24:19,480 heard anyone say that. And on the other end of the 446 00:24:19,480 --> 00:24:22,480 scale, so you know, we're churning out code much faster. 447 00:24:22,680 --> 00:24:26,040 Who's reviewing it all because if you've played around with 448 00:24:26,040 --> 00:24:29,080 these tools from a code generation perspective, you will 449 00:24:29,080 --> 00:24:32,400 be aware of how much faster it can generate large pull 450 00:24:32,400 --> 00:24:34,640 requests. And I think the Dora report 451 00:24:34,640 --> 00:24:38,920 talks about that, right as they are wondering, they hypothesize 452 00:24:39,160 --> 00:24:43,680 whether some of that delivery stability decrease that they've 453 00:24:43,880 --> 00:24:50,000 started to see might be in part at least related to much larger 454 00:24:50,000 --> 00:24:54,800 PRS that are being generated. And I've felt that myself, to be 455 00:24:54,800 --> 00:24:57,880 honest. That's actually I saw the other 456 00:24:57,880 --> 00:25:00,800 day. Patrick, the plan, the godfather 457 00:25:00,800 --> 00:25:04,760 or grandfather, wife of dev OPS, you know, fabulous man. 458 00:25:04,960 --> 00:25:09,920 He commented on LinkedIn that I think he said it was windsurf or 459 00:25:09,920 --> 00:25:14,000 cursor have introduced this new feature where it makes a sound 460 00:25:14,320 --> 00:25:18,440 when it's completed doing the task that you've given to it. 461 00:25:18,920 --> 00:25:22,280 And I thought, oh, here we go. It's an important feature 462 00:25:22,280 --> 00:25:25,560 because we are already getting used to asking it to do so much 463 00:25:25,560 --> 00:25:28,520 more for us that it can actually take a bit of time. 464 00:25:28,720 --> 00:25:31,840 And I've found myself also sort of waiting there patiently 465 00:25:31,840 --> 00:25:35,560 watching it, thinking, hurry up already, why are you taking so 466 00:25:35,560 --> 00:25:37,600 long? Maybe I need to spin up another 467 00:25:37,960 --> 00:25:41,160 instance of windsurf to get it to do more in parallel. 468 00:25:41,160 --> 00:25:43,680 It's taking too long, so then you get distracted and you go 469 00:25:43,680 --> 00:25:46,360 and do something else. You start reading e-mail or 470 00:25:46,360 --> 00:25:49,440 answering a message. And so you don't know when to 471 00:25:49,440 --> 00:25:51,040 come back unless it makes a little sound. 472 00:25:51,280 --> 00:25:54,120 But that just goes to show that we're moving into this phase now 473 00:25:54,120 --> 00:25:56,560 where it's just generating so much code. 474 00:25:57,000 --> 00:25:59,640 Are you really going to go and have a look at all of the code 475 00:25:59,640 --> 00:26:01,560 it's generated? Like we say we should. 476 00:26:01,600 --> 00:26:05,360 We say it's important, but it will just over time. 477 00:26:05,880 --> 00:26:08,080 Vigilance. We're going to. 478 00:26:08,560 --> 00:26:13,360 Yeah, vigilance decrement is a problem because, well, honestly, 479 00:26:13,360 --> 00:26:15,680 there's vigilance fatigue. You know that when you're 480 00:26:15,680 --> 00:26:18,240 writing it, you're going at the pace of like you're thinking 481 00:26:18,240 --> 00:26:20,680 about it, you're writing the code, you're looking at what 482 00:26:20,680 --> 00:26:23,280 you've written, you're maybe writing a unit test and testing 483 00:26:23,280 --> 00:26:25,520 that what you've written works the way it should. 484 00:26:25,880 --> 00:26:28,080 And there's a sort of a natural pace to it. 485 00:26:28,080 --> 00:26:31,480 But now, if you can generate so much more code, so much quicker, 486 00:26:32,040 --> 00:26:35,040 when do you go and check or that you know what it's done? 487 00:26:35,040 --> 00:26:39,400 And how does the person who now needs to review your code deal 488 00:26:39,400 --> 00:26:43,080 with a much, much larger PR? So I think there will probably 489 00:26:43,080 --> 00:26:46,680 be some new tooling that comes out that helps us with that, you 490 00:26:46,680 --> 00:26:50,760 know, that helps identify what maybe what the areas to focus 491 00:26:50,880 --> 00:26:56,080 our review on might be. And I think this is possibly 492 00:26:56,280 --> 00:26:58,320 part of the brave new world that we're stepping into. 493 00:26:58,320 --> 00:27:04,040 We don't yet know what tooling, what extra phases within the 494 00:27:04,040 --> 00:27:07,480 software development life cycle may need to be introduced or 495 00:27:07,480 --> 00:27:12,840 augmented, you know, to deal with the removal of a bottleneck 496 00:27:12,840 --> 00:27:18,800 that has existed for, I want to say about 50 years, you know, 50 497 00:27:18,800 --> 00:27:22,880 odd years that the software industry has been trying to find 498 00:27:22,880 --> 00:27:27,200 ways to reduce the time it takes to build software by, you know, 499 00:27:27,200 --> 00:27:30,520 little things like the libraries and frameworks that we use today 500 00:27:30,520 --> 00:27:34,200 that abstract away as much of the heavy lifting or the 501 00:27:34,200 --> 00:27:37,160 thinking that we previously would have had to have done. 502 00:27:37,160 --> 00:27:40,120 Like we can't even think about what it must have been like to 503 00:27:40,120 --> 00:27:43,360 do all of this on punch cards, you know, or, or even writing 504 00:27:43,720 --> 00:27:46,680 assembly code directly, how much slower that must have been, 505 00:27:46,680 --> 00:27:48,560 right? So we take for granted what we 506 00:27:48,560 --> 00:27:51,760 have today. And even today we feel like it 507 00:27:51,760 --> 00:27:53,160 could be faster, it could be faster. 508 00:27:53,160 --> 00:27:55,280 So we have a need for speed, right? 509 00:27:55,280 --> 00:27:59,960 Like this speed is exciting, it's fun, but then we need to, 510 00:28:00,040 --> 00:28:03,640 you know, move on to the next. OK, So if that's true now, then 511 00:28:03,640 --> 00:28:05,120 what's the next thing to think about? 512 00:28:05,120 --> 00:28:09,600 So I reckon developers will end up spending more time thinking 513 00:28:09,600 --> 00:28:13,760 about how you verify that the work that you've, the code 514 00:28:13,760 --> 00:28:15,760 that's been written does what it's supposed to do. 515 00:28:15,760 --> 00:28:19,960 And I'm not talking about unit tests or even test automation, 516 00:28:20,280 --> 00:28:23,680 they're a part of it. But we're starting to see, I 517 00:28:23,680 --> 00:28:26,960 only learnt last week, this concept of product managers 518 00:28:27,480 --> 00:28:33,360 building what they call evals. So apparently in 2025, evals are 519 00:28:33,400 --> 00:28:36,840 the new thing that product managers are starting to think 520 00:28:36,840 --> 00:28:40,040 about to evaluate whether the product that they're building 521 00:28:40,040 --> 00:28:44,840 that has AI embedded in it does what it's supposed to do safely 522 00:28:44,840 --> 00:28:48,160 and, and all of that. And it's not quite the same as 523 00:28:48,360 --> 00:28:51,960 the test automation that we've as coders we've traditionally 524 00:28:52,240 --> 00:28:54,840 built and relied on, but it's broader than that. 525 00:28:55,280 --> 00:28:58,000 And it's super interesting. Like already like that didn't 526 00:28:58,000 --> 00:29:01,760 exist two years ago, you know, So I think new things like that 527 00:29:01,760 --> 00:29:06,160 will come up and we'll have to find innovative ways to automate 528 00:29:06,160 --> 00:29:09,880 that and stitch it into our pipelines and anything and 529 00:29:09,880 --> 00:29:14,080 everything around that. Yet even just figuring out when 530 00:29:14,080 --> 00:29:19,320 to trust an AI coding assistant, that in itself, that probably 531 00:29:19,600 --> 00:29:21,760 requires a huge amount of research. 532 00:29:22,000 --> 00:29:26,000 How do you know? Trust is a big topic and it's 533 00:29:26,000 --> 00:29:28,800 not one that we typically talk about in software engineering. 534 00:29:29,440 --> 00:29:32,360 I would say, I'm sure it might come up, but almost like as a 535 00:29:32,360 --> 00:29:36,720 side effect of something else. But trust is kind of a very 536 00:29:36,720 --> 00:29:39,480 heavy psychological thing. You know, you have the person 537 00:29:39,960 --> 00:29:42,040 that trusts or, and the trustee, I guess. 538 00:29:42,040 --> 00:29:45,640 So you have to try and work out, you know, when I say I trust a 539 00:29:45,640 --> 00:29:50,440 tool, what's my belief in its ability, you know, and versus 540 00:29:50,760 --> 00:29:54,400 the other side of the equation. So, so maybe we can find ways to 541 00:29:54,400 --> 00:29:57,920 systematize that trust so that it results in a better 542 00:29:57,920 --> 00:30:03,560 calibrated trust so that we don't end up almost washing 543 00:30:03,680 --> 00:30:06,760 ethics, you know, like ethic washing, where we think we've 544 00:30:06,920 --> 00:30:10,520 trusted, it's trustworthy. But are we just overlooking 545 00:30:10,520 --> 00:30:12,960 something deeper? There will be a whole bunch of 546 00:30:12,960 --> 00:30:16,400 new jobs that arise out of all of this, like specialties, I 547 00:30:16,400 --> 00:30:17,960 suppose. But for now, I think 548 00:30:18,560 --> 00:30:23,160 generalizing your knowledge is to, to get through this near 549 00:30:23,160 --> 00:30:26,280 term phase. I would say that becoming, 550 00:30:26,520 --> 00:30:29,880 moving back towards being a bit more of a generalist, certainly 551 00:30:29,880 --> 00:30:33,760 less specialising on the libraries or the syntax, because 552 00:30:33,760 --> 00:30:36,360 the bots can do that rather well, is a good idea. 553 00:30:36,360 --> 00:30:40,000 And in future we might find new specialist roles that crop up. 554 00:30:40,600 --> 00:30:43,400 The world's our oyster. I think we're creating these 555 00:30:43,440 --> 00:30:47,080 paths, so I think that's an important aspect to consider. 556 00:30:47,880 --> 00:30:50,240 Yeah. So I think the need for speed is 557 00:30:50,680 --> 00:30:52,320 in everyone's mind these days, right? 558 00:30:52,320 --> 00:30:55,920 So they think with AI we can be more productive, we can work on 559 00:30:55,920 --> 00:30:59,360 more, we can deliver faster. And I think the temptation is 560 00:30:59,360 --> 00:31:01,440 there. So when I say like, I'm becoming 561 00:31:01,440 --> 00:31:03,720 much more hands on, right? I become a developer, right? 562 00:31:03,920 --> 00:31:07,280 So supercharged with disability, we would like to think that we 563 00:31:07,280 --> 00:31:09,480 want to be faster as well. Like, you know, churning out 564 00:31:09,840 --> 00:31:12,520 code sometimes even if the requirements is a little bit 565 00:31:12,520 --> 00:31:15,720 vague, we can speed up just by taking our assumptions and 566 00:31:15,720 --> 00:31:18,040 deliver things, at least get something done, right? 567 00:31:18,320 --> 00:31:19,600 So I think the temptation is there. 568 00:31:19,720 --> 00:31:21,520 And you mentioned about vigilance, right? 569 00:31:21,520 --> 00:31:25,840 So every time when you write a code and you ask AI so many 570 00:31:25,840 --> 00:31:29,280 times, right, I would assume that everyone will start to 571 00:31:29,280 --> 00:31:33,320 build some kind of trust simply because it becomes like a close 572 00:31:33,320 --> 00:31:36,160 friend of yours, right? And I remember back then when we 573 00:31:36,160 --> 00:31:38,880 Googled as well, right? Anything that is top ranked, 574 00:31:38,880 --> 00:31:40,680 we'll just assume that it's true, right? 575 00:31:40,680 --> 00:31:43,720 It's more accurate. So I think these days especially 576 00:31:43,720 --> 00:31:47,960 people will just easily trust AI and especially with coding, it 577 00:31:47,960 --> 00:31:50,760 becomes much more. The feedback loop is very short, 578 00:31:50,760 --> 00:31:52,120 right? When you ask question, it gives 579 00:31:52,120 --> 00:31:55,160 back and then you continue and build features on top of that. 580 00:31:55,800 --> 00:31:59,040 So I think you remind us a very good point about vigilance and 581 00:31:59,040 --> 00:32:02,880 trust, right? So try not to assume everything 582 00:32:02,880 --> 00:32:04,680 that AI gives us is actually true. 583 00:32:05,040 --> 00:32:07,200 Which comes to the next discussion that I want to bring 584 00:32:07,200 --> 00:32:09,200 up. It's about over reliance on AII 585 00:32:09,200 --> 00:32:10,880 can actually feel it myself as well. 586 00:32:11,600 --> 00:32:15,520 So every time I need to do something, be big or small, I 587 00:32:15,520 --> 00:32:19,160 would need to ask AI, you know, some as a like giving me a fresh 588 00:32:19,160 --> 00:32:22,640 pair of perspectives or maybe even criticize, summarise and 589 00:32:22,640 --> 00:32:25,080 all those things. So what do you think is the 590 00:32:25,080 --> 00:32:27,240 danger of this over reliance on AI? 591 00:32:28,160 --> 00:32:30,760 Well, that's a great question. It's one of the things I think 592 00:32:30,760 --> 00:32:34,680 about a lot because like you, I also feel it in myself like it. 593 00:32:35,040 --> 00:32:39,040 It's too appealing. You know, you've got this tool 594 00:32:39,040 --> 00:32:43,920 that is never tired. It's always available as long as 595 00:32:43,920 --> 00:32:48,480 you, you know, pay for it or the free one or and don't exhaust 596 00:32:48,480 --> 00:32:52,280 your obtaining limits. But it's just so helpful to have 597 00:32:52,280 --> 00:32:55,000 this sort of second pair of eyes there that's there as your 598 00:32:55,000 --> 00:32:58,280 helper. But I'm a firm believer that we 599 00:32:58,280 --> 00:33:01,400 learn through experience, hands on experience. 600 00:33:01,400 --> 00:33:07,080 You know, you can read, watch all the YouTube videos about 601 00:33:07,080 --> 00:33:10,720 something that you can find. But at the end of the day, if 602 00:33:10,720 --> 00:33:15,000 you don't have a variety of experiences, and I think we 603 00:33:15,000 --> 00:33:17,840 spoke about this four years ago, I think I talked about the 604 00:33:17,840 --> 00:33:21,800 Dreyfus model and how you go through from novice to expert 605 00:33:22,120 --> 00:33:25,200 over 10 years apparently of skills acquisition. 606 00:33:25,440 --> 00:33:29,560 And that skills acquisition is built through experience through 607 00:33:29,560 --> 00:33:34,240 10 years of experiencing a variety of at the different 608 00:33:34,240 --> 00:33:36,760 angles of the same skill, right? You can't just practice the 609 00:33:36,760 --> 00:33:40,840 same, the same song on a guitar and become a, you know, the best 610 00:33:40,840 --> 00:33:44,920 guitar player in on the world. So from that perspective, if you 611 00:33:44,960 --> 00:33:49,760 are outsourcing or delegating a lot of what you do to this bot, 612 00:33:49,760 --> 00:33:52,200 then you yourself are not experiencing it. 613 00:33:52,200 --> 00:33:56,360 You're not experiencing the friction, the frustration, the 614 00:33:56,360 --> 00:34:00,800 pain, the resolution that you come up with through your own 615 00:34:00,800 --> 00:34:03,080 critical thinking. You're kind of offloading that 616 00:34:03,560 --> 00:34:06,640 to something else. And what does that mean for our 617 00:34:06,640 --> 00:34:11,880 ability to gain those skills? Now, I'm really not sure because 618 00:34:12,360 --> 00:34:15,800 research will tell you that deliberate practice is indeed 619 00:34:15,800 --> 00:34:20,440 the best way to learn something. Yes, simply reviewing something 620 00:34:20,440 --> 00:34:23,360 else is you'll get a feel for it, but you're not going to have 621 00:34:23,360 --> 00:34:25,040 that confidence to go and do it yourself. 622 00:34:25,719 --> 00:34:31,679 So I am quite sure that if we keep going down the path that 623 00:34:31,679 --> 00:34:35,719 we're going, which I can't see us not, and we all become super 624 00:34:35,719 --> 00:34:40,719 reliant on this, these new tools that over time we'll all become 625 00:34:40,719 --> 00:34:43,400 quite rusty at the thing that we used to do by hand. 626 00:34:43,679 --> 00:34:45,840 And we'll become very dependent on these tools. 627 00:34:45,840 --> 00:34:49,000 And to some degree it's a bit like the calculator, you know, 628 00:34:49,000 --> 00:34:54,239 like how many of us even, to sum up a pretty simple, you know, 629 00:34:54,239 --> 00:34:57,040 additional subtraction, we pull out the calculator just in case, 630 00:34:57,040 --> 00:35:01,640 you know, or GPS, right? It's quite funny, but my husband 631 00:35:01,640 --> 00:35:06,040 is not good with directions and fully relies on GPS. 632 00:35:06,040 --> 00:35:08,960 Without GPS, he's just not confident about getting where he 633 00:35:08,960 --> 00:35:11,200 needs to go. There's plenty of other skills, 634 00:35:11,200 --> 00:35:14,240 but directions and following them, you know, has to be the 635 00:35:14,240 --> 00:35:16,320 GPS. And that's something that 636 00:35:16,320 --> 00:35:19,080 previously you would have figured out himself, you know, 637 00:35:19,080 --> 00:35:21,600 and he would have learnt the way to a place and then never 638 00:35:21,600 --> 00:35:23,360 forgotten it. But now he never needs to learn 639 00:35:23,360 --> 00:35:26,480 it because the GPS will take him there every time until his phone 640 00:35:26,480 --> 00:35:28,760 runs out of battery and then he just can't get there. 641 00:35:28,880 --> 00:35:30,880 It's OK, I'm not going there today because I can't. 642 00:35:32,160 --> 00:35:34,360 You solve the right. In that case, you would solve 643 00:35:34,360 --> 00:35:37,520 the problem of putting giving your battery a bit more juice so 644 00:35:37,520 --> 00:35:39,840 that you can then use GPS to then get to the place that 645 00:35:39,840 --> 00:35:43,040 you're going to, right? So I wonder if at some point 646 00:35:43,040 --> 00:35:45,880 we're all going to become so dependent on these tools, not 647 00:35:45,880 --> 00:35:49,120 just for coding, but for a lot of aspects of our lives that 648 00:35:49,440 --> 00:35:51,680 we're going to have this like virtual assistant in our back 649 00:35:51,680 --> 00:35:55,720 pocket that is just always on and ready to help us. 650 00:35:56,320 --> 00:36:00,800 And I do wonder if we might head to a place where we each have 651 00:36:00,800 --> 00:36:04,920 our own AI that knows us intimately, you know, and that 652 00:36:04,920 --> 00:36:09,200 we can kind of bring to work almost, you know, so we're 653 00:36:09,200 --> 00:36:12,520 already seeing it to some degree with the GitHub copilot. 654 00:36:12,520 --> 00:36:15,560 Like it's allows you to connect to a number of different models, 655 00:36:15,920 --> 00:36:19,400 but what if it could also connect to a model that is 656 00:36:19,560 --> 00:36:23,000 yours, is your personal one that knows the way that you think, 657 00:36:23,400 --> 00:36:27,280 your interests, you know, the way that you tend to think about 658 00:36:27,280 --> 00:36:33,320 things and is more tuned to you. I suspect Open AI and Sam 659 00:36:33,320 --> 00:36:36,120 Altman, they're going in that direction from what I heard in 660 00:36:36,120 --> 00:36:37,640 the video that I watched yesterday. 661 00:36:37,640 --> 00:36:40,880 Like he really sees a world in which everyone has their own 662 00:36:40,880 --> 00:36:47,600 personalized ChatGPT in terms of raw skills, maybe the skill that 663 00:36:47,600 --> 00:36:50,600 we'll learn really. And actually that's a good point 664 00:36:50,600 --> 00:36:54,400 because there's the World Economic Forum produced a report 665 00:36:54,400 --> 00:36:59,520 on the future of jobs. And in that it's like a 290 page 666 00:37:00,080 --> 00:37:02,760 PDF. So without getting an AI to 667 00:37:02,760 --> 00:37:06,640 summarize it for me, I don't know if we're actually sit down 668 00:37:06,640 --> 00:37:07,920 and read the whole thing into it. 669 00:37:08,320 --> 00:37:11,080 I don't have enough time. But there's some really nice 670 00:37:11,280 --> 00:37:15,000 visualizations and one in particular, they show the skills 671 00:37:15,000 --> 00:37:18,240 that are becoming more important, that are already 672 00:37:18,240 --> 00:37:21,480 important and they project will become more important are things 673 00:37:21,480 --> 00:37:27,480 like resilience, agility, flexibility, you know, basically 674 00:37:27,480 --> 00:37:31,160 embracing change. These not skills that we really 675 00:37:31,320 --> 00:37:34,200 learn as a subject at school or at university, you know, they're 676 00:37:34,200 --> 00:37:38,040 kind of just life skills that you build up as you mature. 677 00:37:38,040 --> 00:37:42,720 So how, how do we teach people these things earlier on in their 678 00:37:42,720 --> 00:37:46,520 lives? How do we kind of leapfrog 679 00:37:46,920 --> 00:37:51,520 experience and all of that that life gives you naturally and 680 00:37:51,520 --> 00:37:54,800 sort of try and instill that type of thinking in the next 681 00:37:54,800 --> 00:37:57,640 generation? There's some really big problems 682 00:37:57,640 --> 00:38:02,960 to solve, you know, to fully understand how we can leverage 683 00:38:02,960 --> 00:38:07,640 these tools without giving away everything that, you know, as 684 00:38:07,640 --> 00:38:11,640 humans, we've been used to doing ourselves, which is what makes 685 00:38:11,640 --> 00:38:14,600 this very exciting and scary at the same time. 686 00:38:15,760 --> 00:38:18,040 Yeah. So I think, yeah, open AI is 687 00:38:18,360 --> 00:38:21,080 releasing this thing called memories probably right in maybe 688 00:38:21,080 --> 00:38:24,400 in your ChatGPT. And also the trend seems to be 689 00:38:24,400 --> 00:38:27,920 like the larger and the larger context that AI model is able to 690 00:38:27,920 --> 00:38:29,960 use, right? Yeah, I think maybe one day, 691 00:38:29,960 --> 00:38:33,840 like what you said, we can have a personal clone of us maybe 692 00:38:34,120 --> 00:38:37,240 that can think of feel just like us feel is probably a bit too 693 00:38:37,240 --> 00:38:39,280 far. But I guess the thinking, you 694 00:38:39,280 --> 00:38:41,960 know, the kind of rationale, the kind of styles that we use in 695 00:38:41,960 --> 00:38:45,160 terms of our output, right? Bit writing code, it's something 696 00:38:45,160 --> 00:38:48,160 that AI can learn as well. So definitely it's something 697 00:38:48,600 --> 00:38:50,920 maybe exciting but also scary at the same time. 698 00:38:51,320 --> 00:38:55,600 And you mentioned that we human would love to learn by doing, 699 00:38:55,720 --> 00:38:57,640 you know, like the experiential aspect, right. 700 00:38:57,920 --> 00:38:59,920 And I think also software engineering is like a 701 00:38:59,920 --> 00:39:03,120 craftsmanship thing, right? You cannot just, you know, learn 702 00:39:03,120 --> 00:39:05,680 software by reading book type hello world and you know, you 703 00:39:05,680 --> 00:39:08,040 become a software engineer. So there's so many different 704 00:39:08,040 --> 00:39:11,000 aspects like design, you know, maybe non functional 705 00:39:11,000 --> 00:39:14,240 requirements, maybe be it the deployment network, things like 706 00:39:14,240 --> 00:39:16,560 that. So in your blocks, actually you 707 00:39:16,560 --> 00:39:20,440 mentioned about this trend of software engineering or software 708 00:39:20,440 --> 00:39:24,280 engineer now has this kind of identity crisis this day, right? 709 00:39:24,280 --> 00:39:26,640 Like what you mentioned, right? So many things now is available 710 00:39:26,640 --> 00:39:29,440 to us. Maybe our roles have become much 711 00:39:29,440 --> 00:39:32,520 more, I would say changing a lot, like demanding. 712 00:39:32,520 --> 00:39:36,120 At the same time, everyone is scared of their job because the 713 00:39:36,120 --> 00:39:38,720 news always say, OK, we can reduce the number of of 714 00:39:38,720 --> 00:39:41,200 engineers. Junior engineers will not be 715 00:39:41,200 --> 00:39:44,080 needed anymore. So tell us about this identity 716 00:39:44,080 --> 00:39:46,000 crisis. I'm sure everyone is feeling it. 717 00:39:46,000 --> 00:39:48,960 I'm feeling it myself as well. What is your take about this 718 00:39:48,960 --> 00:39:51,640 identity crisis? Well, it's funny you should 719 00:39:51,640 --> 00:39:53,480 mention that particular blog post. 720 00:39:53,480 --> 00:39:57,960 It's the the last one that I published and it's by far the 721 00:39:57,960 --> 00:40:02,720 most popular blog post I have ever, ever written or even could 722 00:40:02,720 --> 00:40:04,640 have conceivably imagined writing. 723 00:40:05,040 --> 00:40:08,760 Just so you know, to be completely transparent, I've had 724 00:40:08,760 --> 00:40:11,320 a blog in the past which went nowhere. 725 00:40:11,320 --> 00:40:12,800 I was just yelling into the wind. 726 00:40:12,800 --> 00:40:16,080 Nobody cared. And I started off a blog again a 727 00:40:16,320 --> 00:40:19,360 couple of years ago, maybe a year and a half ago, figured, Oh 728 00:40:19,360 --> 00:40:22,480 well, at least it'll get me writing. 729 00:40:22,480 --> 00:40:24,480 You know, I have a lot of thoughts about things and I've 730 00:40:25,040 --> 00:40:28,200 aside from. Filling my poor husband's ears 731 00:40:28,240 --> 00:40:31,040 with all of my thoughts. I thought maybe I can just pour 732 00:40:31,040 --> 00:40:32,800 it into my writing and put it in. 733 00:40:33,160 --> 00:40:35,600 But for the most part, my blog doesn't get all that much 734 00:40:35,600 --> 00:40:37,360 traffic. But this particular blog post 735 00:40:37,360 --> 00:40:41,480 I've had, I think at the last count, 43,000 views of it in 736 00:40:41,480 --> 00:40:44,840 less than a month. And yeah, I couldn't have 737 00:40:44,840 --> 00:40:47,520 imagined such a response. And also people reaching out to 738 00:40:47,520 --> 00:40:50,720 me about it as well. They're wanting to either thank 739 00:40:50,720 --> 00:40:55,520 me for writing it because it resonated for them or, you know, 740 00:40:55,680 --> 00:40:59,760 sharing it with their own, their own networks because it meant 741 00:40:59,760 --> 00:41:01,880 something to them. And other people are commenting 742 00:41:01,880 --> 00:41:03,960 on it. So it's clearly hit a nerve, 743 00:41:04,200 --> 00:41:07,640 which is quite interesting. But what does that mean? 744 00:41:08,720 --> 00:41:14,000 See, I wrote that from, it was a very heartfelt blog post from my 745 00:41:14,000 --> 00:41:17,240 perspective because like I said, I started coding when, well, 746 00:41:17,240 --> 00:41:19,960 started coding. I started playing with code when 747 00:41:19,960 --> 00:41:22,800 I was six years old, and it was BASIC at the time, right? 748 00:41:23,360 --> 00:41:28,360 But the feeling that I got when I saw my work turn into 749 00:41:28,360 --> 00:41:32,960 something on the computer was probably, you know, it was 750 00:41:32,960 --> 00:41:36,400 certainly at that age, the best feeling I'd ever had, you know, 751 00:41:37,000 --> 00:41:40,120 to see this. It was a smiling bounce a :) 752 00:41:40,120 --> 00:41:41,680 that bounced around the screen, a Sprite. 753 00:41:42,160 --> 00:41:44,400 And it was just one of the examples in the BASIC manual 754 00:41:44,400 --> 00:41:48,240 that came with the Commodore 64. And ever since then, you know, 755 00:41:48,240 --> 00:41:51,320 I've really held onto that. The craft itself, the knowledge 756 00:41:51,320 --> 00:41:55,200 that you've written some words that have turned into this cool 757 00:41:55,200 --> 00:41:57,960 thing. It's like it's more than just 758 00:41:57,960 --> 00:42:00,920 solving a puzzle, right? And there's a beauty behind it. 759 00:42:01,440 --> 00:42:04,000 And I think a lot of software engineers feel that way about 760 00:42:04,000 --> 00:42:06,280 their code. There's a reason that things 761 00:42:06,280 --> 00:42:11,840 like lead code websites exist because we like to solve 762 00:42:12,240 --> 00:42:16,440 existing problems in new and imaginative ways using a 763 00:42:16,440 --> 00:42:19,640 slightly different technique that is that much faster, more 764 00:42:19,640 --> 00:42:21,360 efficient or more memory efficient. 765 00:42:21,360 --> 00:42:25,640 Or, you know, and we like to you know, some of us might even, we 766 00:42:25,640 --> 00:42:27,760 certainly feel some a level of pride. 767 00:42:27,800 --> 00:42:30,440 We might boast about our, you know, being top of the 768 00:42:30,440 --> 00:42:35,280 leaderboard on leet code. I wish like I've never been good 769 00:42:35,280 --> 00:42:37,240 enough to make it anywhere near that. 770 00:42:37,480 --> 00:42:40,800 But, you know, a lot of software engineers tie their at least 771 00:42:40,800 --> 00:42:44,120 their professional identity into their work and into the quality 772 00:42:44,120 --> 00:42:45,440 of the code that they're producing. 773 00:42:46,320 --> 00:42:52,040 And so if that is now kind of not the point because these 774 00:42:52,040 --> 00:42:53,600 tools can generate the code for you. 775 00:42:53,600 --> 00:42:57,400 So really the point becomes more about solving the customer's 776 00:42:57,400 --> 00:43:00,240 problem. But as we spoke about earlier, 777 00:43:00,720 --> 00:43:03,760 over the last 10-15 years, you know, we've even had to 778 00:43:03,760 --> 00:43:08,920 introduce new titles like product engineer to refer to a 779 00:43:08,960 --> 00:43:12,320 software engineer who actually cares about the whole problem 780 00:43:12,320 --> 00:43:14,640 that's being solved, the product that's being built and the 781 00:43:14,640 --> 00:43:17,920 customer at the end of the value chain that, you know, actually 782 00:43:18,120 --> 00:43:20,280 makes use of the solution that you're building. 783 00:43:20,920 --> 00:43:25,680 Because somehow we've lost the sight of perhaps why we were 784 00:43:25,680 --> 00:43:27,720 building that code. Because it just the act of 785 00:43:27,720 --> 00:43:30,160 building the coding and using different libraries and 786 00:43:30,160 --> 00:43:33,800 frameworks and new and interesting ways became, you 787 00:43:33,800 --> 00:43:35,760 know, so interesting for so many people. 788 00:43:35,760 --> 00:43:41,760 So in that regard, how do we take that passion and apply it a 789 00:43:42,200 --> 00:43:45,920 little bit more broadly? Some people might still be able 790 00:43:45,920 --> 00:43:51,200 to spend their energy and their passion building things at the 791 00:43:51,200 --> 00:43:53,840 code level, but maybe even the type of things they build might 792 00:43:53,840 --> 00:43:56,800 change, you know? So maybe if that's the area that 793 00:43:56,800 --> 00:43:59,960 really brings you the most joy, then maybe start looking at 794 00:43:59,960 --> 00:44:03,360 things like Agent Development Kit and these new frameworks 795 00:44:03,360 --> 00:44:06,800 that are coming up that allow you to build solutions that have 796 00:44:07,040 --> 00:44:10,000 AI embedded in them. You know, so there's, we're 797 00:44:10,000 --> 00:44:13,760 still forging new paths around that like MCP great. 798 00:44:13,760 --> 00:44:17,040 But you know, already we're finding some new vulnerabilities 799 00:44:17,040 --> 00:44:19,440 that might creep in. So, you know, get good at 800 00:44:19,440 --> 00:44:22,120 building the tools that protect us from things like that. 801 00:44:22,400 --> 00:44:28,040 There's going to have to be some sort of AI version of SRE, you 802 00:44:28,040 --> 00:44:31,320 know, so when you have AI embedded in your systems, then 803 00:44:31,320 --> 00:44:35,000 how do you do observability and monitoring and resilience and 804 00:44:35,000 --> 00:44:37,000 all of that? So, you know, there will be 805 00:44:37,000 --> 00:44:40,480 things that are maybe closer to the what we would consider 806 00:44:40,480 --> 00:44:45,320 coding today in future for these new solutions we might build. 807 00:44:45,800 --> 00:44:48,760 But for anybody who was already at a point in their career where 808 00:44:48,760 --> 00:44:51,160 they were thinking maybe it's time to consider a management 809 00:44:51,160 --> 00:44:53,880 role because you know, I don't find coding the hard part 810 00:44:53,880 --> 00:44:56,960 anymore. You know, maybe their start 811 00:44:56,960 --> 00:45:00,480 thinking about how best to leverage these tools, how to 812 00:45:00,480 --> 00:45:04,720 manage that calibrated trust and automate some of that eval 813 00:45:04,800 --> 00:45:10,480 verification. How do you layer the kind of a 814 00:45:10,560 --> 00:45:16,720 distributed cognitive support where you end up with layers of 815 00:45:16,840 --> 00:45:20,080 AI agents that check each other's homework? 816 00:45:20,520 --> 00:45:23,680 You know, and I think we see that already in some tools and 817 00:45:23,680 --> 00:45:27,880 that's likely, and Sam Altman said this in his talk that I 818 00:45:28,040 --> 00:45:31,320 listened to yesterday as well. Like he's, we can see that the 819 00:45:31,320 --> 00:45:35,360 current version of LLM's have already revolutionize software 820 00:45:35,360 --> 00:45:38,640 engineering in particular and starting to revolutionize many 821 00:45:38,640 --> 00:45:40,520 other things, but software engineering for sure. 822 00:45:40,800 --> 00:45:43,560 The next thing he's very interested in is agentic AI. 823 00:45:43,560 --> 00:45:47,400 And I completely agree. You know, so LLMS are kind of 824 00:45:47,400 --> 00:45:51,120 like 1 component. That's fine and we're all using 825 00:45:51,120 --> 00:45:54,760 them in our own different ways. But the power will come when you 826 00:45:55,240 --> 00:45:59,720 create a system that is made-up of many components, some of 827 00:45:59,720 --> 00:46:04,600 which might be LLMS and then potentially controlled by an 828 00:46:04,600 --> 00:46:06,760 orchestrator that is an LLM itself. 829 00:46:06,760 --> 00:46:10,520 And that's where you get that kind of extreme non determinism 830 00:46:10,520 --> 00:46:14,240 because you just give the system a task and it figures out for 831 00:46:14,240 --> 00:46:17,760 itself which agent it's going to call, which tool it's going to 832 00:46:17,760 --> 00:46:20,320 use. It may just not go down the path 833 00:46:20,320 --> 00:46:23,120 you expected at all. So those will be very 834 00:46:23,120 --> 00:46:29,080 interesting systems to create, to test, to maintain, to reason 835 00:46:29,080 --> 00:46:32,120 about, you know, so there will be work in that area. 836 00:46:32,120 --> 00:46:36,240 But so that's, yeah, becoming slightly more generalist, 837 00:46:36,720 --> 00:46:40,040 becomes really good at using these tools, building these 838 00:46:40,040 --> 00:46:43,760 tools and also managing these tools, wrangling these tools. 839 00:46:43,760 --> 00:46:46,360 Like how do you fix the system when it stops working? 840 00:46:46,360 --> 00:46:50,600 How do you improve it? These will become some new 841 00:46:50,600 --> 00:46:54,800 interesting jobs in the future. But what do we do about our 842 00:46:54,800 --> 00:46:59,840 junior engineers like the me of, you know, 20-30 years ago who 843 00:46:59,840 --> 00:47:01,360 wanted to be a software engineer? 844 00:47:02,080 --> 00:47:04,080 That's a tough one. It's going to be hard, I think, 845 00:47:04,080 --> 00:47:08,080 for some to break into the industry harder than it was, 846 00:47:08,080 --> 00:47:10,440 say, three years ago. Yeah. 847 00:47:10,440 --> 00:47:14,200 So I think everyone is feeling this kind of identity crisis in 848 00:47:14,200 --> 00:47:16,280 their role, right? Funny enough, probably it hits a 849 00:47:16,280 --> 00:47:19,000 lot of nurse simply because we are feeling it, especially for 850 00:47:19,000 --> 00:47:22,200 people who are using it. But for those people who are not 851 00:47:22,200 --> 00:47:25,640 using it, maybe they don't know yet, but I think it's very, very 852 00:47:25,640 --> 00:47:28,360 concerning if they don't start picking up, you know, at least 853 00:47:28,360 --> 00:47:30,640 understanding what these two is capable of, right. 854 00:47:31,120 --> 00:47:34,720 And I think you mentioned about, you know, juniors, I think the 855 00:47:34,720 --> 00:47:37,440 other day I went into like a meet up, you know, small meet up 856 00:47:37,440 --> 00:47:40,960 within, you know, my area, some managers, some leaders even 857 00:47:40,960 --> 00:47:45,600 saying that, oh, I've tried vibe coding on certain aspect and it 858 00:47:45,600 --> 00:47:47,880 seemed to be doing it's job. And it's cheap. 859 00:47:48,000 --> 00:47:52,120 You just spend a few, you know, maybe 100 of bucks a month just 860 00:47:52,120 --> 00:47:55,160 to use these tools. And compared to hiring a junior, 861 00:47:55,160 --> 00:47:58,640 maybe a fresh grad who just graduated, the economy seems 862 00:47:58,640 --> 00:48:01,440 like very, very, you know, lopsided, right? 863 00:48:01,440 --> 00:48:04,560 So it seems like leaders probably will start to think 864 00:48:04,560 --> 00:48:06,800 more, OK, how can we leverage AI more? 865 00:48:06,800 --> 00:48:09,640 Just like what I mentioned about a Shopify experience, right? 866 00:48:10,120 --> 00:48:13,320 So I think for juniors out there, maybe looking back at 867 00:48:13,320 --> 00:48:17,080 you, you know, maybe 30 years back, what would you say to 868 00:48:17,080 --> 00:48:18,280 them? Is software engineering still a 869 00:48:18,280 --> 00:48:21,240 good career for them? Because it just to breakthrough 870 00:48:21,240 --> 00:48:24,440 and get a first job probably is a little bit tough and 871 00:48:24,440 --> 00:48:27,400 especially now in the very difficult situations globally, 872 00:48:27,400 --> 00:48:30,640 you know, with all this happening in the US and some 873 00:48:30,640 --> 00:48:33,040 parts of the world. So is there anything that you 874 00:48:33,040 --> 00:48:37,000 would want to say maybe to the younger junior engineers of US? 875 00:48:37,800 --> 00:48:42,240 I reckon if it's a path that you, you know, I would encourage 876 00:48:42,240 --> 00:48:46,800 anyone to follow their dreams. Life is there to be lived. 877 00:48:47,400 --> 00:48:50,160 Don't choose a career that you don't think you're going to 878 00:48:50,160 --> 00:48:54,120 enjoy because you're going to spend a lot of your life doing 879 00:48:54,120 --> 00:48:55,920 it. So make sure that whatever it is 880 00:48:55,920 --> 00:48:59,080 that you pick, you know that it's something that you enjoy 881 00:48:59,080 --> 00:49:01,400 doing. And if software engineering is 882 00:49:01,400 --> 00:49:05,480 the path that you had your mind set on, and by all means follow 883 00:49:05,480 --> 00:49:11,240 your heart, but understand that, say, talking to somebody who is 884 00:49:11,240 --> 00:49:14,040 five years into their career, like you won't be able to follow 885 00:49:14,040 --> 00:49:16,400 in their same footsteps. I think it's just going to be 886 00:49:16,400 --> 00:49:19,720 different. So I would encourage them to 887 00:49:20,200 --> 00:49:24,520 simultaneously try and educate themselves on all of the 888 00:49:24,520 --> 00:49:26,360 fundamentals. You know, I'm pretty sure that 889 00:49:26,360 --> 00:49:29,960 universities are still teaching all of that, the basics around 890 00:49:30,520 --> 00:49:35,400 SOLID principles and yeah, like naming conventions and 891 00:49:35,600 --> 00:49:38,640 algorithms and sorting algorithms and all of that. 892 00:49:39,480 --> 00:49:42,160 Familiarize yourself with that, but then get really good at 893 00:49:42,160 --> 00:49:47,040 using these tools to generate code that reflects that 894 00:49:47,040 --> 00:49:50,440 foundational learning. Because I think if you showed up 895 00:49:50,520 --> 00:49:53,120 to a job interview and you said, sure, I'm a junior, I, you know, 896 00:49:53,120 --> 00:49:57,640 I haven't had much experience or really any at writing code, but 897 00:49:57,640 --> 00:50:02,560 I am a master at using all of these other tools, cursor and 898 00:50:02,560 --> 00:50:04,640 lovable, and this tool's good for that. 899 00:50:04,640 --> 00:50:06,880 And then I port it over here and then I open it here. 900 00:50:06,880 --> 00:50:09,560 And then I, you know, I look at it through this and I know it's 901 00:50:09,560 --> 00:50:13,080 important to verify because these are the sorts of mistakes 902 00:50:13,080 --> 00:50:15,640 it tends to make. And, you know, with all of that 903 00:50:15,640 --> 00:50:18,800 knowledge, probably find yourself being quite in demand 904 00:50:18,800 --> 00:50:20,760 actually. And, you know, maybe they're in 905 00:50:20,760 --> 00:50:24,640 a perfect position to really adopt these new technologies at 906 00:50:24,640 --> 00:50:27,320 a time that they're coming in, you know, those further on in 907 00:50:27,320 --> 00:50:29,520 their career. I have a feeling that in my 908 00:50:29,520 --> 00:50:33,920 research, what I'm seeing is people who are well into their 909 00:50:33,920 --> 00:50:37,280 career, they might not have been writing that much code anymore 910 00:50:37,280 --> 00:50:39,240 anyway. So to them, it's kind of like, 911 00:50:39,240 --> 00:50:41,080 oh, yeah, it's nice. But, you know, I might just use 912 00:50:41,080 --> 00:50:43,760 it to play around with a bit. So they almost feel like it's 913 00:50:43,920 --> 00:50:47,840 not that big a deal. But the people kind of part way 914 00:50:47,840 --> 00:50:50,560 through their career, like Mid Korea, I think they're 915 00:50:50,560 --> 00:50:53,440 struggling with it, right? Because they're sort of, OK, So 916 00:50:53,440 --> 00:50:55,800 do I now need to become really good at using this thing? 917 00:50:55,800 --> 00:50:59,080 Like what I've known to be true over the last 10 years is now 918 00:50:59,080 --> 00:51:01,240 changing? That's probably harder than 919 00:51:01,240 --> 00:51:05,040 somebody who's entering the market at this point where, you 920 00:51:05,040 --> 00:51:06,480 know, where the tools are already. 921 00:51:06,760 --> 00:51:08,720 They've landed. They're here to stay. 922 00:51:09,040 --> 00:51:10,280 They're only going to get better. 923 00:51:10,760 --> 00:51:14,320 And the types of systems we're building are changing, how we 924 00:51:14,320 --> 00:51:16,600 build them are changing. So it's done settling. 925 00:51:16,760 --> 00:51:19,520 But they, you know, juniors probably have a unique 926 00:51:19,520 --> 00:51:23,800 opportunity to help create the roles that they want, but 927 00:51:23,800 --> 00:51:26,160 they're probably going to need to have a lot of imagination. 928 00:51:26,400 --> 00:51:29,000 You know, I think that the moment, if you can imagine it, 929 00:51:29,120 --> 00:51:32,800 you can create it. So imagination becomes a really 930 00:51:32,800 --> 00:51:35,240 important skill. But again, we probably don't 931 00:51:35,240 --> 00:51:36,840 teach it at school all that much. 932 00:51:36,840 --> 00:51:39,720 I don't know, I, I never felt like I had much imagination, but 933 00:51:40,280 --> 00:51:43,400 these tools help you be a little bit more, you know, imaginative 934 00:51:43,400 --> 00:51:44,840 if you're struggling in that domain. 935 00:51:45,360 --> 00:51:48,160 But yeah, imagination, creativity, you know, the 936 00:51:48,160 --> 00:51:51,360 confidence to try it, use a tool to help you have the confidence 937 00:51:51,360 --> 00:51:54,320 to, you know, it's very circular in a way like that. 938 00:51:54,320 --> 00:51:58,880 So I wouldn't say all hope is lost for junior engineers, just 939 00:51:58,880 --> 00:52:01,240 that the type of work they're going to be doing will be 940 00:52:01,240 --> 00:52:03,120 different. So they probably won't have as 941 00:52:03,120 --> 00:52:07,200 much guidance as perhaps the last 10-15 years worth of new 942 00:52:07,200 --> 00:52:11,000 engineers have had. But yeah, as leaders, we need to 943 00:52:11,000 --> 00:52:14,480 be able to systematize what that looks like as well, like 944 00:52:14,480 --> 00:52:17,360 potentially change our career ladders and the sorts of skills 945 00:52:17,360 --> 00:52:20,440 we expect people to build up and how we measure their 946 00:52:20,680 --> 00:52:23,120 performance. You know, maybe it becomes a 947 00:52:23,520 --> 00:52:26,640 measurement of how well are you using these tools, which is 948 00:52:27,240 --> 00:52:31,040 quite different to how well do you understand the SOLID 949 00:52:31,040 --> 00:52:34,680 principles, you know? Yeah, we all have to adjust. 950 00:52:35,600 --> 00:52:38,480 Yeah, very interesting that you mentioned about, you know, the 951 00:52:38,480 --> 00:52:42,360 juniors now have the time to actually learn these tools, you 952 00:52:42,360 --> 00:52:45,080 know, from the get go, right. So I was having another 953 00:52:45,080 --> 00:52:46,720 conversation, another episode as well. 954 00:52:46,880 --> 00:52:49,760 We have a term about it like AI native, you know, AI native 955 00:52:49,760 --> 00:52:52,280 developers just like, you know, social media aspect, right? 956 00:52:52,280 --> 00:52:55,240 The youngsters these days are much more savvy than maybe I 957 00:52:55,240 --> 00:52:57,840 would say me, you know, using social media. 958 00:52:57,840 --> 00:53:01,040 They have even different interaction patterns and maybe 959 00:53:01,040 --> 00:53:03,480 the way to leverage social media. 960 00:53:03,800 --> 00:53:06,640 So some of us maybe it's generational thing, right? 961 00:53:06,640 --> 00:53:10,200 Like we were not trained using that and now we have to adapt in 962 00:53:10,200 --> 00:53:12,400 the middle of what we used to believe. 963 00:53:12,760 --> 00:53:15,400 And now adopting these new techniques, these new tools, 964 00:53:15,720 --> 00:53:18,440 they seem a little bit icky for some of us there because, like, 965 00:53:18,440 --> 00:53:19,880 what do you mean? I used to do this. 966 00:53:19,880 --> 00:53:23,280 Now I have to change, you know, all my habits, all my techniques 967 00:53:23,280 --> 00:53:25,000 and all that. Yeah. 968 00:53:25,280 --> 00:53:27,160 So I think juniors have this opportunity. 969 00:53:27,160 --> 00:53:29,680 I would say that they can learn this tool natively, you know, 970 00:53:29,680 --> 00:53:33,560 like the first thing that they can leverage on in their job. 971 00:53:34,120 --> 00:53:37,320 So you mentioned about the aspects of management as well, 972 00:53:37,320 --> 00:53:39,240 right. And management maybe needs to 973 00:53:39,280 --> 00:53:42,280 tweak their, I don't know, performance appraisal, maybe 974 00:53:42,280 --> 00:53:45,920 hiring, maybe giving tasks to, you know, their team. 975 00:53:46,480 --> 00:53:49,560 What do you see will change a lot by for managers? 976 00:53:49,560 --> 00:53:53,080 And what's the danger? And now that everyone is using 977 00:53:53,080 --> 00:53:56,000 AI, right, as a manager, at the end of the day, you're kind of 978 00:53:56,000 --> 00:53:58,520 like responsible for the delivery of your team, you know, 979 00:53:58,520 --> 00:54:01,240 the organization. What aspects do you think 980 00:54:01,520 --> 00:54:03,720 management will need to consider these days? 981 00:54:04,680 --> 00:54:12,040 I think the most important aspect is not to focus on at 982 00:54:12,040 --> 00:54:15,000 that speed, right? Like, as if that's all we care 983 00:54:15,000 --> 00:54:19,640 about is are we moving faster, Then you'll probably find that 984 00:54:19,640 --> 00:54:21,680 you have a big price to pay later on. 985 00:54:21,680 --> 00:54:25,280 You know, like, it will hit you in the face at a later stage 986 00:54:26,000 --> 00:54:30,800 when you're not prepared for it. So really I would say leaders 987 00:54:30,800 --> 00:54:34,120 today should be thinking about how to leverage these tools for 988 00:54:34,400 --> 00:54:38,280 optimizing quality as much as possible, as counterintuitive as 989 00:54:38,280 --> 00:54:40,840 that feels, because the speed is so seductive, right? 990 00:54:41,600 --> 00:54:44,720 Use it to leverage the quality of the systems that you're 991 00:54:44,720 --> 00:54:48,520 building because that will in turn create sustainable 992 00:54:48,520 --> 00:54:50,560 productivity gains into the future, right? 993 00:54:51,080 --> 00:54:55,120 That's probably somewhere where engineering managers need to 994 00:54:55,120 --> 00:54:59,480 really reflect on how they perceive good performance, 995 00:54:59,480 --> 00:55:01,920 right? How do you even measure building 996 00:55:02,000 --> 00:55:04,320 high quality systems? Well, we have some tools 997 00:55:04,320 --> 00:55:07,240 available like Sonar cube can help you with, you know, it's 998 00:55:07,440 --> 00:55:12,280 was that latest PR increasing or decreasing Cyclomatic complexity 999 00:55:12,280 --> 00:55:15,160 or you know, code smells or things like that. 1000 00:55:15,160 --> 00:55:18,840 But I would urge people to look at that angle of it more than 1001 00:55:18,840 --> 00:55:22,320 ever because of the speed at which we can erode. 1002 00:55:22,960 --> 00:55:26,000 Like how these reports that we see Dora and the get clear one 1003 00:55:26,000 --> 00:55:29,480 are pretty clear indicators that things are not moving in the 1004 00:55:29,480 --> 00:55:32,240 right direction in that sense. And I don't know if we keep 1005 00:55:32,240 --> 00:55:34,600 going down the path that we're going, we're probably just going 1006 00:55:34,600 --> 00:55:38,920 to create the next phase will be a clean up phase where we pay 1007 00:55:38,920 --> 00:55:43,480 the price of the speed gains that we all experimented with in 1008 00:55:43,480 --> 00:55:47,920 2024 or 2025, you know. But if we want to avoid doing 1009 00:55:47,920 --> 00:55:51,920 that, then I think we can already maybe start to give a 1010 00:55:51,920 --> 00:55:56,080 little bit more attention to the quality aspect of things. 1011 00:55:56,640 --> 00:56:01,360 And then through that, I suppose is where you shift the needle on 1012 00:56:01,360 --> 00:56:05,080 what it is that you're expecting your teams to really focus on. 1013 00:56:05,560 --> 00:56:07,520 And it has to be adaptive as well, right? 1014 00:56:07,520 --> 00:56:11,240 Because these tools are evolving so much that maybe today the 1015 00:56:11,240 --> 00:56:16,120 sort of code that it generates is hallucinates and qualities a 1016 00:56:16,200 --> 00:56:20,120 bit random. But maybe in a year's time, a 1017 00:56:20,120 --> 00:56:23,000 lot of that will be resolved. And actually what what we then 1018 00:56:23,000 --> 00:56:27,280 need to focus on is how do these LLMS reason about distributed 1019 00:56:27,280 --> 00:56:28,560 systems? You know, because right now 1020 00:56:28,560 --> 00:56:32,040 they're pretty good at staying within the one repository, you 1021 00:56:32,040 --> 00:56:34,120 know, and they can reason about the logic and the workspace 1022 00:56:34,120 --> 00:56:37,280 there, some more than others. But you know, most of our 1023 00:56:37,280 --> 00:56:40,360 systems are far bigger than just one repo these days, you know. 1024 00:56:40,360 --> 00:56:43,240 So how do you reason about bigger systems? 1025 00:56:43,240 --> 00:56:47,960 That's where your human software engineer has an edge today. 1026 00:56:47,960 --> 00:56:52,320 So being able to reason about systems like that and frame 1027 00:56:52,320 --> 00:56:55,640 problems correctly, those are the sorts of skills that 1028 00:56:56,200 --> 00:56:59,240 software engineers, you know, those who can do that well will 1029 00:56:59,240 --> 00:57:02,400 be at an advantage. And therefore, maybe that's the 1030 00:57:02,400 --> 00:57:05,840 type of the direction that engineering managers need to be 1031 00:57:05,840 --> 00:57:09,280 meeting their teams towards. Because as you say, like every 1032 00:57:09,280 --> 00:57:12,720 individual is still very much responsible for the code they 1033 00:57:12,720 --> 00:57:15,120 commit. Irrespective of whether the bot 1034 00:57:15,120 --> 00:57:18,600 wrote it or you wrote it or you kind of collaborated on it, at 1035 00:57:18,600 --> 00:57:19,760 the end of the day, you're committing it. 1036 00:57:19,760 --> 00:57:23,400 Your name is behind that, at least for now, till we have 1037 00:57:23,400 --> 00:57:27,120 fully automated agents that are spinning off and writing code 1038 00:57:27,120 --> 00:57:29,920 for us while we sleep. But for now, it's still your 1039 00:57:29,920 --> 00:57:31,360 code, right? Which means it's your 1040 00:57:31,360 --> 00:57:33,320 responsibility. And then it's your manager's 1041 00:57:33,320 --> 00:57:36,800 responsibility. And so, yeah, work back from OK, 1042 00:57:36,800 --> 00:57:38,840 if that's still my responsibility, what needs to be 1043 00:57:38,840 --> 00:57:42,200 true for me to be proud of that work and for it to be good 1044 00:57:42,200 --> 00:57:46,000 quality and focus on the skills that would help people move in 1045 00:57:46,000 --> 00:57:48,720 that direction. So I think a very good reminder 1046 00:57:48,720 --> 00:57:51,720 you just said, right, as tempting as it is, right, don't 1047 00:57:51,880 --> 00:57:54,680 focus solely on speed and productivity, right, Because 1048 00:57:54,680 --> 00:57:57,600 everyone seems to be still focusing a lot of these aspects, 1049 00:57:57,600 --> 00:57:59,400 right? Productivity gain, you know, 1050 00:57:59,480 --> 00:58:02,760 less the number of people, speed off delivery and things like 1051 00:58:02,760 --> 00:58:04,360 that. So I think there will be time 1052 00:58:04,360 --> 00:58:06,840 probably where this becomes unsustainable, right? 1053 00:58:07,200 --> 00:58:09,720 Maybe your software becomes unmaintainable. 1054 00:58:09,720 --> 00:58:13,040 Nobody can reason about simply because nobody has that 1055 00:58:13,040 --> 00:58:17,080 cognitive load to actually understand what AI is suggesting 1056 00:58:17,080 --> 00:58:19,480 to your code, especially if nobody's reviewing it properly, 1057 00:58:19,480 --> 00:58:21,040 right? So I think there will come a 1058 00:58:21,040 --> 00:58:24,800 time that probably, you know, your software is unmaintainable, 1059 00:58:24,800 --> 00:58:26,360 right? And maybe you can try to 1060 00:58:26,360 --> 00:58:29,280 leverage AI more, but I don't know where it will go to, right? 1061 00:58:29,600 --> 00:58:32,640 And putting in guard rails like what you mentioned, I think this 1062 00:58:32,640 --> 00:58:35,960 is also important, right? Because when you turn out code, 1063 00:58:36,440 --> 00:58:37,840 you need to have the guard rails, right? 1064 00:58:37,840 --> 00:58:40,960 Be it, you know, the design, you know, cyclomatic complexity, 1065 00:58:40,960 --> 00:58:43,960 security issues as well, right? Especially if you are not 1066 00:58:43,960 --> 00:58:47,280 knowledgeable about it and you adopt, you know, insecure code. 1067 00:58:47,440 --> 00:58:50,120 Probably it's also one thing that we need to think about. 1068 00:58:50,560 --> 00:58:53,600 And I think another thing that I would say for managers out 1069 00:58:53,600 --> 00:58:56,720 there, leaders, right, try to be hands on leveraging this tool, 1070 00:58:56,720 --> 00:58:59,120 right? Because especially for people 1071 00:58:59,120 --> 00:59:03,640 like me, you know, we are in the 40s sometimes, you know, we are 1072 00:59:03,640 --> 00:59:06,760 into management a lot, right? And we forget our hands on 1073 00:59:06,760 --> 00:59:09,280 experience. So at least adapting, trying to 1074 00:59:09,280 --> 00:59:13,680 learn what these two is capable of that kind of gives you the 1075 00:59:13,680 --> 00:59:15,760 first effect is like, you know what these two is capable of. 1076 00:59:15,760 --> 00:59:18,960 But the other aspect is like, how can you think about the 1077 00:59:18,960 --> 00:59:22,880 impact of using these tools? You know, what are the effect to 1078 00:59:22,880 --> 00:59:26,480 your maybe your team, maybe to your process, right? 1079 00:59:26,800 --> 00:59:29,720 And I think the most important thing as well now becomes like 1080 00:59:29,720 --> 00:59:31,840 first principle thinking and systems thinking. 1081 00:59:31,840 --> 00:59:34,160 I think you also mentioned it in your block, right? 1082 00:59:34,600 --> 00:59:37,480 So I think manager will have a tougher job, I would say, 1083 00:59:37,800 --> 00:59:41,240 especially trying to keep up with the AI changes and putting 1084 00:59:41,240 --> 00:59:45,280 guardrails and make sure that the team is as productive as 1085 00:59:45,280 --> 00:59:47,720 they can be. So we have talked a lot about 1086 00:59:47,720 --> 00:59:50,240 all these aspects. Is there any, maybe one thing 1087 00:59:50,240 --> 00:59:53,640 that you think you would still want to convey for people who 1088 00:59:53,640 --> 00:59:58,280 listen to this conversation? I think I would, yes. 1089 00:59:58,280 --> 01:00:01,560 I would say that you know you're not alone if you are. 1090 01:00:01,840 --> 01:00:04,040 If you're listening to this and you've been wondering what's 1091 01:00:04,040 --> 01:00:08,000 going to happen to your own job or whether you you have a path 1092 01:00:08,000 --> 01:00:11,120 forward from wherever you are today, you're certainly not 1093 01:00:11,120 --> 01:00:12,960 alone. There's many, many of us. 1094 01:00:12,960 --> 01:00:16,440 There's something like 27,000,000 developers around the 1095 01:00:16,440 --> 01:00:17,960 world. So there's a lot of software 1096 01:00:17,960 --> 01:00:20,880 engineers who are in one way, shape or form currently 1097 01:00:20,880 --> 01:00:22,720 wondering what's going to happen. 1098 01:00:23,560 --> 01:00:26,480 Be curious to learn to use these tools, right? 1099 01:00:26,480 --> 01:00:33,120 Like, get good at them and imagine, use your imagination to 1100 01:00:33,120 --> 01:00:36,880 work out how you'd like to see it go and see if you can make it 1101 01:00:36,880 --> 01:00:37,800 happen that way. Right. 1102 01:00:37,800 --> 01:00:40,880 But play. It doesn't all need to be doom 1103 01:00:40,880 --> 01:00:42,480 and gloom. There's a lot of excitement 1104 01:00:42,480 --> 01:00:45,680 around. But yeah, just know that you're 1105 01:00:45,680 --> 01:00:49,440 not alone. I think is probably a good myth. 1106 01:00:50,440 --> 01:00:53,600 Yeah, that's a good thing. So Annie, it's been a pleasant 1107 01:00:53,600 --> 01:00:55,840 conversation. So I have one last question for 1108 01:00:55,840 --> 01:00:57,640 you. I asked this last time as well. 1109 01:00:57,640 --> 01:01:00,320 So what I call the tree technical leadership wisdom. 1110 01:01:00,800 --> 01:01:03,040 So just think like an advice that you want to give to the 1111 01:01:03,040 --> 01:01:05,440 listeners. Maybe is there anything 1112 01:01:05,440 --> 01:01:08,240 different for this episode? Maybe can you share with us? 1113 01:01:09,160 --> 01:01:13,360 Yeah, so I reflected a bit on this and it's funny to think 1114 01:01:13,360 --> 01:01:16,600 back four years ago, but time has changed. 1115 01:01:16,800 --> 01:01:19,000 Well a lot has changed in the time in between. 1116 01:01:19,000 --> 01:01:23,800 And so I think this time my advice would be or the take lead 1117 01:01:23,800 --> 01:01:27,200 wisdom would be more around AI is what we've just spoken about. 1118 01:01:27,200 --> 01:01:31,080 So the first point would be for our software engineering 1119 01:01:31,080 --> 01:01:34,880 community to really embrace the full role. 1120 01:01:34,880 --> 01:01:38,000 Again. You know, there was a time when 1121 01:01:38,240 --> 01:01:43,320 you specialized on the code, as we spoke about before on the 1122 01:01:43,320 --> 01:01:46,320 syntax on all of that minor, minor detail. 1123 01:01:46,760 --> 01:01:49,440 Keep that in your back pocket, but start looking for 1124 01:01:49,440 --> 01:01:54,240 opportunities to expand your horizons and you know, have 1125 01:01:54,240 --> 01:01:59,120 opportunities to experience and practice requirements gathering, 1126 01:01:59,440 --> 01:02:02,760 you know, software design. Think about that systems 1127 01:02:02,760 --> 01:02:06,240 thinking that higher level thinking that we keep being told 1128 01:02:06,240 --> 01:02:08,760 is, you know, where we will now go and spend more of our time. 1129 01:02:08,760 --> 01:02:12,040 We'll start looking for opportunities to use that skill 1130 01:02:12,080 --> 01:02:14,160 because it's the muscle that you have to build up. 1131 01:02:14,560 --> 01:02:17,800 So I think it's important for everyone to start looking for 1132 01:02:17,800 --> 01:02:20,680 those opportunities. So basically move back towards 1133 01:02:20,800 --> 01:02:24,320 more generalist than specialist for the time being. 1134 01:02:24,840 --> 01:02:28,800 And #2 is repeating what we have spoken about a fair bit. 1135 01:02:28,800 --> 01:02:32,520 But that speed really is very seductive and fun. 1136 01:02:33,120 --> 01:02:37,560 But fragility and complexity is very expensive. 1137 01:02:37,960 --> 01:02:42,440 And it may not be obvious immediately, but we are starting 1138 01:02:42,440 --> 01:02:46,760 to see signs of that. So temper your enthusiasm for 1139 01:02:46,760 --> 01:02:50,840 that huge productivity gain that you're reading about in social 1140 01:02:50,840 --> 01:02:55,120 media and in academic papers. It's there to be gained, but 1141 01:02:55,720 --> 01:03:00,160 it's a trade off between that and the potential cost of trying 1142 01:03:00,160 --> 01:03:01,840 to maintain this software later on. 1143 01:03:01,840 --> 01:03:05,360 So find your balance. It'll be different for everyone. 1144 01:03:05,360 --> 01:03:09,400 At a startup you might be able to take more risks, but in more 1145 01:03:10,200 --> 01:03:15,040 institutionalized large code bases, you have to take that 1146 01:03:15,040 --> 01:03:20,440 into account a lot I would say. And the third one is an 1147 01:03:20,440 --> 01:03:23,760 interesting one, and there might be opinions around this, but I 1148 01:03:23,760 --> 01:03:27,880 would suggest to start thinking of these AI tools as not just 1149 01:03:27,880 --> 01:03:33,120 mere tools, but as team members. I feel like that's the way that 1150 01:03:33,120 --> 01:03:35,280 it's going. You know, we are already seeing 1151 01:03:35,280 --> 01:03:40,200 people talking about human AI teams and what that means. 1152 01:03:40,200 --> 01:03:43,560 You know, there's research out there about some papers say that 1153 01:03:43,880 --> 01:03:49,320 a human plus an AI can be more productive or perform better 1154 01:03:49,400 --> 01:03:52,720 than two humans together, and others find the opposite. 1155 01:03:52,720 --> 01:03:55,960 So it's an evolving landscape at the moment. 1156 01:03:56,440 --> 01:04:00,320 But if you think about it, the way that you interact with these 1157 01:04:00,320 --> 01:04:04,840 tools is more similar to how you would interact with a team 1158 01:04:04,840 --> 01:04:06,960 member perhaps. You know, some people describe 1159 01:04:06,960 --> 01:04:11,280 them as it's like working with a very eager junior engineer who 1160 01:04:11,280 --> 01:04:13,760 might make a lot of mistakes but is eager to please. 1161 01:04:13,760 --> 01:04:16,440 You know, others actually describe it as having like a 1162 01:04:16,440 --> 01:04:20,280 principal engineer on demand. And it all depends on the 1163 01:04:20,280 --> 01:04:24,240 perspective of the person in the equation, right? 1164 01:04:24,680 --> 01:04:29,600 But for both software engineers and software engineering leaders 1165 01:04:29,960 --> 01:04:34,040 alike, there may well come a time when your team will be 1166 01:04:34,040 --> 01:04:38,120 made-up of some humans and some AI. 1167 01:04:38,800 --> 01:04:41,400 Start thinking about how that changes things for you. 1168 01:04:41,400 --> 01:04:44,040 If you're not treating it just like a tool, but instead you're 1169 01:04:44,040 --> 01:04:47,760 actually treating it a bit more like a team member, then how 1170 01:04:47,760 --> 01:04:50,720 does that collaboration work? What roles would they fill and 1171 01:04:50,720 --> 01:04:52,360 how do you manage their performance? 1172 01:04:52,360 --> 01:04:55,720 You know, and maybe maybe there's some rubrics around that 1173 01:04:55,720 --> 01:04:57,280 that we need to start thinking about. 1174 01:04:57,920 --> 01:05:00,160 So that's sort of the futuristic thinking. 1175 01:05:00,960 --> 01:05:04,960 Food for thought. A very interesting wisdom. 1176 01:05:04,960 --> 01:05:07,440 So I would say maybe it's like a tech leadership wisdom in the 1177 01:05:07,440 --> 01:05:10,520 age of AI. So I think I like the last one 1178 01:05:10,520 --> 01:05:14,400 that you mentioned. We have to now think that AI is 1179 01:05:14,480 --> 01:05:17,880 part of the team, right? So it's a members of the team 1180 01:05:18,160 --> 01:05:22,040 because if everyone is using it right and they kind of like take 1181 01:05:22,240 --> 01:05:25,720 a lot of part in the delivery aspect of the team, so they 1182 01:05:25,720 --> 01:05:27,280 should be treated as a member as well. 1183 01:05:27,520 --> 01:05:29,680 So I think that's a very intriguing thoughts for people 1184 01:05:29,680 --> 01:05:33,480 to think about South, for people who would love to maybe continue 1185 01:05:33,480 --> 01:05:36,520 this discussion, asking you more questions, follow you online, is 1186 01:05:36,520 --> 01:05:38,480 that the best place for them to reach out? 1187 01:05:39,080 --> 01:05:42,920 Sure, probably LinkedIn is the one that I seem to gravitate 1188 01:05:43,040 --> 01:05:47,240 towards the most, but I'm also on Blue Sky these days and I'm 1189 01:05:47,520 --> 01:05:51,360 probably less active on X. But yeah, those other two or 1190 01:05:51,640 --> 01:05:54,600 just through my website, you should be able to track me down 1191 01:05:54,600 --> 01:05:57,600 to any of those others. Annie valor.com, that's my blog. 1192 01:05:58,760 --> 01:06:01,680 Right, so I would say for people who are intrigued by the 1193 01:06:01,680 --> 01:06:04,480 software identity thing, read Annie's blog post as well. 1194 01:06:04,480 --> 01:06:07,800 I think it's a very good food for thought, especially in this 1195 01:06:07,840 --> 01:06:10,680 era where everyone seems to be concerned about their role. 1196 01:06:11,160 --> 01:06:13,080 So thank you so much again Annie for your time. 1197 01:06:13,080 --> 01:06:16,960 So yeah, thank you and wish you good luck with all the research 1198 01:06:16,960 --> 01:06:20,360 for this AI related stuff. Thank you so much, Henry. 1199 01:06:20,360 --> 01:06:23,120 And maybe let's not leave it four years until we speak again, 1200 01:06:23,120 --> 01:06:26,280 but thank you very much for the opportunity to be a repeat guest 1201 01:06:26,280 --> 01:06:29,880 on your podcast. And yeah, I guess we'll catch 1202 01:06:29,880 --> 01:06:32,040 you around. Yeah, looking forward for that. 1203 01:06:32,600 --> 01:06:33,040 Thank you.