Pełna transkrypcja odcinka In Conversation with Christian Dippel: Developing Competitive Edge with AI

that if you build it, 78 00:03:15,660 --> 00:03:16,675 they won't just come. 79 00:03:16,675 --> 00:03:18,050 So in other words, you can't just 80 00:03:18,050 --> 00:03:19,717 build the thing that you're in love with 81 00:03:19,717 --> 00:03:22,050 and assume that it can stand on its own two feet. 82 00:03:22,050 --> 00:03:24,990 And I had to quickly learn a lot of strategy. 83 00:03:24,990 --> 00:03:28,040 And then, really, it sort of became this natural confluence 84 00:03:28,040 --> 00:03:31,110 where I was building projects in data science. 85 00:03:31,110 --> 00:03:32,550 I had to learn the strategy. 86 00:03:32,550 --> 00:03:34,740 And then right around the time when that happened, 87 00:03:34,740 --> 00:03:35,850 AI came around. 88 00:03:35,850 --> 00:03:38,400 And so for me, strategy, almost overnight, 89 00:03:38,400 --> 00:03:40,680 very quickly just became AI strategy. 90 00:03:40,680 --> 00:03:43,260 So in other words, for me, for what I was doing, 91 00:03:43,260 --> 00:03:46,140 there was no strategy that wasn't AI strategy. 92 00:03:46,140 --> 00:03:48,950 HOST: Thank you for the introduction. 93 00:03:48,950 --> 00:03:52,460 I'm smiling because it's two comments here at the-- build it, 94 00:03:52,460 --> 00:03:55,530 and they will come doesn't necessarily really happen. 95 00:03:55,530 --> 00:03:58,050 It's how do you go beyond that, and how do you engage? 96 00:03:58,050 --> 00:04:03,060 And given the topic at hand, even when you reference 2019, 97 00:04:03,060 --> 00:04:05,330 when it comes to AI, that feels like it's centuries 98 00:04:05,330 --> 00:04:09,150 ago in terms of how fast everything is moving. 99 00:04:09,150 --> 00:04:13,940 So let's go back to the topic or even the headline. 100 00:04:13,940 --> 00:04:17,180 We talk about, creating a competitive advantage, 101 00:04:17,180 --> 00:04:19,800 and how can I play a role? 102 00:04:19,800 --> 00:04:22,740 I actually talked about moving away from buzzwords. 103 00:04:22,740 --> 00:04:26,030 Competitive advantage almost feels like a buzzword, 104 00:04:26,030 --> 00:04:27,570 to some extent. 105 00:04:27,570 --> 00:04:29,810 It's obviously something that we strive 106 00:04:29,810 --> 00:04:31,640 for when you're leading organizations 107 00:04:31,640 --> 00:04:33,840 to create that edge. 108 00:04:33,840 --> 00:04:35,930 So how can we think about it maybe 109 00:04:35,930 --> 00:04:39,210 in a more practical or pragmatic term? 110 00:04:39,210 --> 00:04:41,730 CHRISTIAN DIPPEL: I got into strategy in a roundabout way, 111 00:04:41,730 --> 00:04:42,313 you would say. 112 00:04:42,313 --> 00:04:43,860 I wasn't trained as a strategist. 113 00:04:43,860 --> 00:04:45,960 I was trained as an economist in global macro, 114 00:04:45,960 --> 00:04:47,820 and then the strategy came later. 115 00:04:47,820 --> 00:04:52,820 And when I was at UCLA, I had this very renowned 116 00:04:52,820 --> 00:04:56,750 senior strategy colleague who retired midway through my stint 117 00:04:56,750 --> 00:05:00,028 at UCLA, Richard Rumelt. And he used to tell the story of-- 118 00:05:00,028 --> 00:05:02,570 he would have another senior colleague come into his strategy 119 00:05:02,570 --> 00:05:06,210 class to maybe get into teaching strategy himself. 120 00:05:06,210 --> 00:05:09,440 And after sitting there for 10 classes, he walked up to Richard 121 00:05:09,440 --> 00:05:12,380 and said, I think I can summarize your entire strategy 122 00:05:12,380 --> 00:05:15,000 class in one question, what's going on here? 123 00:05:15,000 --> 00:05:17,460 The point that he was making is, at the end of the day, 124 00:05:17,460 --> 00:05:18,350 strategy is-- 125 00:05:18,350 --> 00:05:21,890 it's a fairly unstructured thing in the sense 126 00:05:21,890 --> 00:05:23,520 that it's so context-specific. 127 00:05:23,520 --> 00:05:25,580 And this was an interesting experience for me 128 00:05:25,580 --> 00:05:28,440 as a macroeconomist who is really a data scientist. 129 00:05:28,440 --> 00:05:30,507 You're always in the business of generalities, 130 00:05:30,507 --> 00:05:32,090 so you're always trying to figure out, 131 00:05:32,090 --> 00:05:33,780 what does the average do? 132 00:05:33,780 --> 00:05:35,713 What are like the patterns in the data? 133 00:05:35,713 --> 00:05:37,130 But when you start to think about, 134 00:05:37,130 --> 00:05:40,850 what is going to make my one thing successful in this sea 135 00:05:40,850 --> 00:05:43,220 of other things around it, it's like your whole way 136 00:05:43,220 --> 00:05:45,380 of thinking totally pivots into-- everything 137 00:05:45,380 --> 00:05:46,800 becomes about context. 138 00:05:46,800 --> 00:05:49,200 And the data is a background thing. 139 00:05:49,200 --> 00:05:50,820 You need to know some business trends. 140 00:05:50,820 --> 00:05:53,445 You need to have a sense for consumer behavior, 141 00:05:53,445 --> 00:05:55,820 competitive landscape behavior, all these things that can 142 00:05:55,820 --> 00:05:57,068 be expressed quantitatively. 143 00:05:57,068 --> 00:05:58,610 But at the end of the day, it becomes 144 00:05:58,610 --> 00:06:02,510 this highly context-specific, very qualitative endeavor 145 00:06:02,510 --> 00:06:06,000 and really thinking deeply about, what is going on here? 146 00:06:06,000 --> 00:06:08,330 The answer to the question, what is going on here, 147 00:06:08,330 --> 00:06:10,860 is going to be different for every kind of business. 148 00:06:10,860 --> 00:06:13,100 There will be businesses where AI really hits you 149 00:06:13,100 --> 00:06:14,262 on the process innovation. 150 00:06:14,262 --> 00:06:16,220 There'll be businesses where it really hits you 151 00:06:16,220 --> 00:06:17,460 on the product innovation. 152 00:06:17,460 --> 00:06:18,950 And I think it's really about-- 153 00:06:18,950 --> 00:06:20,750 the way that I think about it is really 154 00:06:20,750 --> 00:06:23,870 to

not be too heavy on the playbooks, like, let's 155 00:06:23,870 --> 00:06:25,550 apply this framework or that framework. 156 00:06:25,550 --> 00:06:28,280 Frameworks have their use, for sure, but nothing, I think, 157 00:06:28,280 --> 00:06:31,350 replaces deeply thinking about your situation. 158 00:06:31,350 --> 00:06:33,570 HOST: Great way to frame it. 159 00:06:33,570 --> 00:06:36,870 I find sometimes when we use these bigger terms-- 160 00:06:36,870 --> 00:06:38,690 and even I had a recent conversation 161 00:06:38,690 --> 00:06:41,690 with another faculty where we dug in specifically 162 00:06:41,690 --> 00:06:44,070 on strategy and scenario planning 163 00:06:44,070 --> 00:06:48,540 and, how do you stay ahead in today's rapidly changing world? 164 00:06:48,540 --> 00:06:51,270 It's sort of broke it down to say-- 165 00:06:51,270 --> 00:06:54,710 but it doesn't have to be this big grandiose piece where 166 00:06:54,710 --> 00:06:57,200 you almost run the risk of making it bigger than it is 167 00:06:57,200 --> 00:06:58,680 and creating that inertia. 168 00:06:58,680 --> 00:07:01,740 It's like, OK, so what's going on today? 169 00:07:01,740 --> 00:07:03,090 What's going on tomorrow? 170 00:07:03,090 --> 00:07:06,540 Continue to ask myself these questions. 171 00:07:06,540 --> 00:07:08,540 So if we go to your work, we look 172 00:07:08,540 --> 00:07:12,600 at drawing on political economy, institutional economics. 173 00:07:12,600 --> 00:07:17,870 How do these lenses illustrate the strategic role 174 00:07:17,870 --> 00:07:19,560 that AI can play? 175 00:07:19,560 --> 00:07:22,010 And maybe I'll put in a second part 176 00:07:22,010 --> 00:07:25,610 to this question around hidden assumptions 177 00:07:25,610 --> 00:07:28,763 that might be at play as well. 178 00:07:28,763 --> 00:07:29,930 What are your thoughts here? 179 00:07:29,930 --> 00:07:31,305 CHRISTIAN DIPPEL: In my ventures, 180 00:07:31,305 --> 00:07:32,600 let's say-- let's start there. 181 00:07:32,600 --> 00:07:36,200 Those ventures, as well as my research, really are in-- 182 00:07:36,200 --> 00:07:40,410 political economy people refer to that as government relations. 183 00:07:40,410 --> 00:07:41,820 It's really similar things. 184 00:07:41,820 --> 00:07:44,400 And I saw that transition of-- 185 00:07:44,400 --> 00:07:46,860 2011 was when I got my PhD. 186 00:07:46,860 --> 00:07:50,840 2014, '15 is when it was cutting-edge to be 187 00:07:50,840 --> 00:07:54,140 using keyword extraction to characterize 188 00:07:54,140 --> 00:07:55,710 certain bodies of text. 189 00:07:55,710 --> 00:07:58,880 By 2016, you can show off with your machine learning 190 00:07:58,880 --> 00:07:59,990 algorithms. 191 00:07:59,990 --> 00:08:03,090 By 2017, you do sentiment analysis. 192 00:08:03,090 --> 00:08:05,550 And so the bar kept getting raised but in a somewhat, 193 00:08:05,550 --> 00:08:07,020 I would say, incremental manner. 194 00:08:07,020 --> 00:08:09,350 But it's really 2023 when it hits you. 195 00:08:09,350 --> 00:08:11,280 Suddenly, there's this level shift. 196 00:08:11,280 --> 00:08:12,860 There's a million things that you 197 00:08:12,860 --> 00:08:14,850 can do that you couldn't previously do. 198 00:08:14,850 --> 00:08:18,260 And from a business perspective, anyone with a lot of exposure 199 00:08:18,260 --> 00:08:20,390 to data or a lot of in-house data 200 00:08:20,390 --> 00:08:23,570 or needing to grapple with a lot of data-- there's text as data 201 00:08:23,570 --> 00:08:24,960 and there's quantitative data. 202 00:08:24,960 --> 00:08:27,500 I would say the manifestation of AI 203 00:08:27,500 --> 00:08:31,170 that we're talking about mostly today, which is not agentic, 204 00:08:31,170 --> 00:08:32,990 but let's say just like LLMs, that's 205 00:08:32,990 --> 00:08:34,470 mostly about text as data. 206 00:08:34,470 --> 00:08:36,679 And text as data is really an application where, 207 00:08:36,679 --> 00:08:41,000 if your business interfaces with text as data, which mine did-- 208 00:08:41,000 --> 00:08:42,830 with text as data, you're constantly 209 00:08:42,830 --> 00:08:46,250 in this world of not being able to see the forest for the trees 210 00:08:46,250 --> 00:08:48,030 because there's just so much text. 211 00:08:48,030 --> 00:08:49,830 So if you're interfacing with government, 212 00:08:49,830 --> 00:08:52,010 there's hundreds of bills every year 213 00:08:52,010 --> 00:08:55,970 that affect hundreds of X, which in turn affect thousands 214 00:08:55,970 --> 00:08:57,180 of regulations. 215 00:08:57,180 --> 00:09:00,230 There's thousands of pieces of formal commenting, lobby 216 00:09:00,230 --> 00:09:02,070 meeting, committee meeting transcripts. 217 00:09:02,070 --> 00:09:04,520 There's just such a big ocean of text 218 00:09:04,520 --> 00:09:07,830 that no traditional method can really get a good handle on, 219 00:09:07,830 --> 00:09:12,560 and that's really where AI it lowers the bar so much in terms 220 00:09:12,560 --> 00:09:14,630 of even a very small business being 221 00:09:14,630 --> 00:09:17,090 able to do so much with a lot of data, 222 00:09:17,090 --> 00:09:18,530 where you went from being able to, 223 00:09:18,530 --> 00:09:20,750 do more or less, nothing at all five years ago 224 00:09:20,750 --> 00:09:24,080 to being able to do enterprise-level applications 225 00:09:24,080 --> 00:09:27,560 with a tiny team if you are in this kind of very data 226 00:09:27,560 --> 00:09:29,780 heavy environment, which, in my case, 227 00:09:29,780 --> 00:09:32,930 in the political economy, institutional economics type 228 00:09:32,930 --> 00:09:34,710 environment, we work. 229 00:09:34,710 --> 00:09:37,430 HOST: I have conversations frequently 230 00:09:37,430 --> 00:09:41,990 with leaders who are grappling with almost too much. 231 00:09:41,990 --> 00:09:44,100 There's so much information. 232 00:09:44,100 --> 00:09:45,530

There's so much data. 233 00:09:45,530 --> 00:09:48,890 I don't even know where to start. 234 00:09:48,890 --> 00:09:50,390 So I'm keen. 235 00:09:50,390 --> 00:09:52,830 I want to take advantage of all the data. 236 00:09:52,830 --> 00:09:55,280 I want to be able to analyze and look at it 237 00:09:55,280 --> 00:09:57,080 and get insights that maybe I wasn't 238 00:09:57,080 --> 00:10:02,040 able to get before or get them in a much more expedited manner. 239 00:10:02,040 --> 00:10:07,160 Do you have an example of an organization or maybe someone 240 00:10:07,160 --> 00:10:09,950 that you work with that's doing this well? 241 00:10:09,950 --> 00:10:12,990 And how do they start, or what are they doing? 242 00:10:12,990 --> 00:10:16,280 Are there any secrets but are not 243 00:10:16,280 --> 00:10:18,890 really secret that we could share to just help 244 00:10:18,890 --> 00:10:22,370 those that are listening in who want to move forward, know where 245 00:10:22,370 --> 00:10:23,220 to start? 246 00:10:23,220 --> 00:10:25,070 CHRISTIAN DIPPEL: There's general, almost 247 00:10:25,070 --> 00:10:27,110 irony you could say, where-- 248 00:10:27,110 --> 00:10:30,310 I always think, in strategy, there's two buckets of strategy, 249 00:10:30,310 --> 00:10:32,310 or at least that's a frame you can impose on it. 250 00:10:32,310 --> 00:10:34,070 You can say there's quantitative decision 251 00:10:34,070 --> 00:10:37,980 making, which is operating in data-rich environments. 252 00:10:37,980 --> 00:10:41,937 So a typical example of that would be pricing decisions. 253 00:10:41,937 --> 00:10:43,520 For the most part, if you're operating 254 00:10:43,520 --> 00:10:45,480 in a market that's well established, 255 00:10:45,480 --> 00:10:48,630 you can make very data-informed decisions 256 00:10:48,630 --> 00:10:49,930 about pricing decisions. 257 00:10:49,930 --> 00:10:52,870 You can call that strategy, but it's very quantitative, 258 00:10:52,870 --> 00:10:54,190 very data-driven strategy. 259 00:10:54,190 --> 00:10:58,360 And then there's very imagination-driven strategy, 260 00:10:58,360 --> 00:11:01,410 which is the strategy you apply to settings for markets 261 00:11:01,410 --> 00:11:03,390 that don't exist yet. 262 00:11:03,390 --> 00:11:05,790 Who knows what the market for space travel 263 00:11:05,790 --> 00:11:07,510 is going to look like in 10 years? 264 00:11:07,510 --> 00:11:08,710 We can't know this. 265 00:11:08,710 --> 00:11:09,850 We can only imagine it. 266 00:11:09,850 --> 00:11:13,080 And I think, one of the tensions that people 267 00:11:13,080 --> 00:11:14,970 seem to have when it comes to AI is-- 268 00:11:14,970 --> 00:11:16,630 AI is all about the data. 269 00:11:16,630 --> 00:11:19,230 It's all about ingesting and transforming the data. 270 00:11:19,230 --> 00:11:21,090 Yet when we think about, what are 271 00:11:21,090 --> 00:11:23,940 going to be the repercussions of AI for my business 272 00:11:23,940 --> 00:11:27,900 or in my industry-- it's really a very qualitative exercise 273 00:11:27,900 --> 00:11:30,870 of imagining the future that is not-- inherently, 274 00:11:30,870 --> 00:11:33,550 because it is the future, can't be very data-informed. 275 00:11:33,550 --> 00:11:35,225 And because you can't really interpolate 276 00:11:35,225 --> 00:11:37,600 from what's been going on in the last two or three years, 277 00:11:37,600 --> 00:11:39,070 there's just not enough data. 278 00:11:39,070 --> 00:11:40,360 It's not established enough. 279 00:11:40,360 --> 00:11:41,410 It's too exponential. 280 00:11:41,410 --> 00:11:44,550 The process is too to accelerating to really use 281 00:11:44,550 --> 00:11:47,415 data to extrapolate into what's that going to look like in two 282 00:11:47,415 --> 00:11:48,040 or three years. 283 00:11:48,040 --> 00:11:49,870 So really, you just need to think about it. 284 00:11:49,870 --> 00:11:51,100 HOST: Yeah, for sure. 285 00:11:51,100 --> 00:11:54,390 You mentioned earlier around frameworks, which sometimes 286 00:11:54,390 --> 00:11:57,070 can be a challenge if we over rely, 287 00:11:57,070 --> 00:12:02,080 also can be an enabler as a way to get things structured. 288 00:12:02,080 --> 00:12:06,180 So let's play out strategy and maybe 289 00:12:06,180 --> 00:12:08,670 sort of bold or future thinking. 290 00:12:08,670 --> 00:12:12,180 Frameworks that either you've used or approaches that you've 291 00:12:12,180 --> 00:12:16,320 seen that help leaders process their thinking 292 00:12:16,320 --> 00:12:19,230 or start to get things organized so that they 293 00:12:19,230 --> 00:12:23,400 can make better decisions, they can lean into the data 294 00:12:23,400 --> 00:12:25,330 to create insights-- 295 00:12:25,330 --> 00:12:28,240 CHRISTIAN DIPPEL: So strategy is heavy on frameworks 296 00:12:28,240 --> 00:12:30,010 and I think for good reason. 297 00:12:30,010 --> 00:12:32,370 I think frameworks are always a little bit 298 00:12:32,370 --> 00:12:34,410 of a double edged sword in the sense 299 00:12:34,410 --> 00:12:36,880 that a framework offers you a playbook. 300 00:12:36,880 --> 00:12:39,340 And I think at the end of the day, 301 00:12:39,340 --> 00:12:42,010 you need to think deeply about your situation. 302 00:12:42,010 --> 00:12:44,040 And I think playbooks or frameworks 303 00:12:44,040 --> 00:12:47,800 can help you get there quicker, or they can help you-- 304 00:12:47,800 --> 00:12:49,440 It's almost like, you know, you're 305 00:12:49,440 --> 00:12:52,260 trying on a Phillips screwdriver and a flathead screwdriver, 306 00:12:52,260 --> 00:12:54,130 and you see which one fits the screw. 307 00:12:54,130 --> 00:12:56,280 So you have tools at your disposal 308 00:12:56,280 --> 00:12:58,030 that helped you solve the problem. 309 00:12:58,030 --> 00:13:01,660 Where I caution always when I work with clients, 310 00:13:01,660 --> 00:13:03,360 when I work with

students in exec ed, 311 00:13:03,360 --> 00:13:06,510 is it also satisfies a little bit our bias 312 00:13:06,510 --> 00:13:09,210 towards not having to think deeply and just having 313 00:13:09,210 --> 00:13:10,510 a playbook to play. 314 00:13:10,510 --> 00:13:13,980 If someone just hands me my five forces canvas or my business 315 00:13:13,980 --> 00:13:16,270 model canvas and I can fill in the thing, 316 00:13:16,270 --> 00:13:18,340 I can feel like I've achieved something. 317 00:13:18,340 --> 00:13:21,750 But I think ultimately, a person that doesn't have those tools 318 00:13:21,750 --> 00:13:25,020 but is really willing to sit there still and think deeply 319 00:13:25,020 --> 00:13:26,820 about their situation, they will come up 320 00:13:26,820 --> 00:13:28,770 with better solutions than the person that 321 00:13:28,770 --> 00:13:31,810 is just superficially filling in their business model canvas. 322 00:13:31,810 --> 00:13:34,500 And that's not a criticism on the business model canvas. 323 00:13:34,500 --> 00:13:37,920 It's human nature that I think it can be a little bit offering 324 00:13:37,920 --> 00:13:40,800 us sort of this like fake ticking boxes sense 325 00:13:40,800 --> 00:13:42,580 of getting a handle on a situation. 326 00:13:42,580 --> 00:13:44,050 So I love frameworks. 327 00:13:44,050 --> 00:13:45,130 They're very useful. 328 00:13:45,130 --> 00:13:49,840 But I always caution students or in consulting work with clients. 329 00:13:49,840 --> 00:13:51,040 They're a means to an end. 330 00:13:51,040 --> 00:13:52,450 They're not the end in itself. 331 00:13:52,450 --> 00:13:55,750 HOST: I like your analogy around the screwdrivers. 332 00:13:55,750 --> 00:14:00,450 You know, we sometimes confuse motion with progress. 333 00:14:00,450 --> 00:14:05,010 So someone can actually be moving but not necessarily 334 00:14:05,010 --> 00:14:05,980 advancing. 335 00:14:05,980 --> 00:14:07,300 But it satisfies this. 336 00:14:07,300 --> 00:14:07,900 Oh, I'm busy. 337 00:14:07,900 --> 00:14:09,490 I'm doing things, right? 338 00:14:09,490 --> 00:14:10,990 I'm making stuff happen. 339 00:14:10,990 --> 00:14:14,920 How would you help someone sort of shift between the two? 340 00:14:14,920 --> 00:14:17,800 So are there any cues that I can say, 341 00:14:17,800 --> 00:14:21,090 like maybe I'm not doing enough of the hard work, 342 00:14:21,090 --> 00:14:23,500 and I'm sort of relying too much on. 343 00:14:23,500 --> 00:14:25,240 OK, I filled out the sheet. 344 00:14:25,240 --> 00:14:28,513 I've done work, pat myself on the back. 345 00:14:28,513 --> 00:14:30,180 What are some things I should be holding 346 00:14:30,180 --> 00:14:33,100 myself accountable to maybe or paying attention to? 347 00:14:33,100 --> 00:14:34,440 CHRISTIAN DIPPEL: If you look back in history, and that's 348 00:14:34,440 --> 00:14:36,232 always like when you look back at the past, 349 00:14:36,232 --> 00:14:39,870 things appear very obvious, but they're not obvious at the time 350 00:14:39,870 --> 00:14:43,810 or for us looking forward for a new technological disruption. 351 00:14:43,810 --> 00:14:45,820 But with past technological disruptions, 352 00:14:45,820 --> 00:14:49,090 you often have these first order effects that are quite obvious. 353 00:14:49,090 --> 00:14:52,410 The radio comes around, and it totally disrupts the newspaper 354 00:14:52,410 --> 00:14:53,230 industry. 355 00:14:53,230 --> 00:14:56,880 TV comes around, and it completely disrupts the radio 356 00:14:56,880 --> 00:14:58,000 station industry. 357 00:14:58,000 --> 00:15:00,810 The railway comes around, and it kills off 358 00:15:00,810 --> 00:15:02,950 the Pony Express in the American West, 359 00:15:02,950 --> 00:15:04,860 and it completely destroys anyone 360 00:15:04,860 --> 00:15:07,860 who is invested in canal building, which 10 years prior, 361 00:15:07,860 --> 00:15:09,180 was incredibly profitable. 362 00:15:09,180 --> 00:15:13,150 Because these are sort of first order obvious replacements. 363 00:15:13,150 --> 00:15:16,650 And many in what some management people 364 00:15:16,650 --> 00:15:18,220 like to call fast history. 365 00:15:18,220 --> 00:15:21,030 So recent technological history, whether that's 366 00:15:21,030 --> 00:15:23,260 in microchips or PCs or whatever. 367 00:15:23,260 --> 00:15:26,320 You get these examples like Netflix killing Blockbuster. 368 00:15:26,320 --> 00:15:27,990 And they clearly, Blockbuster just 369 00:15:27,990 --> 00:15:30,360 didn't think about the first order effect 370 00:15:30,360 --> 00:15:32,470 of home delivery of DVDs. 371 00:15:32,470 --> 00:15:34,530 And then subsequent to that, the long game 372 00:15:34,530 --> 00:15:36,085 was always going to be streaming. 373 00:15:36,085 --> 00:15:37,210 They didn't think about it. 374 00:15:37,210 --> 00:15:38,170 They got killed by it. 375 00:15:38,170 --> 00:15:40,420 Kodak getting killed by digicams. 376 00:15:40,420 --> 00:15:42,120 But then oftentimes, when you think 377 00:15:42,120 --> 00:15:45,750 about the really fundamental technological changes that 378 00:15:45,750 --> 00:15:49,290 really changed the way we live, the second order effects 379 00:15:49,290 --> 00:15:51,210 are actually much more interesting 380 00:15:51,210 --> 00:15:53,102 or, well, depending-- if you're Kodak, 381 00:15:53,102 --> 00:15:54,310 they're not more interesting. 382 00:15:54,310 --> 00:15:55,510 Depending on what business you run, 383 00:15:55,510 --> 00:15:57,460 you might be more affected by the first order effect. 384 00:15:57,460 --> 00:15:59,400 But what I'm trying to get at is you might be operating 385 00:15:59,400 --> 00:16:02,350 in a business where you say, AI doesn't affect me all that much. 386 00:16:02,350 --> 00:16:04,900 But when you think through the second order effects, 387 00:16:04,900 --> 00:16:07,620 think about something like the tractor coming around 388

00:16:07,620 --> 00:16:11,080 in the 1920s completely changed agriculture, 389 00:16:11,080 --> 00:16:14,257 completely changed employment opportunities in agriculture. 390 00:16:14,257 --> 00:16:15,840 Those are like the first order effects 391 00:16:15,840 --> 00:16:17,410 that people could see right away. 392 00:16:17,410 --> 00:16:19,260 But then within 10 years, what happened 393 00:16:19,260 --> 00:16:22,110 is you got massive rural urban migration, 394 00:16:22,110 --> 00:16:25,420 and you got complete changing of the urban landscape, 395 00:16:25,420 --> 00:16:28,150 the rise of superstar cities, urban manufacturing, 396 00:16:28,150 --> 00:16:30,960 the rise of the service sector, and just this like infinite 397 00:16:30,960 --> 00:16:33,498 plethora of new opportunities coming around. 398 00:16:33,498 --> 00:16:35,040 And that's the part where I sometimes 399 00:16:35,040 --> 00:16:37,110 think, because I was fortunate enough 400 00:16:37,110 --> 00:16:39,820 to have some training as an economic historian, 401 00:16:39,820 --> 00:16:41,770 you have a bit of this longer run lens of, 402 00:16:41,770 --> 00:16:44,970 OK, let's try to really think through what this thing is going 403 00:16:44,970 --> 00:16:47,020 to do in five years, in 10 years, 404 00:16:47,020 --> 00:16:49,480 and not just get hung up on the difference of 405 00:16:49,480 --> 00:16:52,050 does my competitor have a chatbot today, 406 00:16:52,050 --> 00:16:54,240 and I don't have a chatbot, and what 407 00:16:54,240 --> 00:16:56,880 that's going to do to my perception of my brand 408 00:16:56,880 --> 00:16:58,040 in the marketplace today? 409 00:16:58,040 --> 00:17:08,130 410 00:17:08,130 --> 00:17:11,260 HOST: So let's build on that last point, right? 411 00:17:11,260 --> 00:17:14,210 And I think most leaders endeavor to learn from the past. 412 00:17:14,210 --> 00:17:14,710 Right. 413 00:17:14,710 --> 00:17:17,579 We can't change it, but we can certainly 414 00:17:17,579 --> 00:17:18,940 glean insights from it. 415 00:17:18,940 --> 00:17:22,380 So how do you challenge and get someone 416 00:17:22,380 --> 00:17:27,132 to think five and 10 years when we're talking about AI? 417 00:17:27,132 --> 00:17:28,590 And I'm thinking, jeez, something's 418 00:17:28,590 --> 00:17:30,640 going to change in five or 10 days. 419 00:17:30,640 --> 00:17:32,820 And that idea of, oh, I don't have 420 00:17:32,820 --> 00:17:35,920 a chatbot that is on my site. 421 00:17:35,920 --> 00:17:37,500 And if someone goes to my site, I'm 422 00:17:37,500 --> 00:17:39,690 going to look like I'm behind the times. 423 00:17:39,690 --> 00:17:41,940 So how do you toggle between what 424 00:17:41,940 --> 00:17:45,460 do I need to do right now while anticipating? 425 00:17:45,460 --> 00:17:48,370 Because if I don't think five years plus out, 426 00:17:48,370 --> 00:17:51,090 I might be in for a much bigger surprise than not having 427 00:17:51,090 --> 00:17:52,202 a

chatbot today. 428 00:17:52,202 --> 00:17:53,910 CHRISTIAN DIPPEL: There's a lot of myopia 429 00:17:53,910 --> 00:17:57,310 in how humans think in general. 430 00:17:57,310 --> 00:17:59,460 I think the value of trying to bring 431 00:17:59,460 --> 00:18:02,460 a little bit of economic history into whether that's 432 00:18:02,460 --> 00:18:05,240 a one on one consultation or whether that's a classroom 433 00:18:05,240 --> 00:18:05,740 setting. 434 00:18:05,740 --> 00:18:07,907 Bringing a little bit of that historical perspective 435 00:18:07,907 --> 00:18:09,970 can allow you to do a little bit of back casting. 436 00:18:09,970 --> 00:18:13,170 You can say, OK, what is it that people didn't see 437 00:18:13,170 --> 00:18:18,030 in the '60s or in the '70s about what the arrival of all sorts 438 00:18:18,030 --> 00:18:21,660 of home appliances would do to female labor force 439 00:18:21,660 --> 00:18:24,730 participation, and then female educational attainments, 440 00:18:24,730 --> 00:18:27,040 and then changing gender roles 10, 20, 441 00:18:27,040 --> 00:18:30,070 30 years out from those technological innovations? 442 00:18:30,070 --> 00:18:32,340 What are the kinds of biases that are making 443 00:18:32,340 --> 00:18:34,450 it hard to see out like that? 444 00:18:34,450 --> 00:18:37,380 Really trying to, I think, lean into just sitting 445 00:18:37,380 --> 00:18:41,020 back and imagining what is my industry going to look like, 446 00:18:41,020 --> 00:18:43,200 either in free form or doing that through something 447 00:18:43,200 --> 00:18:44,877 like Porter's Five Forces framework. 448 00:18:44,877 --> 00:18:46,210 I think that's really important. 449 00:18:46,210 --> 00:18:48,180 HOST: So how do you feel about using 450 00:18:48,180 --> 00:18:53,260 AI to cross-check your own human-based deep thinking, 451 00:18:53,260 --> 00:18:53,760 right? 452 00:18:53,760 --> 00:18:57,610 So you've sort of challenged to say, carve space. 453 00:18:57,610 --> 00:18:59,050 Push yourself. 454 00:18:59,050 --> 00:19:01,110 Exercise that what's going to happen 455 00:19:01,110 --> 00:19:04,660 to my industry in my space five years and 10 years? 456 00:19:04,660 --> 00:19:06,700 How can AI help me do that? 457 00:19:06,700 --> 00:19:08,500 So I'm still going to do the hard work, 458 00:19:08,500 --> 00:19:12,720 but maybe it can fact check me or cross-reference me or keep me 459 00:19:12,720 --> 00:19:13,780 on track. 460 00:19:13,780 --> 00:19:15,780 CHRISTIAN DIPPEL: AI means many different things 461 00:19:15,780 --> 00:19:19,390 to different people, different strategic business units. 462 00:19:19,390 --> 00:19:21,300 My own perspective and a little bit 463 00:19:21,300 --> 00:19:23,730 that's shaped our conversations so far 464 00:19:23,730 --> 00:19:26,280 is when you're operating in a kind of business that 465 00:19:26,280 --> 00:19:30,090 is more data-rich, there are a lot of operational things 466 00:19:30,090 --> 00:19:31,690 that you can do with AI. 467 00:19:31,690 --> 00:19:34,380 One thing that I'm finding in just some work with clients 468 00:19:34,380 --> 00:19:37,470 is there's a lot of small and medium-sized businesses 469 00:19:37,470 --> 00:19:39,360 that have a ton of data. 470 00:19:39,360 --> 00:19:41,230 That could be manufacturing data. 471 00:19:41,230 --> 00:19:43,680 It could be order processing data, client 472 00:19:43,680 --> 00:19:45,940 facing data, all types of data. 473 00:19:45,940 --> 00:19:50,200 But up until now, it's never been worth hiring, 474 00:19:50,200 --> 00:19:52,210 let's say, two data scientists. 475 00:19:52,210 --> 00:19:54,330 Like if you're not like a data-driven enterprise 476 00:19:54,330 --> 00:19:56,860 per se, you just sort of accidentally, you might say, 477 00:19:56,860 --> 00:19:59,410 or just as a side product of your business operations, 478 00:19:59,410 --> 00:20:01,170 I think a lot of enterprises are sort 479 00:20:01,170 --> 00:20:03,750 of moving into a world where you've accumulated 480 00:20:03,750 --> 00:20:05,530 an incredible treasure trove of data, 481 00:20:05,530 --> 00:20:07,330 have never really done anything with it, 482 00:20:07,330 --> 00:20:10,750 and now, it's a matter of can I build a back end 483 00:20:10,750 --> 00:20:13,620 maybe in a very sandboxed way that 484 00:20:13,620 --> 00:20:16,710 doesn't take a lot of resources, where I can really come up 485 00:20:16,710 --> 00:20:20,310 with new business offerings or just really process 486 00:20:20,310 --> 00:20:22,410 optimization, maybe knowing which 487 00:20:22,410 --> 00:20:25,150 clients are due a maintenance, which clients are due, 488 00:20:25,150 --> 00:20:26,110 a new product, 489 00:20:26,110 --> 00:20:28,140 having an automated email pipeline 490 00:20:28,140 --> 00:20:30,480 that automates part of the outreach 491 00:20:30,480 --> 00:20:32,340 or standardizes some of the procedures 492 00:20:32,340 --> 00:20:33,940 we haven't standardized in the past. 493 00:20:33,940 --> 00:20:37,180 So that's the part where even as a small business, 494 00:20:37,180 --> 00:20:40,023 you can use AI to automate a lot of things. 495 00:20:40,023 --> 00:20:41,940 The thing that I think you're alluding to more 496 00:20:41,940 --> 00:20:46,390 is, is almost like a slightly more individualistic enterprise, 497 00:20:46,390 --> 00:20:49,440 which is you might just sit there at your desk, 498 00:20:49,440 --> 00:20:50,910 and you have to make a decision. 499 00:20:50,910 --> 00:20:54,610 And you can go speak to your colleagues at the water cooler, 500 00:20:54,610 --> 00:20:55,600 and that's useful. 501 00:20:55,600 --> 00:20:58,080 But maybe you're worried that everyone 502 00:20:58,080 --> 00:21:00,660 is drinking the same Kool-Aid to an extent when 503 00:21:00,660 --> 00:21:03,450 it comes to certain strategic business decisions. 504 00:21:03,450 --> 00:21:04,950 Now you have the ability, and that's

505 00:21:04,950 --> 00:21:07,350 a very individualistic exercise that 506 00:21:07,350 --> 00:21:10,170 is almost something that I think in a business culture can 507 00:21:10,170 --> 00:21:11,380 be encouraged. 508 00:21:11,380 --> 00:21:13,800 But early on, it's not really actually that clear 509 00:21:13,800 --> 00:21:15,670 how to even standardize it. 510 00:21:15,670 --> 00:21:17,700 Encouraging could be something as simple 511 00:21:17,700 --> 00:21:22,380 as the company will pay for your $20 a month subscription 512 00:21:22,380 --> 00:21:25,230 to one of the models to have a more premium 513 00:21:25,230 --> 00:21:28,530 access to the capabilities, like deeper thinking, deeper research 514 00:21:28,530 --> 00:21:29,138 capabilities. 515 00:21:29,138 --> 00:21:30,930 That's not a lot of money you need to spend 516 00:21:30,930 --> 00:21:32,860 to use those kind of tools. 517 00:21:32,860 --> 00:21:36,850 But what they allow you to do is really check your own thinking, 518 00:21:36,850 --> 00:21:38,238 scrutinize your own thinking. 519 00:21:38,238 --> 00:21:40,030 And of course, any individual can use that. 520 00:21:40,030 --> 00:21:42,520 You can use that to prepare for job interviews. 521 00:21:42,520 --> 00:21:45,220 You can use that to prepare for this particular interview. 522 00:21:45,220 --> 00:21:46,950 I didn't do it, but I could have tried 523 00:21:46,950 --> 00:21:49,860 to characterize Brian Benjamin to a chatbot 524 00:21:49,860 --> 00:21:52,480 and ask what kind of questions, is he going to ask me? 525 00:21:52,480 --> 00:21:54,460 And so there's a lot of capability to do that. 526 00:21:54,460 --> 00:21:57,940 And one area where I see that a lot is in the classroom now. 527 00:21:57,940 --> 00:22:01,530 Just in the last, let's say, 12 months, the ability to just 528 00:22:01,530 --> 00:22:05,640 use really any LLM, I was hesitant to just say ChatGPT. 529 00:22:05,640 --> 00:22:08,520 ChatGPT is becoming a little bit like Xerox for copying. 530 00:22:08,520 --> 00:22:10,780 And getting a really good strategic partner. 531 00:22:10,780 --> 00:22:14,250 So when I teach geopolitics, I might throw a question out there 532 00:22:14,250 --> 00:22:16,650 like, OK, imagine there's a scenario where 533 00:22:16,650 --> 00:22:19,690 we are an electric vehicle producer in Ontario, 534 00:22:19,690 --> 00:22:23,010 and there's a civil war in the Democratic Republic of Congo 535 00:22:23,010 --> 00:22:24,400 where all the cobalt comes from. 536 00:22:24,400 --> 00:22:26,850 Chances are zero people in the audience will 537 00:22:26,850 --> 00:22:28,090 have deep knowledge on this. 538 00:22:28,090 --> 00:22:30,070 But you can do a scenario analysis. 539 00:22:30,070 --> 00:22:32,040 You can describe the thing to ChatGPT 540 00:22:32,040 --> 00:22:34,587 and do a scenario analysis within five minutes. 541 00:22:34,587 --> 00:22:36,420 And within five minutes, we can have a group 542 00:22:36,420 -->

00:22:38,280 strategic discussion. 543 00:22:38,280 --> 00:22:41,910 So we get elevated in our industry knowledge 544 00:22:41,910 --> 00:22:45,180 and in our sort of high level strategic thinking and an issue. 545 00:22:45,180 --> 00:22:47,770 Within minutes, you can get elevated so much. 546 00:22:47,770 --> 00:22:50,860 And then of course, that can't be all and say all. 547 00:22:50,860 --> 00:22:54,910 There needs to be additional guardrails around the process, 548 00:22:54,910 --> 00:22:57,660 and there's all sorts of things you need to guard against and be 549 00:22:57,660 --> 00:22:58,300 worried about. 550 00:22:58,300 --> 00:23:01,990 But as a quick way of just getting deep into a topic, 551 00:23:01,990 --> 00:23:04,360 scrutinizing your own thinking, it's just incredible. 552 00:23:04,360 --> 00:23:06,870 Predictive nature can be an interesting tool 553 00:23:06,870 --> 00:23:10,450 to grab from the past and context. 554 00:23:10,450 --> 00:23:11,800 And so forth. 555 00:23:11,800 --> 00:23:17,370 How can AI be used to help me imagine 10 years out 556 00:23:17,370 --> 00:23:20,790 if I am trying to see where my industry is going 557 00:23:20,790 --> 00:23:21,970 or what's going to happen? 558 00:23:21,970 --> 00:23:24,183 Is it a tool I can use in that kind of capacity? 559 00:23:24,183 --> 00:23:26,350 CHRISTIAN DIPPEL: You often hear this one liner when 560 00:23:26,350 --> 00:23:28,725 it comes to AI, and I think there's a lot of truth in it, 561 00:23:28,725 --> 00:23:32,000 that it really raises the value of good questions. 562 00:23:32,000 --> 00:23:34,190 In other words, your job as a leader, 563 00:23:34,190 --> 00:23:36,340 whether it's at the c-suite level 564 00:23:36,340 --> 00:23:40,550 or leading a strategic business unit or whatever the case, 565 00:23:40,550 --> 00:23:43,690 I think it's going to become more about asking good questions 566 00:23:43,690 --> 00:23:47,810 and less so about being the person to provide the answers. 567 00:23:47,810 --> 00:23:50,540 And so there's a lot of debates in the AI space. 568 00:23:50,540 --> 00:23:52,690 Like with my companies, I interface more 569 00:23:52,690 --> 00:23:54,220 with the sort of technical elements 570 00:23:54,220 --> 00:23:56,860 of prompt engineering and automating 571 00:23:56,860 --> 00:24:01,030 prompt engineering through the various LLMs, APIs, et cetera. 572 00:24:01,030 --> 00:24:03,790 So asking good questions is becoming 573 00:24:03,790 --> 00:24:04,850 more and more important. 574 00:24:04,850 --> 00:24:07,730 Some of those questions will be automated questions. 575 00:24:07,730 --> 00:24:10,840 So if you just have a data feed that operationally matters 576 00:24:10,840 --> 00:24:13,450 for your company, you want to have an automated pipeline 577 00:24:13,450 --> 00:24:15,850 in the back end that queries the data feed 578 00:24:15,850 --> 00:24:17,990 and generates the answer from the data, 579 00:24:17,990 --> 00:24:19,760 and it's all about asking good questions. 580 00:24:19,760 --> 00:24:21,427 That's sort of like the engineering side 581 00:24:21,427 --> 00:24:22,490 of asking good questions. 582 00:24:22,490 --> 00:24:24,420 But I think at the more qualitative level, 583 00:24:24,420 --> 00:24:26,170 if you're just a person making a decision, 584 00:24:26,170 --> 00:24:27,670 you're sitting at your desk, and you 585 00:24:27,670 --> 00:24:31,590 can use AI as a copilot in a non-technical sense, so not 586 00:24:31,590 --> 00:24:34,810 as an explicit copilot but just as someone to run ideas by. 587 00:24:34,810 --> 00:24:37,020 So much of the way we ask questions 588 00:24:37,020 --> 00:24:38,740 is informed by our context. 589 00:24:38,740 --> 00:24:40,330 And I'll give you one example. 590 00:24:40,330 --> 00:24:43,410 When I teach an executive education class in Canada, 591 00:24:43,410 --> 00:24:46,350 and we talk about how is the labor market going 592 00:24:46,350 --> 00:24:48,400 to be impacted by AI? 593 00:24:48,400 --> 00:24:50,850 One thing that you often get is a lot of people 594 00:24:50,850 --> 00:24:53,370 assume that one area that is going 595 00:24:53,370 --> 00:24:55,650 to be relatively unaffected by AI 596 00:24:55,650 --> 00:24:58,740 is old age care because that is something 597 00:24:58,740 --> 00:25:00,730 that requires the human touch. 598 00:25:00,730 --> 00:25:03,005 And that might be entirely true. 599 00:25:03,005 --> 00:25:05,130 But what I found really interesting is the contrast 600 00:25:05,130 --> 00:25:08,170 to if you talk to people in East Asia, let's say in Hong Kong, 601 00:25:08,170 --> 00:25:11,970 about AI, that is their number one use case for AI, 602 00:25:11,970 --> 00:25:13,210 is old age care. 603 00:25:13,210 --> 00:25:16,050 Because the demographics are so fundamentally different 604 00:25:16,050 --> 00:25:17,620 in those two geographies. 605 00:25:17,620 --> 00:25:19,930 We have relatively young demographics. 606 00:25:19,930 --> 00:25:22,390 Of course, we talk about aging and things like that. 607 00:25:22,390 --> 00:25:25,840 But it's not a very pressing concern in a macroeconomic sense 608 00:25:25,840 --> 00:25:28,510 when you look at our population trees in North America. 609 00:25:28,510 --> 00:25:30,210 When you look at population trees 610 00:25:30,210 --> 00:25:34,240 in Hong Kong, Japan, Singapore, oh, my gosh, 611 00:25:34,240 --> 00:25:35,890 they look very different from ours. 612 00:25:35,890 --> 00:25:40,050 And old age care is viewed as the number one-- well, maybe not 613 00:25:40,050 --> 00:25:42,030 the number one, but one of the number one 614 00:25:42,030 --> 00:25:45,600 use cases where I can actually help in the form of literally 615 00:25:45,600 --> 00:25:46,990 AI-informed robotics. 616 00:25:46,990 --> 00:25:50,790 And so it's just striking how people's predictions 617 00:25:50,790 --> 00:25:53,490 about future use cases are so

socially 618 00:25:53,490 --> 00:25:55,060 informed by the context. 619 00:25:55,060 --> 00:25:57,940 And I think AI can help you question your own assumptions. 620 00:25:57,940 --> 00:25:59,620 In when you're asking those questions. 621 00:25:59,620 --> 00:26:00,480 HOST: You're right. 622 00:26:00,480 --> 00:26:03,280 We kind of naturally go to what we know. 623 00:26:03,280 --> 00:26:04,920 And one would hope that we can learn 624 00:26:04,920 --> 00:26:08,400 from each other because as things shift in North America 625 00:26:08,400 --> 00:26:12,378 and aging does become an even bigger sort of prominent piece, 626 00:26:12,378 --> 00:26:14,670 what have we learned from other parts of the world that 627 00:26:14,670 --> 00:26:16,270 have already been there first? 628 00:26:16,270 --> 00:26:21,180 I wrote down, you're asking good questions, 629 00:26:21,180 --> 00:26:23,500 and we talk about prompt engineers. 630 00:26:23,500 --> 00:26:26,460 I'm fortunate to work with a lot of exceptional coaches 631 00:26:26,460 --> 00:26:27,730 here at Ivey. 632 00:26:27,730 --> 00:26:30,210 So I've had the pleasure of being 633 00:26:30,210 --> 00:26:32,040 asked really good questions. 634 00:26:32,040 --> 00:26:37,780 Very specific, often open-ended, leaving space for conversation. 635 00:26:37,780 --> 00:26:40,720 And so I think that your comment and maybe the art 636 00:26:40,720 --> 00:26:44,820 of good questions is going to be an even more prominent skill 637 00:26:44,820 --> 00:26:48,540 that leaders are going to need to hone and to build 638 00:26:48,540 --> 00:26:50,110 and to refine. 639 00:26:50,110 --> 00:26:53,250 So I'm going to go back a little bit to the conversation 640 00:26:53,250 --> 00:26:56,860 around sort of executives and maybe even park executives. 641 00:26:56,860 --> 00:27:00,060 I think leaders at multiple levels in the organization 642 00:27:00,060 --> 00:27:02,530 are involved in strategic planning. 643 00:27:02,530 --> 00:27:03,160 So yes. 644 00:27:03,160 --> 00:27:05,050 It can be organizational strategic planning, 645 00:27:05,050 --> 00:27:08,650 but it also be departmental or even team strategic planning. 646 00:27:08,650 --> 00:27:11,850 So when we hear about AI initiatives, most strategic 647 00:27:11,850 --> 00:27:14,670 planning exercises, we got to leverage AI 648 00:27:14,670 --> 00:27:16,540 to some capacity or another. 649 00:27:16,540 --> 00:27:21,250 So let's toggle between is this sort of a process innovation? 650 00:27:21,250 --> 00:27:24,790 So can AI save me time? 651 00:27:24,790 --> 00:27:26,170 Can it save me money? 652 00:27:26,170 --> 00:27:28,800 Can it create speed and efficiency 653 00:27:28,800 --> 00:27:33,030 with you sort of reshaping maybe our business model 654 00:27:33,030 --> 00:27:35,650 or our team model altogether? 655 00:27:35,650 --> 00:27:38,080 CHRISTIAN DIPPEL: I used to have a little bit this idea, 656 00:27:38,080 --> 00:27:40,210 I think, early

on in thinking about AI, 657 00:27:40,210 --> 00:27:42,720 that product innovation is somehow 658 00:27:42,720 --> 00:27:44,770 better than process innovation. 659 00:27:44,770 --> 00:27:47,500 Process innovation, it feels a little boring. 660 00:27:47,500 --> 00:27:50,920 It's sort of like you're wringing the towel dry, 661 00:27:50,920 --> 00:27:52,840 like getting the efficiencies you can get, 662 00:27:52,840 --> 00:27:55,390 but product innovation is really the exciting part. 663 00:27:55,390 --> 00:27:57,550 But I've really changed my thinking on that. 664 00:27:57,550 --> 00:28:00,510 And actually, some of our colleagues in Hong Kong 665 00:28:00,510 --> 00:28:03,070 had a helping hand in a discussion. 666 00:28:03,070 --> 00:28:05,070 I remember well where the framework that 667 00:28:05,070 --> 00:28:06,840 was applied in that conversation was more 668 00:28:06,840 --> 00:28:10,830 about where is AI a homogenizer, and where 669 00:28:10,830 --> 00:28:12,220 is it a differentiator? 670 00:28:12,220 --> 00:28:17,040 And the chatbot is a great example of a product innovation 671 00:28:17,040 --> 00:28:19,600 that is not going to be a differentiator. 672 00:28:19,600 --> 00:28:25,420 Chatbot technology is a very commoditized type of technology. 673 00:28:25,420 --> 00:28:27,340 I've been involved in building chatbots. 674 00:28:27,340 --> 00:28:29,200 It's a very standardized product. 675 00:28:29,200 --> 00:28:31,630 And yes, there's better ones and less good ones. 676 00:28:31,630 --> 00:28:34,030 And yes, we've seen with Air Canada, 677 00:28:34,030 --> 00:28:36,240 we've seen examples of it horribly 678 00:28:36,240 --> 00:28:38,460 backfiring because they didn't execute 679 00:28:38,460 --> 00:28:39,952 very well on their chatbot. 680 00:28:39,952 --> 00:28:42,160 But at the end of the day, it's not a differentiator. 681 00:28:42,160 --> 00:28:45,730 No business will stand out for having a chatbot. 682 00:28:45,730 --> 00:28:48,640 Some businesses might be worried that in the short run, 683 00:28:48,640 --> 00:28:51,190 they might stand out negatively for not having one. 684 00:28:51,190 --> 00:28:53,890 I think those are relatively minor concerns. 685 00:28:53,890 --> 00:28:56,550 At the same time, sometimes, the process innovation 686 00:28:56,550 --> 00:29:00,640 is your real differentiator and your real business model. 687 00:29:00,640 --> 00:29:04,360 So when you think about famous examples like IKEA, Walmart, 688 00:29:04,360 --> 00:29:09,160 to a lesser extent, the network dynamics of Netflix or Amazon, 689 00:29:09,160 --> 00:29:11,760 those are really all businesses where the business model 690 00:29:11,760 --> 00:29:14,550 is on the process innovation more than the product 691 00:29:14,550 --> 00:29:15,400 innovation. 692 00:29:15,400 --> 00:29:19,540 And so that is really an area where my thinking has changed 693 00:29:19,540 --> 00:29:22,390 from thinking product innovation somehow 694 00:29:22,390 --> 00:29:25,930 better than process innovation, to really being agnostic 695 00:29:25,930 --> 00:29:28,540 about the two and thinking about where is 696 00:29:28,540 --> 00:29:30,350 the true competitive advantage. 697 00:29:30,350 --> 00:29:33,310 Is it plausible for us to build a product that really sets us 698 00:29:33,310 --> 00:29:35,170 apart, or is it maybe the case that there's 699 00:29:35,170 --> 00:29:36,880 something in our processes that is 700 00:29:36,880 --> 00:29:39,640 going to really be able to leverage AI in a way that 701 00:29:39,640 --> 00:29:40,730 sets us apart? 702 00:29:40,730 --> 00:29:44,020 HOST: We're often looking for the bigger splash or the bigger 703 00:29:44,020 --> 00:29:46,060 bold innovation, and maybe it's actually 704 00:29:46,060 --> 00:29:48,070 something a little closer to home 705 00:29:48,070 --> 00:29:50,090 that we need to pay attention to. 706 00:29:50,090 --> 00:29:52,690 I want to dig in because you're involved 707 00:29:52,690 --> 00:29:56,870 in a lot of really cool ventures and sort of consulting work. 708 00:29:56,870 --> 00:30:03,520 A couple of live examples of where AI is being strategically 709 00:30:03,520 --> 00:30:04,160 applied. 710 00:30:04,160 --> 00:30:06,110 So it's not hype. 711 00:30:06,110 --> 00:30:08,410 So don't tell me about the ChatGPT bot 712 00:30:08,410 --> 00:30:11,240 that Company A implemented. 713 00:30:11,240 --> 00:30:15,800 Something that be interesting for our listeners 714 00:30:15,800 --> 00:30:18,690 to hear that you've been involved with firsthand. 715 00:30:18,690 --> 00:30:20,690 CHRISTIAN DIPPEL: One example that comes to mind 716 00:30:20,690 --> 00:30:23,270 is in one of my own businesses, which is really 717 00:30:23,270 --> 00:30:25,940 a newsletter business, that takes an enormous amount 718 00:30:25,940 --> 00:30:27,480 of government text. 719 00:30:27,480 --> 00:30:31,400 And we knew with that idea that no one really 720 00:30:31,400 --> 00:30:34,040 wants to have a newsletter that is everything 721 00:30:34,040 --> 00:30:35,637 that is going on in Ottawa. 722 00:30:35,637 --> 00:30:37,220 No one really needs to know everything 723 00:30:37,220 --> 00:30:39,000 that is going on in Ottawa. 724 00:30:39,000 --> 00:30:41,070 The value really comes from saying, 725 00:30:41,070 --> 00:30:42,770 if you're in oil and gas, you want 726 00:30:42,770 --> 00:30:44,480 to have a newsletter that says everything 727 00:30:44,480 --> 00:30:47,720 that's going on in Ottawa that affects the oil and gas 728 00:30:47,720 --> 00:30:48,360 industry. 729 00:30:48,360 --> 00:30:50,712 If you're in utilities, you want a newsletter for that. 730 00:30:50,712 --> 00:30:52,920 If you're in telecoms, you want a newsletter in that. 731 00:30:52,920 --> 00:30:55,128 If you're in finance, you want a newsletter for that. 732 00:30:55,128 --> 00:30:58,580 And so we had this idea, but we had this forest for the trees 733

00:30:58,580 --> 00:31:00,770 problem, that it was, well, government just 734 00:31:00,770 --> 00:31:04,760 generates too much text, like the editorial team that it would 735 00:31:04,760 --> 00:31:09,410 require to like, parse all those things into 20 different topics 736 00:31:09,410 --> 00:31:10,530 is just enormous. 737 00:31:10,530 --> 00:31:15,310 And it's a good actually example of this process versus product 738 00:31:15,310 --> 00:31:18,610 innovation, and how fuzzy that line sometimes can be with AI 739 00:31:18,610 --> 00:31:21,320 is we have a very, very simple AI stack. 740 00:31:21,320 --> 00:31:24,820 All it does is take all the data that we're ingesting 741 00:31:24,820 --> 00:31:28,190 and funnel it into 20 plus funnels. 742 00:31:28,190 --> 00:31:29,900 The technology is not that hard. 743 00:31:29,900 --> 00:31:32,150 It's not a particularly sophisticated model. 744 00:31:32,150 --> 00:31:34,160 It's using very off the shelf tools. 745 00:31:34,160 --> 00:31:36,310 But it allows you to go from 0 to 1 746 00:31:36,310 --> 00:31:39,560 in being able to not being able to do it at all, 747 00:31:39,560 --> 00:31:42,680 to being able to do it in an extremely automated way. 748 00:31:42,680 --> 00:31:45,280 And so now, we're in a place where I sit there on a Sunday 749 00:31:45,280 --> 00:31:49,040 night, editing 24 newsletters in the space of an hour and a half. 750 00:31:49,040 --> 00:31:51,190 Everything else is AI generated, and I just 751 00:31:51,190 --> 00:31:53,690 look through them, do a little bit of polish, and that's it. 752 00:31:53,690 --> 00:31:56,710 So your capability as a small business 753 00:31:56,710 --> 00:31:58,832 to do an enormous amount, obviously, 754 00:31:58,832 --> 00:32:00,290 depending on your line of business, 755 00:32:00,290 --> 00:32:02,900 it's harder in hardware than it is in software. 756 00:32:02,900 --> 00:32:05,710 In certain applications, you get completely supercharged in what 757 00:32:05,710 --> 00:32:07,510 you can do, and a lot of it can be 758 00:32:07,510 --> 00:32:09,100 much more about the process innovation 759 00:32:09,100 --> 00:32:10,400 than the product innovation. 760 00:32:10,400 --> 00:32:11,950 HOST: Yeah, interesting. 761 00:32:11,950 --> 00:32:13,760 You think about an hour and a half. 762 00:32:13,760 --> 00:32:16,490 What would it have taken without AI support? 763 00:32:16,490 --> 00:32:17,870 How long would it have been? 764 00:32:17,870 --> 00:32:18,370 CHRISTIAN DIPPEL: Well, it's almost 765 00:32:18,370 --> 00:32:20,350 like it's your process is changed in a way 766 00:32:20,350 --> 00:32:22,372 where it would have been straight up impossible 767 00:32:22,372 --> 00:32:24,830 because you would have had to see the forest for the trees. 768 00:32:24,830 --> 00:32:26,620 But there were too many trees, and you could have never 769 00:32:26,620 --> 00:32:27,710 done it to begin with. 770 00:32:27,710 --> 00:32:29,920 Then there is a

maybe intermediate solution 771 00:32:29,920 --> 00:32:32,440 that's like six years ago technology where 772 00:32:32,440 --> 00:32:34,220 you define a set of keywords. 773 00:32:34,220 --> 00:32:37,730 You parse through all the text based on buckets of keywords 774 00:32:37,730 --> 00:32:39,802 and to try to do an initial funneling. 775 00:32:39,802 --> 00:32:41,260 But oh, my gosh, you would have had 776 00:32:41,260 --> 00:32:44,300 a lot of work left to do to then make that good. 777 00:32:44,300 --> 00:32:45,910 And then you need to still summarize 778 00:32:45,910 --> 00:32:48,190 all the content, which four or five years ago, would 779 00:32:48,190 --> 00:32:49,880 have been very hard to do. 780 00:32:49,880 --> 00:32:51,805 There would have been a lot of hallucinations, 781 00:32:51,805 --> 00:32:53,930 the quality of the text would have been horrendous. 782 00:32:53,930 --> 00:32:55,960 And so all of those problems just go away. 783 00:32:55,960 --> 00:32:58,700 And you're left with just polishing, editing work. 784 00:32:58,700 --> 00:32:59,200 HOST: Yeah. 785 00:32:59,200 --> 00:33:01,540 So saving time on something that you might not 786 00:33:01,540 --> 00:33:05,110 have even been able to do because of the complexity 787 00:33:05,110 --> 00:33:05,830 involved. 788 00:33:05,830 --> 00:33:09,850 So what's one single takeaway that you would leave, something 789 00:33:09,850 --> 00:33:13,030 that I can do as a leader, when I'm 790 00:33:13,030 --> 00:33:16,720 thinking about AI in the context of creating 791 00:33:16,720 --> 00:33:18,290 competitive advantage? 792 00:33:18,290 --> 00:33:20,590 CHRISTIAN DIPPEL: I think one big takeaway 793 00:33:20,590 --> 00:33:25,750 that I think a lot of businesses, especially 794 00:33:25,750 --> 00:33:28,760 businesses that interface with Ivey a lot, 795 00:33:28,760 --> 00:33:30,910 I think have traditionally operated 796 00:33:30,910 --> 00:33:33,580 on a model where a wait and see approach 797 00:33:33,580 --> 00:33:37,070 is often the right approach when it comes to new technologies. 798 00:33:37,070 --> 00:33:39,040 And you know, in academia, the fancy language 799 00:33:39,040 --> 00:33:41,900 for that is like the real option value of waiting. 800 00:33:41,900 --> 00:33:43,910 And oftentimes, with new technologies, 801 00:33:43,910 --> 00:33:46,870 the idea is, well, some standardized gold standard 802 00:33:46,870 --> 00:33:48,890 is going to emerge. 803 00:33:48,890 --> 00:33:51,700 And maybe I need to wait for two or three years for the dust 804 00:33:51,700 --> 00:33:52,490 to settle. 805 00:33:52,490 --> 00:33:54,160 But then there'll be an off the shelf 806 00:33:54,160 --> 00:33:56,840 product that will do what I need to do. 807 00:33:56,840 --> 00:33:58,250 And so it's better to wait. 808 00:33:58,250 --> 00:34:00,620 And especially in the world of atoms, 809 00:34:00,620 --> 00:34:02,630 like in the physical hardware world, 810 00:34:02,630 -->

00:34:04,390 that is often the case, right? 811 00:34:04,390 --> 00:34:06,250 We all know Moore's law. 812 00:34:06,250 --> 00:34:09,110 The cost of computing decreases exponentially. 813 00:34:09,110 --> 00:34:11,469 Let's say things like the cost of solar panels 814 00:34:11,469 --> 00:34:13,510 has also been decreasing exponentially 815 00:34:13,510 --> 00:34:14,510 over the last 10 years. 816 00:34:14,510 --> 00:34:16,219 So lots of technologies. 817 00:34:16,219 --> 00:34:18,800 Absolutely, it makes sense to wait and see. 818 00:34:18,800 --> 00:34:21,159 I think AI is really different, and it's 819 00:34:21,159 --> 00:34:22,630 different for two reasons. 820 00:34:22,630 --> 00:34:25,047 I'm sure it's different for more than two reasons, but two 821 00:34:25,047 --> 00:34:26,630 that are sort of prominent in my head. 822 00:34:26,630 --> 00:34:28,820 One is it just keeps accelerating. 823 00:34:28,820 --> 00:34:30,440 And so if you wait and do nothing, 824 00:34:30,440 --> 00:34:33,850 the thing is going to accelerate away from you in a way 825 00:34:33,850 --> 00:34:35,449 that you might come to regret two 826 00:34:35,449 --> 00:34:36,730 or three years down the road. 827 00:34:36,730 --> 00:34:38,230 And then the other thing is the cost 828 00:34:38,230 --> 00:34:40,659 of an option is just so darn low compared 829 00:34:40,659 --> 00:34:42,560 to most other technologies. 830 00:34:42,560 --> 00:34:45,550 All it really takes is start using the thing 831 00:34:45,550 --> 00:34:47,840 in a potentially very low tech way, 832 00:34:47,840 --> 00:34:51,550 maybe designating two or three critical people 833 00:34:51,550 --> 00:34:54,920 in your organizations to say, explore some options. 834 00:34:54,920 --> 00:34:56,800 Free up one day a week where you say 835 00:34:56,800 --> 00:35:00,400 today is like your moonshot research day, where you just 836 00:35:00,400 --> 00:35:02,390 think about things we can do with AI. 837 00:35:02,390 --> 00:35:06,070 So the barrier to adoption is really so low 838 00:35:06,070 --> 00:35:07,540 that I think it really makes sense 839 00:35:07,540 --> 00:35:11,380 to engage today and not wait, even for organizations that 840 00:35:11,380 --> 00:35:14,470 have been well-served in the past, with taking a wait 841 00:35:14,470 --> 00:35:17,883 and see approach when it comes to adopting new technologies. 842 00:35:17,883 --> 00:35:19,300 And then the second thing is going 843 00:35:19,300 --> 00:35:20,990 to be quicker because I already said it. 844 00:35:20,990 --> 00:35:23,050 I think a lot of people I find in conversations 845 00:35:23,050 --> 00:35:25,930 grapple with this tension, that this idea that AI 846 00:35:25,930 --> 00:35:27,140 is all about data. 847 00:35:27,140 --> 00:35:29,500 But thinking about what AI will do 848 00:35:29,500 --> 00:35:32,300 is all about qualitative imagination. 849 00:35:32,300 --> 00:35:35,020 And I feel like that creates a real barrier to

engaging 850 00:35:35,020 --> 00:35:37,190 with it, and I think getting comfortable with that, 851 00:35:37,190 --> 00:35:40,790 with the idea that, yes, AI is all about transforming data, 852 00:35:40,790 --> 00:35:43,390 but that doesn't mean that you can't just sit there 853 00:35:43,390 --> 00:35:45,400 and think carefully about what AI 854 00:35:45,400 --> 00:35:48,170 will do to your business or your industry. 855 00:35:48,170 --> 00:35:50,680 HOST: And Thank you for sort of making the comparison 856 00:35:50,680 --> 00:35:54,350 to sometimes, it did make sense to wait and see, 857 00:35:54,350 --> 00:35:57,020 let the dust settle, let the kinks get worked out. 858 00:35:57,020 --> 00:35:59,200 And whereas other cases, I sort of 859 00:35:59,200 --> 00:36:01,090 pictured this train going down the tracks, 860 00:36:01,090 --> 00:36:03,730 and it's picking up speed, and it's like, I'm not ever 861 00:36:03,730 --> 00:36:05,050 going to catch up. 862 00:36:05,050 --> 00:36:07,280 I gotta jump now and get on. 863 00:36:07,280 --> 00:36:11,260 For individuals, do you feel it's worth 864 00:36:11,260 --> 00:36:13,910 paying for premium AI models. 865 00:36:13,910 --> 00:36:16,218 And if so, or if not, do you have a favorite? 866 00:36:16,218 --> 00:36:18,760 CHRISTIAN DIPPEL: One of the things that I found most amazing 867 00:36:18,760 --> 00:36:20,570 is in working with engineers. 868 00:36:20,570 --> 00:36:22,970 When you use something like GitHub Copilot, 869 00:36:22,970 --> 00:36:25,690 GitHub Copilot is like integrated with five or six 870 00:36:25,690 --> 00:36:29,290 different AI models, and just seeing an engineer switching 871 00:36:29,290 --> 00:36:31,300 between models and one model will 872 00:36:31,300 --> 00:36:33,800 be incredible at changing your code. 873 00:36:33,800 --> 00:36:37,090 Another model will be incredible at giving you code from scratch 874 00:36:37,090 --> 00:36:39,200 if you're just describing what you want to do. 875 00:36:39,200 --> 00:36:41,710 And so there's a lot of nuance, and I 876 00:36:41,710 --> 00:36:44,630 think the deeper down that rabbit hole you go, 877 00:36:44,630 --> 00:36:46,123 the more those nuances matter. 878 00:36:46,123 --> 00:36:48,040 But then at the end of the day, it's up to you 879 00:36:48,040 --> 00:36:50,420 to decide if those nuances matter to you. 880 00:36:50,420 --> 00:36:53,470 So I routinely use four different models 881 00:36:53,470 --> 00:36:54,980 just in conversation. 882 00:36:54,980 --> 00:36:56,780 I have premium subscriptions to two. 883 00:36:56,780 --> 00:36:59,600 I don't have premium subscriptions to the other two, 884 00:36:59,600 --> 00:37:00,980 and I just experiment. 885 00:37:00,980 --> 00:37:04,000 And certain models I find very little difference between them 886 00:37:04,000 --> 00:37:05,488 depending on the context too. 887 00:37:05,488 --> 00:37:07,030 So if you just ask, here's the things 888 00:37:07,030 --> 00:37:08,450 I have left over in my fridge. 889 00:37:08,450 --> 00:37:10,460 Give me some suggestions for dinner tonight, 890 00:37:10,460 --> 00:37:13,220 it's not going to matter all that much what you use it for. 891 00:37:13,220 --> 00:37:15,850 If you want to do like really deep thinking 892 00:37:15,850 --> 00:37:18,400 analysis of an industry profile, maybe there 893 00:37:18,400 --> 00:37:20,410 will be some real differences between a premium 894 00:37:20,410 --> 00:37:22,070 model and a non-premium model. 895 00:37:22,070 --> 00:37:24,233 HOST: So what I take away is it depends. 896 00:37:24,233 --> 00:37:25,150 CHRISTIAN DIPPEL: Yes. 897 00:37:25,150 --> 00:37:25,940 It depends. 898 00:37:25,940 --> 00:37:27,790 And maybe further to that point, I 899 00:37:27,790 --> 00:37:30,220 don't think you need professional advice 900 00:37:30,220 --> 00:37:34,260 on the decision of whether to spend $20 a month on a premium 901 00:37:34,260 --> 00:37:34,760 model. 902 00:37:34,760 --> 00:37:37,910 Just spend $20 on a premium model, and if you don't like it, 903 00:37:37,910 --> 00:37:39,470 unsubscribe a month later. 904 00:37:39,470 --> 00:37:41,608 It's not an irreversible decision. 905 00:37:41,608 --> 00:37:43,900 HOST: You know, someone might like this particular tool 906 00:37:43,900 --> 00:37:46,940 because they're using it for this very specific reason, 907 00:37:46,940 --> 00:37:51,550 whereas someone else might have success with a different tool 908 00:37:51,550 --> 00:37:54,320 because again, it's a different purpose. 909 00:37:54,320 --> 00:37:55,910 And given how fast things are going, 910 00:37:55,910 --> 00:37:59,170 I'm sure new tools will be emerging all the time. 911 00:37:59,170 --> 00:38:02,020 CHRISTIAN DIPPEL: To your point, in a classroom setting, if we 912 00:38:02,020 --> 00:38:03,895 have a session where we actually say, OK, now 913 00:38:03,895 --> 00:38:05,540 everyone go and prompt an AI. 914 00:38:05,540 --> 00:38:07,220 People tend to use very different ones. 915 00:38:07,220 --> 00:38:09,070 Some people are on their work laptop. 916 00:38:09,070 --> 00:38:12,380 And so they often use just Microsoft Copilot. 917 00:38:12,380 --> 00:38:14,170 Other people use GPT, some people 918 00:38:14,170 --> 00:38:15,860 use Perplexity, Claude, whatever. 919 00:38:15,860 --> 00:38:18,975 You get a pretty wide range of results in the answers. 920 00:38:18,975 --> 00:38:20,350 But then at the same time, if you 921 00:38:20,350 --> 00:38:22,510 ask the same question of the same LLM 922 00:38:22,510 --> 00:38:25,640 5 minutes later, you can also get a wide range of results. 923 00:38:25,640 --> 00:38:26,750 HOST: It's like us, right? 924 00:38:26,750 --> 00:38:28,610 If you ask me the same question tomorrow, 925 00:38:28,610 --> 00:38:31,480 I might get a different answer depending 926 00:38:31,480 --> 00:38:34,180 on what we were talking about. 927 00:38:34,180 --> 00:38:36,770 Thanks for listening to Learning in Action. 928 00:38:36,770 --> 00:38:40,400 If Christian's insights sparked new thinking on AI strategies, 929 00:38:40,400 --> 00:38:41,570 don't stop here. 930 00:38:41,570 --> 00:38:46,120 Head over to ivexeced.com to explore upcoming programs 931 00:38:46,120 --> 00:38:48,440 and deepen your fluency with AI. 932 00:38:48,440 --> 00:38:51,520 And follow us on your favorite social media platform 933 00:38:51,520 --> 00:38:55,210 to catch more real time thought leadership and exclusive podcast 934 00:38:55,210 --> 00:38:55,940 drops. 935 00:38:55,940 --> 00:38:58,380 We'll see you next time. 936 00:38:58,380 --> 00:39:02,000