1 00:00:00,000 --> 00:00:01,080 Welcome back everybody. 2 00:00:01,080 --> 00:00:04,160 Ready to dive into more AI goodness. 3 00:00:04,160 --> 00:00:07,480 Today, it's all about generative AI 4 00:00:07,480 --> 00:00:09,720 and how it's really shaking things up 5 00:00:09,720 --> 00:00:10,560 in the business world. 6 00:00:10,560 --> 00:00:11,720 Yeah, it's getting real now. 7 00:00:11,720 --> 00:00:12,560 Definitely. 8 00:00:12,560 --> 00:00:15,600 And we've got two really interesting reports 9 00:00:15,600 --> 00:00:17,200 to kind of unpack and look at. 10 00:00:17,200 --> 00:00:20,600 We've got the Deloitte state of generative AI 11 00:00:20,600 --> 00:00:22,540 in the enterprise report and then some insights 12 00:00:22,540 --> 00:00:25,680 from this MIT Technology Review Insights report 13 00:00:25,680 --> 00:00:27,280 called the Great Acceleration, 14 00:00:27,280 --> 00:00:29,720 CIO Perspectives on Generative AI. 15 00:00:29,720 --> 00:00:31,680 And the interesting thing here is that 16 00:00:31,680 --> 00:00:34,200 we're past the hype phase. 17 00:00:34,200 --> 00:00:36,800 AI is not just this shiny new toy anymore. 18 00:00:36,800 --> 00:00:38,560 It's like companies are really, 19 00:00:38,560 --> 00:00:40,000 they're starting to demand results, right? 20 00:00:40,000 --> 00:00:42,880 They're looking for how is this actually gonna 21 00:00:42,880 --> 00:00:43,960 help my bottom line? 22 00:00:43,960 --> 00:00:45,200 How's this gonna, you know? 23 00:00:45,200 --> 00:00:46,520 It's like the honeymoon's over, right? 24 00:00:46,520 --> 00:00:48,880 Exactly, yeah, the honeymoon's over and it's like, okay. 25 00:00:48,880 --> 00:00:49,720 Show me the money. 26 00:00:49,720 --> 00:00:51,080 Show me the money, yeah, exactly. 27 00:00:51,080 --> 00:00:52,960 And both of these reports are kind of hinting 28 00:00:52,960 --> 00:00:54,960 that while there's a ton of enthusiasm, 29 00:00:54,960 --> 00:00:58,080 obviously tons of investment going into this area 30 00:00:58,080 --> 00:01:00,280 actually scaling up those AI projects, 31 00:01:00,280 --> 00:01:02,200 it's proving to be a real challenge. 32 00:01:02,200 --> 00:01:03,960 Yeah, it's that gap between the potential 33 00:01:03,960 --> 00:01:06,120 and the actual rubber hitting the road, so to speak. 34 00:01:06,120 --> 00:01:08,720 Exactly, and the Deloitte report actually found 35 00:01:08,720 --> 00:01:12,560 that a whopping 67% of companies are increasing 36 00:01:12,560 --> 00:01:14,600 their investments in generative AI. 37 00:01:14,600 --> 00:01:18,080 But here's the catch, almost 70% haven't even been able 38 00:01:18,080 --> 00:01:21,600 to move 30% of their AI experiments into full production. 39 00:01:21,600 --> 00:01:23,760 So like everyone's excited about it, 40 00:01:23,760 --> 00:01:25,640 but then actually getting it to work, you know? 41 00:01:25,640 --> 00:01:26,720 Yeah, we can get real. 42 00:01:26,720 --> 00:01:27,560 At scale. 43 00:01:27,560 --> 00:01:28,400 Yeah. 44 00:01:28,400 --> 00:01:29,280 That's the tough part. 45 00:01:29,280 --> 00:01:31,000 Yeah, that's a hurdle. 46 00:01:31,000 --> 00:01:34,480 And I think the MIT report gives us some good perspective 47 00:01:34,480 --> 00:01:38,080 here because it seems that even before we had this whole 48 00:01:38,080 --> 00:01:40,920 explosion of generative AI, a lot of businesses 49 00:01:40,920 --> 00:01:44,960 were already facing hurdles when it came to integrating AI, 50 00:01:44,960 --> 00:01:47,360 any kind of AI to their core operations 51 00:01:47,360 --> 00:01:49,760 and generative AI just kind of like amplified 52 00:01:49,760 --> 00:01:51,120 those challenges. 53 00:01:51,120 --> 00:01:53,880 So it's not just a generative AI problem, 54 00:01:53,880 --> 00:01:55,000 it's an AI problem. 55 00:01:55,000 --> 00:01:57,640 Yeah, it's a bigger AI adoption problem, 56 00:01:57,640 --> 00:01:58,680 I think that's right. 57 00:01:58,680 --> 00:01:59,920 And what is it they're saying? 58 00:01:59,920 --> 00:02:03,440 The reports are saying data is a big part of that? 59 00:02:03,440 --> 00:02:06,800 Yeah, the lack of the right kind of data, 60 00:02:06,800 --> 00:02:08,600 I think that's a huge piece of it. 61 00:02:08,600 --> 00:02:11,080 And the Deloitte report actually mentions 62 00:02:11,080 --> 00:02:14,160 that something like 75% of organizations 63 00:02:14,160 --> 00:02:18,360 have increased their investments in data lifecycle 64 00:02:18,360 --> 00:02:21,160 management specifically to support 65 00:02:21,160 --> 00:02:22,640 their generative AI initiatives. 66 00:02:22,640 --> 00:02:24,320 OK, so data lifecycle management now, 67 00:02:24,320 --> 00:02:27,560 for those of us who aren't data scientists, 68 00:02:27,560 --> 00:02:29,160 break that down, what does that even mean? 69 00:02:29,160 --> 00:02:30,160 Yeah, sure. 70 00:02:30,160 --> 00:02:33,560 So think of it as the entire journey of data. 71 00:02:33,560 --> 00:02:35,040 Within a company, it's everything 72 00:02:35,040 --> 00:02:38,720 from how you collect it, how you store it, how you clean it, 73 00:02:38,720 --> 00:02:41,200 how you analyze it, and then ultimately how you use it 74 00:02:41,200 --> 00:02:42,600 to actually make decisions. 75 00:02:42,600 --> 00:02:44,240 So it's not just about having a lot of data, 76 00:02:44,240 --> 00:02:46,640 it's about having the right processes and systems 77 00:02:46,640 --> 00:02:48,120 to manage that data well. 78 00:02:48,120 --> 00:02:49,040 Yeah, exactly. 79 00:02:49,040 --> 00:02:51,240 And this is where it gets really tricky for a lot of companies, 80 00:02:51,240 --> 00:02:53,560 because these generative AI models, 81 00:02:53,560 --> 00:02:55,720 they're just incredibly data hungry. 82 00:02:55,720 --> 00:02:58,320 They need high quality, well-organized data 83 00:02:58,320 --> 00:02:59,840 to actually function properly. 84 00:02:59,840 --> 00:03:02,080 It's like trying to bake a cake without the right ingredients 85 00:03:02,080 --> 00:03:03,920 or knowing how to use them properly. 86 00:03:03,920 --> 00:03:05,280 You just end up with a mess. 87 00:03:05,280 --> 00:03:06,760 Yeah, that's a great analogy. 88 00:03:06,760 --> 00:03:07,480 It is like that. 89 00:03:07,480 --> 00:03:10,040 And the consequences can be pretty significant too. 90 00:03:10,040 --> 00:03:12,440 I mean, the Deloitte report found that over half of the companies, 91 00:03:12,440 --> 00:03:17,160 they surveyed 55% admitted to actually avoiding certain AI 92 00:03:17,160 --> 00:03:20,200 projects altogether just because of all these data roadblocks. 93 00:03:20,200 --> 00:03:21,400 Yeah, that's a big number. 94 00:03:21,400 --> 00:03:22,960 So data is a big bottleneck. 95 00:03:22,960 --> 00:03:24,720 But then what about the risks of AI? 96 00:03:24,720 --> 00:03:29,080 Right, we hear about algorithmic bias, data privacy, all of that. 97 00:03:29,080 --> 00:03:30,720 Yeah, that's a key point. 98 00:03:30,720 --> 00:03:33,840 And it's something that both reports address, 99 00:03:33,840 --> 00:03:37,680 specifically the need for really strong governance frameworks 100 00:03:37,680 --> 00:03:40,120 to help mitigate those risks. 101 00:03:40,120 --> 00:03:41,520 OK, well, let's unpack that a little bit. 102 00:03:41,520 --> 00:03:44,120 Because what does governance, when we're talking about AI, 103 00:03:44,120 --> 00:03:45,560 what does that actually mean? 104 00:03:45,560 --> 00:03:48,840 Yeah, so think of it as this set of rules, processes, 105 00:03:48,840 --> 00:03:51,600 guidelines that companies put in place to make sure 106 00:03:51,600 --> 00:03:54,320 that their AI systems are developed and used responsibly. 107 00:03:54,320 --> 00:03:56,880 And that includes things like protecting data privacy, 108 00:03:56,880 --> 00:03:59,920 preventing algorithmic bias, and really making sure 109 00:03:59,920 --> 00:04:03,560 that those AI systems are aligned with the company's values 110 00:04:03,560 --> 00:04:05,560 and ethical principles. 111 00:04:05,560 --> 00:04:07,760 So it's not just making sure that the AI works, 112 00:04:07,760 --> 00:04:12,000 it's making sure it's working in an ethical way. 113 00:04:12,000 --> 00:04:12,480 That's right. 114 00:04:12,480 --> 00:04:15,920 And it gets interesting here because there's often this gap 115 00:04:15,920 --> 00:04:18,800 between recognizing that you need governance 116 00:04:18,800 --> 00:04:21,680 and actually feeling like you're prepared to do it. 117 00:04:21,680 --> 00:04:22,880 Like, OK, I know this is important. 118 00:04:22,880 --> 00:04:23,720 Yeah, exactly. 119 00:04:23,720 --> 00:04:27,560 And the Delay Report found that only 23% of companies 120 00:04:27,560 --> 00:04:29,960 felt like they were highly prepared to kind of deal 121 00:04:29,960 --> 00:04:31,680 with the governance challenges. 122 00:04:31,680 --> 00:04:33,720 So there's that disconnect. 123 00:04:33,720 --> 00:04:37,080 We know we should be doing it, but how do we do it? 124 00:04:37,080 --> 00:04:37,560 Exactly. 125 00:04:37,560 --> 00:04:40,120 And this is where I think the MIT Report is really valuable 126 00:04:40,120 --> 00:04:42,000 because it highlights how important it 127 00:04:42,000 --> 00:04:47,920 is to have clear leadership and this culture of responsible AI 128 00:04:47,920 --> 00:04:49,560 within an organization. 129 00:04:49,560 --> 00:04:51,040 So it's not just about the policies. 130 00:04:51,040 --> 00:04:53,840 It's about having the right people and the right mindset. 131 00:04:53,840 --> 00:04:54,720 Yeah, that's right. 132 00:04:54,720 --> 00:04:58,360 And that requires a real shift in things, 133 00:04:58,360 --> 00:05:02,400 not just within the tech teams, but at every level 134 00:05:02,400 --> 00:05:03,400 of the organization. 135 00:05:03,400 --> 00:05:05,600 It sounds like a lot of companies still have some work to do. 136 00:05:05,600 --> 00:05:08,480 Yeah, they do, but it's also a huge opportunity, right? 137 00:05:08,480 --> 00:05:10,640 The companies that can figure out 138 00:05:10,640 --> 00:05:15,080 how to balance that power of AI with the need 139 00:05:15,080 --> 00:05:18,960 for responsible ethical development, 140 00:05:18,960 --> 00:05:21,520 those are the ones that are going to thrive in this new era. 141 00:05:21,520 --> 00:05:22,520 That's a great point. 142 00:05:22,520 --> 00:05:25,320 It seems like the future of AI success 143 00:05:25,320 --> 00:05:28,120 isn't just about having the coolest algorithms. 144 00:05:28,120 --> 00:05:31,640 It's about having the right approach to governance 145 00:05:31,640 --> 00:05:33,240 and ethics and data management. 146 00:05:33,240 --> 00:05:34,840 Yeah, it's all of those things together. 147 00:05:34,840 --> 00:05:35,600 It's a whole package. 148 00:05:35,600 --> 00:05:36,280 Exactly. 149 00:05:36,280 --> 00:05:37,400 It is a whole package. 150 00:05:37,400 --> 00:05:39,560 And it's those factors, I think, that are really 151 00:05:39,560 --> 00:05:42,640 going to determine which companies can truly 152 00:05:42,640 --> 00:05:45,480 harness this transformative power of AI. 153 00:05:45,480 --> 00:05:48,120 OK, this is a great place to kind of pause and reflect 154 00:05:48,120 --> 00:05:49,080 for a second. 155 00:05:49,080 --> 00:05:51,200 We've uncovered some really interesting insights, I think, 156 00:05:51,200 --> 00:05:53,440 about the state of AI adoption. 157 00:05:53,440 --> 00:05:54,880 Yeah, we've got a lot to think about. 158 00:05:54,880 --> 00:05:56,000 In the enterprise. 159 00:05:56,000 --> 00:05:57,800 So let's take a quick break and we'll come back and explore 160 00:05:57,800 --> 00:05:59,240 some of these ideas even further. 161 00:05:59,240 --> 00:05:59,720 Welcome back. 162 00:05:59,720 --> 00:06:02,120 We were talking before the break about that tension 163 00:06:02,120 --> 00:06:06,800 between the excitement for AI and then the very real challenges 164 00:06:06,800 --> 00:06:10,080 companies are facing in terms of implementing it responsibly 165 00:06:10,080 --> 00:06:10,760 at scale. 166 00:06:10,760 --> 00:06:12,240 Yeah, and one of the biggest things 167 00:06:12,240 --> 00:06:14,960 is just the sheer cost and complexity 168 00:06:14,960 --> 00:06:18,760 of actually developing and deploying 169 00:06:18,760 --> 00:06:20,480 these large language models that are really 170 00:06:20,480 --> 00:06:23,280 the engine behind a lot of these generative AI tools. 171 00:06:23,280 --> 00:06:25,240 And the MIT report highlighted this, right? 172 00:06:25,240 --> 00:06:28,200 You talked about just the enormous resources 173 00:06:28,200 --> 00:06:30,560 required to train these massive models. 174 00:06:30,560 --> 00:06:31,240 It's insane. 175 00:06:31,240 --> 00:06:33,000 It's not something that everyone can just jump into, right? 176 00:06:33,000 --> 00:06:33,920 No, not at all. 177 00:06:33,920 --> 00:06:36,560 I mean, we're talking about millions of dollars 178 00:06:36,560 --> 00:06:39,120 in computing power and energy consumption 179 00:06:39,120 --> 00:06:41,200 just to get these models up and running. 180 00:06:41,200 --> 00:06:43,160 And even then, there's no guarantee 181 00:06:43,160 --> 00:06:46,120 that they're going to perform exactly as expected. 182 00:06:46,120 --> 00:06:49,160 So for companies that don't have those kinds of resources, 183 00:06:49,160 --> 00:06:50,200 what are their options? 184 00:06:50,200 --> 00:06:52,360 Are there other ways to get into this game? 185 00:06:52,360 --> 00:06:54,360 Yeah, I mean, the MIT report seems 186 00:06:54,360 --> 00:06:58,160 to suggest that maybe smaller, more specialized AI models 187 00:06:58,160 --> 00:07:00,960 could be a viable option. 188 00:07:00,960 --> 00:07:01,680 Interesting. 189 00:07:01,680 --> 00:07:04,160 Yeah, so the thinking there is that instead 190 00:07:04,160 --> 00:07:07,360 of trying to create this one gigantic model that 191 00:07:07,360 --> 00:07:09,640 can do everything, you focus on building 192 00:07:09,640 --> 00:07:13,880 smaller, more targeted models that are specifically designed 193 00:07:13,880 --> 00:07:16,160 for particular tasks or industries. 194 00:07:16,160 --> 00:07:18,040 So it's like having a team of specialists 195 00:07:18,040 --> 00:07:20,520 instead of one generalist who's trying to do everything. 196 00:07:20,520 --> 00:07:21,400 Exactly, yeah. 197 00:07:21,400 --> 00:07:22,240 It's a great analogy. 198 00:07:22,240 --> 00:07:25,640 And actually, these smaller, more focused models, 199 00:07:25,640 --> 00:07:27,320 they offer a number of advantages. 200 00:07:27,320 --> 00:07:30,240 For one thing, they're a lot cheaper and faster to train, 201 00:07:30,240 --> 00:07:33,920 which makes them much more accessible to a wider 202 00:07:33,920 --> 00:07:35,120 range of companies. 203 00:07:35,120 --> 00:07:35,920 That makes sense. 204 00:07:35,920 --> 00:07:39,920 But are these smaller models really powerful enough? 205 00:07:39,920 --> 00:07:42,680 I mean, can they actually compete with the big guys 206 00:07:42,680 --> 00:07:44,040 like ChatGPT and all that? 207 00:07:44,040 --> 00:07:45,920 Well, you might be surprised the report actually 208 00:07:45,920 --> 00:07:47,440 highlights a really interesting example. 209 00:07:47,440 --> 00:07:50,440 This model called Dali, which was developed by Databricks. 210 00:07:50,440 --> 00:07:53,400 It's an open source model that was trained for less than $30. 211 00:07:53,400 --> 00:07:54,520 Wait, less than $30? 212 00:07:54,520 --> 00:07:55,720 Yeah, less than $30. 213 00:07:55,720 --> 00:07:59,400 And it's showing some pretty impressive capabilities 214 00:07:59,400 --> 00:08:03,320 that actually rival some of the bigger, more expensive models 215 00:08:03,320 --> 00:08:03,840 out there. 216 00:08:03,840 --> 00:08:05,520 That's amazing, especially when you compare it 217 00:08:05,520 --> 00:08:08,440 to the millions we were talking about before. 218 00:08:08,440 --> 00:08:09,280 Yeah, it's crazy. 219 00:08:09,280 --> 00:08:11,760 And that's why these open source models like Dali 220 00:08:11,760 --> 00:08:13,840 are generating so much excitement. 221 00:08:13,840 --> 00:08:17,600 They have this potential to really democratize access 222 00:08:17,600 --> 00:08:20,960 to AI so that businesses of all sizes 223 00:08:20,960 --> 00:08:24,000 can start experimenting with it and benefiting from it. 224 00:08:24,000 --> 00:08:26,280 But going the open source route, you're 225 00:08:26,280 --> 00:08:27,800 kind of giving up some control. 226 00:08:27,800 --> 00:08:30,440 Like companies are very protective of their data, 227 00:08:30,440 --> 00:08:31,480 their intellectual property. 228 00:08:31,480 --> 00:08:32,920 Yeah, that's a valid concern. 229 00:08:32,920 --> 00:08:36,080 And it's something that both reports kind of acknowledge. 230 00:08:36,080 --> 00:08:39,200 There's definitely a trade-off to consider 231 00:08:39,200 --> 00:08:41,600 between using these pre-trained models 232 00:08:41,600 --> 00:08:44,400 that you get from third party providers 233 00:08:44,400 --> 00:08:45,920 versus building your own right. 234 00:08:45,920 --> 00:08:47,640 Because with a third party model, 235 00:08:47,640 --> 00:08:49,640 you might have concerns about data security 236 00:08:49,640 --> 00:08:53,160 limited customization options or even potential biases that 237 00:08:53,160 --> 00:08:54,720 are baked into the model itself. 238 00:08:54,720 --> 00:08:57,080 So building your own gives you more control, 239 00:08:57,080 --> 00:08:59,840 but it's more expensive and it's more complex. 240 00:08:59,840 --> 00:09:00,440 Exactly. 241 00:09:00,440 --> 00:09:02,120 It's a tough decision for companies to make. 242 00:09:02,120 --> 00:09:05,040 And as the MIT report emphasizes, 243 00:09:05,040 --> 00:09:07,560 even if you do decide to build your own model, 244 00:09:07,560 --> 00:09:10,800 you still need the right infrastructure to support it. 245 00:09:10,800 --> 00:09:14,360 You need powerful hardware specialized software 246 00:09:14,360 --> 00:09:19,000 and a team of skilled engineers to kind of manage it all. 247 00:09:19,000 --> 00:09:20,200 So it's not just about the AI. 248 00:09:20,200 --> 00:09:23,760 It's about building this whole ecosystem to support it. 249 00:09:23,760 --> 00:09:24,440 It really is. 250 00:09:24,440 --> 00:09:27,040 And both reports talk about these things, 251 00:09:27,040 --> 00:09:29,400 kind of data lake houses as being a key part 252 00:09:29,400 --> 00:09:30,360 of that infrastructure. 253 00:09:30,360 --> 00:09:31,440 OK, data lake houses. 254 00:09:31,440 --> 00:09:31,960 What are those? 255 00:09:31,960 --> 00:09:34,480 So think of it as like the centralized hub 256 00:09:34,480 --> 00:09:36,320 for all of a company's data. 257 00:09:36,320 --> 00:09:39,040 It's like a supercharged data warehouse 258 00:09:39,040 --> 00:09:40,760 that can handle both structured data, 259 00:09:40,760 --> 00:09:43,280 like sales figures, customer demographics, 260 00:09:43,280 --> 00:09:45,040 and unstructured data, like, you know, 261 00:09:45,040 --> 00:09:46,920 tech documents, images, videos. 262 00:09:46,920 --> 00:09:50,360 So it's a more modern way to manage all of the different types 263 00:09:50,360 --> 00:09:51,640 of data that companies are collecting. 264 00:09:51,640 --> 00:09:52,560 Yeah, exactly. 265 00:09:52,560 --> 00:09:55,040 And that flexibility is really crucial when it comes to AI 266 00:09:55,040 --> 00:09:58,600 because these models, they need to be able to access and analyze 267 00:09:58,600 --> 00:10:01,640 a wide variety of data to learn and improve. 268 00:10:01,640 --> 00:10:05,400 So it really seems like data is the foundation for all of this. 269 00:10:05,400 --> 00:10:05,920 Yeah. 270 00:10:05,920 --> 00:10:08,440 I mean, without the right data, even the best AI model 271 00:10:08,440 --> 00:10:10,040 is just going to flounder. 272 00:10:10,040 --> 00:10:10,800 Oh, absolutely. 273 00:10:10,800 --> 00:10:13,400 I mean, you can have the most brilliant AI algorithm 274 00:10:13,400 --> 00:10:14,000 in the world. 275 00:10:14,000 --> 00:10:16,400 But if you don't have the right data to feed it, 276 00:10:16,400 --> 00:10:18,120 you're not going to get very far. 277 00:10:18,120 --> 00:10:19,560 And it's not just about having the data. 278 00:10:19,560 --> 00:10:21,360 It's about being able to clean it, organize it, 279 00:10:21,360 --> 00:10:22,480 and make sure it's accurate. 280 00:10:22,480 --> 00:10:23,840 Yeah, if you put garbage in, you're 281 00:10:23,840 --> 00:10:24,880 going to get a garbage out. 282 00:10:24,880 --> 00:10:25,480 Exactly. 283 00:10:25,480 --> 00:10:26,520 Garbage in, garbage out. 284 00:10:26,520 --> 00:10:29,160 And that kind of brings us to another important aspect 285 00:10:29,160 --> 00:10:34,640 that both reports touch on, which is the human element 286 00:10:34,640 --> 00:10:35,400 of all of this. 287 00:10:35,400 --> 00:10:36,200 That's a big one. 288 00:10:36,200 --> 00:10:39,280 Yeah, because there's a lot of anxiety out there 289 00:10:39,280 --> 00:10:42,200 about AI and automation. 290 00:10:42,200 --> 00:10:44,800 People are worried about jobs being displaced. 291 00:10:44,800 --> 00:10:46,040 Yeah, for sure. 292 00:10:46,040 --> 00:10:47,760 What are the reports saying about that? 293 00:10:47,760 --> 00:10:51,160 Well, I think both reports, they acknowledge those concerns, 294 00:10:51,160 --> 00:10:54,200 but they offer, I think, a more nuanced perspective. 295 00:10:54,200 --> 00:10:56,640 They're suggesting that instead of viewing AI 296 00:10:56,640 --> 00:10:59,880 as this job killer, we should think of it 297 00:10:59,880 --> 00:11:04,120 as a tool that can actually augment human capabilities. 298 00:11:04,120 --> 00:11:07,720 So it's about freeing us from these tedious tasks 299 00:11:07,720 --> 00:11:11,360 and allowing us to focus on higher value work that 300 00:11:11,360 --> 00:11:14,040 requires creativity and critical thinking 301 00:11:14,040 --> 00:11:15,920 and emotional intelligence. 302 00:11:15,920 --> 00:11:19,360 So not AI replacing humans, but AI working alongside humans. 303 00:11:19,360 --> 00:11:20,080 Yeah, exactly. 304 00:11:20,080 --> 00:11:23,080 Making us more efficient and effective. 305 00:11:23,080 --> 00:11:26,200 And the experts that were interviewed for the MIT 306 00:11:26,200 --> 00:11:28,400 report, they really emphasize this point. 307 00:11:28,400 --> 00:11:31,520 They argue that AI is not about replacing human judgment 308 00:11:31,520 --> 00:11:32,280 or expertise. 309 00:11:32,280 --> 00:11:35,720 It's about providing us with the tools and insights we need 310 00:11:35,720 --> 00:11:38,800 to make better decisions and solve complex problems. 311 00:11:38,800 --> 00:11:40,800 OK, so that's a more optimistic view. 312 00:11:40,800 --> 00:11:41,560 Yeah, for sure. 313 00:11:41,560 --> 00:11:43,400 And can you give me an example of what 314 00:11:43,400 --> 00:11:45,120 that might look like in practice? 315 00:11:45,120 --> 00:11:48,280 Sure, so let's say you're in the health care industry. 316 00:11:48,280 --> 00:11:52,120 AI could be incredibly helpful in analyzing medical images, 317 00:11:52,120 --> 00:11:54,920 identifying patterns that maybe a human eye might miss, 318 00:11:54,920 --> 00:11:57,440 but it's never going to replace the expertise of a trained 319 00:11:57,440 --> 00:12:00,360 radiologist who can interpret those images in the context 320 00:12:00,360 --> 00:12:02,560 of a patient's history and symptoms. 321 00:12:02,560 --> 00:12:05,040 So the AI is like an assistant in a way. 322 00:12:05,040 --> 00:12:05,600 Exactly. 323 00:12:05,600 --> 00:12:08,200 It's providing information that can help the doctor make 324 00:12:08,200 --> 00:12:09,320 a more informed decision. 325 00:12:09,320 --> 00:12:10,240 Yeah, that's right. 326 00:12:10,240 --> 00:12:16,280 And this idea of AI as a co-pilot rather than a replacement, 327 00:12:16,280 --> 00:12:18,720 that's a theme that runs through both reports. 328 00:12:18,720 --> 00:12:20,960 I mean, for instance, in the creative industry, 329 00:12:20,960 --> 00:12:23,520 AI can help designers generate new ideas 330 00:12:23,520 --> 00:12:25,720 and automate repetitive tasks, but it 331 00:12:25,720 --> 00:12:29,760 can't replicate the human spark of creativity that's 332 00:12:29,760 --> 00:12:33,280 so essential for truly innovative work. 333 00:12:33,280 --> 00:12:35,400 So it's about finding that balance between the human 334 00:12:35,400 --> 00:12:36,240 and the machine. 335 00:12:36,240 --> 00:12:37,200 Yeah, exactly. 336 00:12:37,200 --> 00:12:40,480 Leveraging the strengths of each to get the best possible 337 00:12:40,480 --> 00:12:43,240 outcome, and that requires a shift in mindset, 338 00:12:43,240 --> 00:12:46,720 both within organizations and in society as a whole. 339 00:12:46,720 --> 00:12:49,120 We need to start thinking about AI not as a threat, 340 00:12:49,120 --> 00:12:51,760 but as a tool that can help us solve some of the world's 341 00:12:51,760 --> 00:12:53,080 most pressing challenges. 342 00:12:53,080 --> 00:12:54,920 It's about embracing the potential, 343 00:12:54,920 --> 00:12:57,360 but also recognizing the limitations. 344 00:12:57,360 --> 00:13:00,600 Exactly, and making sure that humans stay 345 00:13:00,600 --> 00:13:02,040 at the center of the equation. 346 00:13:02,040 --> 00:13:02,880 Right at the heart of it. 347 00:13:02,880 --> 00:13:06,600 And that brings us back to this crucial issue of governance 348 00:13:06,600 --> 00:13:08,160 and risk management. 349 00:13:08,160 --> 00:13:09,440 Yeah, we can't forget about that. 350 00:13:09,440 --> 00:13:11,960 Right, because we talked earlier about the need 351 00:13:11,960 --> 00:13:15,280 for these clear governance frameworks 352 00:13:15,280 --> 00:13:18,520 to make sure AI is being developed and used responsibly. 353 00:13:18,520 --> 00:13:20,320 But what does that actually look like? 354 00:13:20,320 --> 00:13:22,880 Yeah, I mean, both reports emphasize 355 00:13:22,880 --> 00:13:24,160 several key elements. 356 00:13:24,160 --> 00:13:27,120 First, there needs to be clear accountability 357 00:13:27,120 --> 00:13:28,360 within organizations. 358 00:13:28,360 --> 00:13:30,600 Somebody needs to be responsible for making sure 359 00:13:30,600 --> 00:13:33,440 that AI systems are developed and used in a way that 360 00:13:33,440 --> 00:13:37,040 aligns with the company's values and ethical principles. 361 00:13:37,040 --> 00:13:38,440 So someone to steer the ship. 362 00:13:38,440 --> 00:13:39,600 Yeah, someone at the helm. 363 00:13:39,600 --> 00:13:42,400 And that person needs to have the authority 364 00:13:42,400 --> 00:13:46,520 to make decisions and enforce policies related to AI. 365 00:13:46,520 --> 00:13:49,440 And the second key element is transparency. 366 00:13:49,440 --> 00:13:51,680 Companies need to be open about how they're using AI, 367 00:13:51,680 --> 00:13:53,640 what data they're collecting, and how they're protecting 368 00:13:53,640 --> 00:13:54,320 that data. 369 00:13:54,320 --> 00:13:55,840 Transparency builds trust. 370 00:13:55,840 --> 00:13:56,600 It does. 371 00:13:56,600 --> 00:13:58,880 It helps to alleviate some of those concerns 372 00:13:58,880 --> 00:14:02,000 that people have about AI being used in ways 373 00:14:02,000 --> 00:14:04,680 that are kind of opaque or potentially harmful. 374 00:14:04,680 --> 00:14:06,720 Right, so it's about being upfront about it. 375 00:14:06,720 --> 00:14:11,200 Yeah, and finally, there needs to be a robust process 376 00:14:11,200 --> 00:14:13,920 for actually identifying and mitigating 377 00:14:13,920 --> 00:14:15,800 the risks associated with AI. 378 00:14:15,800 --> 00:14:18,800 And that includes things like conducting 379 00:14:18,800 --> 00:14:22,800 thorough ethical reviews of AI systems before they're deployed, 380 00:14:22,800 --> 00:14:24,560 monitoring their performance over time, 381 00:14:24,560 --> 00:14:26,280 and having clear protocols in place 382 00:14:26,280 --> 00:14:29,560 for dealing with any unintended consequences or biases that 383 00:14:29,560 --> 00:14:30,720 might emerge. 384 00:14:30,720 --> 00:14:32,960 So it's about being proactive rather than reactive. 385 00:14:32,960 --> 00:14:35,280 Exactly, anticipate potential problems 386 00:14:35,280 --> 00:14:39,840 and have planned to address them before they become major issues. 387 00:14:39,840 --> 00:14:41,920 And this sounds like an area where a lot of companies 388 00:14:41,920 --> 00:14:43,680 are still trying to catch up. 389 00:14:43,680 --> 00:14:44,360 Yeah, you're right. 390 00:14:44,360 --> 00:14:47,160 The Deloitte report found that while most companies recognize 391 00:14:47,160 --> 00:14:50,920 the need for this robust AI governance, 392 00:14:50,920 --> 00:14:54,080 very few feel adequately prepared to actually implement it. 393 00:14:54,080 --> 00:14:56,240 They know it's important, but they don't know how. 394 00:14:56,240 --> 00:14:58,480 Figuring out how to do it is the next step. 395 00:14:58,480 --> 00:15:00,320 Exactly, there's a lot of work to be done 396 00:15:00,320 --> 00:15:04,080 in terms of developing the right frameworks and the tools 397 00:15:04,080 --> 00:15:07,960 and the expertise to really manage AI responsibly. 398 00:15:07,960 --> 00:15:12,120 And we can't forget about the evolving regulatory landscape, 399 00:15:12,120 --> 00:15:12,840 too. 400 00:15:12,840 --> 00:15:16,120 The MIT report mentions the EU's AI Act, which 401 00:15:16,120 --> 00:15:17,600 came into effect earlier this year. 402 00:15:17,600 --> 00:15:20,280 And there are likely to be more regulations 403 00:15:20,280 --> 00:15:22,480 as governments around the world grapple 404 00:15:22,480 --> 00:15:23,960 with all the implications of AI. 405 00:15:23,960 --> 00:15:26,120 Right, so companies need to stay informed about all of that. 406 00:15:26,120 --> 00:15:26,600 They do. 407 00:15:26,600 --> 00:15:27,640 They need to stay informed. 408 00:15:27,640 --> 00:15:29,680 And they need to be prepared to adapt. 409 00:15:29,680 --> 00:15:31,040 You know, to all the changes. 410 00:15:31,040 --> 00:15:31,600 Exactly. 411 00:15:31,600 --> 00:15:35,440 It sounds like navigating this AI world is a real balancing act. 412 00:15:35,440 --> 00:15:35,840 It is. 413 00:15:35,840 --> 00:15:37,400 You've got all this amazing potential, 414 00:15:37,400 --> 00:15:39,800 but then all of these risks that you have to consider. 415 00:15:39,800 --> 00:15:41,280 Yeah, you hit the nail on the head. 416 00:15:41,280 --> 00:15:43,040 It's a balancing act that requires 417 00:15:43,040 --> 00:15:45,440 a thoughtful and proactive approach. 418 00:15:45,440 --> 00:15:48,240 Companies can't just sort of rush into this head first 419 00:15:48,240 --> 00:15:50,880 without thinking about the potential consequences. 420 00:15:50,880 --> 00:15:53,360 And that's a great place to pause. 421 00:15:53,360 --> 00:15:55,960 Now we've covered a lot of ground in this deep dive. 422 00:15:55,960 --> 00:15:58,920 We're back and ready to kind of wrap things up. 423 00:15:58,920 --> 00:16:01,280 We've been talking about generative AI in the enterprise. 424 00:16:01,280 --> 00:16:02,760 And before the break, we were really 425 00:16:02,760 --> 00:16:06,440 focused on that critical need for companies 426 00:16:06,440 --> 00:16:09,480 to establish those strong governance frameworks 427 00:16:09,480 --> 00:16:11,200 as they're adopting AI. 428 00:16:11,200 --> 00:16:13,240 Yeah, it's one thing to say, you know, 429 00:16:13,240 --> 00:16:14,720 responsible AI is important. 430 00:16:14,720 --> 00:16:16,640 But how do you actually put it into practice? 431 00:16:16,640 --> 00:16:17,200 Exactly. 432 00:16:17,200 --> 00:16:18,760 How do you actually make it happen? 433 00:16:18,760 --> 00:16:20,880 Yeah, and the Deloitte Report gives us 434 00:16:20,880 --> 00:16:22,560 some pretty practical guidance here. 435 00:16:22,560 --> 00:16:23,480 OK. 436 00:16:23,480 --> 00:16:25,480 So they really stress the importance 437 00:16:25,480 --> 00:16:29,000 of getting buy-in and oversight from the very top. 438 00:16:29,000 --> 00:16:31,960 You know, the board, the C-suite, 439 00:16:31,960 --> 00:16:36,720 this can't just be like an IT initiative that's off in a silo. 440 00:16:36,720 --> 00:16:38,600 It needs to be a strategic priority 441 00:16:38,600 --> 00:16:39,960 that everyone's on board with. 442 00:16:39,960 --> 00:16:43,960 So AI needs to be aligned with the company's overall goals 443 00:16:43,960 --> 00:16:44,520 and values. 444 00:16:44,520 --> 00:16:45,600 Yeah, exactly. 445 00:16:45,600 --> 00:16:48,800 And there needs to be that sense of ownership 446 00:16:48,800 --> 00:16:51,000 and accountability at every level. 447 00:16:51,000 --> 00:16:53,000 The report also suggests establishing 448 00:16:53,000 --> 00:16:56,400 these cross-functional teams to kind of lead the charge when 449 00:16:56,400 --> 00:17:00,320 it comes to identifying and mitigating the risks associated 450 00:17:00,320 --> 00:17:01,000 with AI. 451 00:17:01,000 --> 00:17:02,680 And this is really make sure you're 452 00:17:02,680 --> 00:17:04,800 getting diverse perspectives and that you're 453 00:17:04,800 --> 00:17:08,080 thinking about the ethical and societal implications. 454 00:17:08,080 --> 00:17:11,080 Right, because AI touches so many parts of a business, 455 00:17:11,080 --> 00:17:15,120 you need a team with a lot of different expertise 456 00:17:15,120 --> 00:17:16,640 to really assess that impact. 457 00:17:16,640 --> 00:17:17,640 Yeah, absolutely. 458 00:17:17,640 --> 00:17:21,880 And this team should be empowered to proactively identify 459 00:17:21,880 --> 00:17:25,080 and address potential risks, not just react to problems 460 00:17:25,080 --> 00:17:26,160 after they've already happened. 461 00:17:26,160 --> 00:17:28,280 Right, so proactive, not reactive. 462 00:17:28,280 --> 00:17:28,880 Exactly. 463 00:17:28,880 --> 00:17:31,680 And another key recommendation from the Deloitte Report 464 00:17:31,680 --> 00:17:33,880 is to actually appoint a single executive 465 00:17:33,880 --> 00:17:36,640 to oversee all AI-related risks. 466 00:17:36,640 --> 00:17:39,520 So you have that one person who's accountable. 467 00:17:39,520 --> 00:17:40,800 Yeah, AI risk office. 468 00:17:40,800 --> 00:17:41,280 Yeah, exactly. 469 00:17:41,280 --> 00:17:44,720 Someone who's specifically responsible for keeping 470 00:17:44,720 --> 00:17:46,520 an eye on the regulations and making sure 471 00:17:46,520 --> 00:17:47,800 the company is compliant. 472 00:17:47,800 --> 00:17:50,480 Having that dedicated person could be a really good way 473 00:17:50,480 --> 00:17:53,720 to streamline things and make sure that those governance 474 00:17:53,720 --> 00:17:55,400 policies are actually happening. 475 00:17:55,400 --> 00:17:56,400 Yeah, I think so. 476 00:17:56,400 --> 00:17:59,880 And the MIT report adds another layer to this conversation. 477 00:17:59,880 --> 00:18:02,040 They emphasize that governance isn't just 478 00:18:02,040 --> 00:18:06,640 about having the right rules and regulations on paper. 479 00:18:06,640 --> 00:18:10,840 It's also about building this culture of responsible AI 480 00:18:10,840 --> 00:18:12,160 within the organization. 481 00:18:12,160 --> 00:18:14,200 Right, it's about education and awareness, 482 00:18:14,200 --> 00:18:16,920 making sure employees understand the ethical issues 483 00:18:16,920 --> 00:18:19,360 and feel empowered to speak up. 484 00:18:19,360 --> 00:18:23,000 Exactly, and this culture of responsible AI, 485 00:18:23,000 --> 00:18:25,360 it should extend beyond the company itself. 486 00:18:25,360 --> 00:18:28,560 It should include suppliers, partners, even customers. 487 00:18:28,560 --> 00:18:29,760 It's all ecosystem thing. 488 00:18:29,760 --> 00:18:30,440 It is. 489 00:18:30,440 --> 00:18:31,080 It really is. 490 00:18:31,080 --> 00:18:34,840 And ultimately, I think it comes down to trust in the technology, 491 00:18:34,840 --> 00:18:37,040 trust in the companies that are developing and using it, 492 00:18:37,040 --> 00:18:39,640 and trust in the people who are interacting with it. 493 00:18:39,640 --> 00:18:41,520 And the trust is so important as AI 494 00:18:41,520 --> 00:18:44,720 becomes more and more a part of our lives. 495 00:18:44,720 --> 00:18:47,200 Absolutely, and that kind of leads to another key takeaway 496 00:18:47,200 --> 00:18:52,400 from both reports, which is this need for ongoing dialogue 497 00:18:52,400 --> 00:18:53,600 and collaboration. 498 00:18:53,600 --> 00:18:56,000 The field is moving so quickly. 499 00:18:56,000 --> 00:18:59,200 The ethical and societal implications 500 00:18:59,200 --> 00:19:00,720 are constantly evolving. 501 00:19:00,720 --> 00:19:03,480 So it's not about finding a one-size-fits-all solution. 502 00:19:03,480 --> 00:19:07,920 It's about having that culture of continuous learning 503 00:19:07,920 --> 00:19:09,000 and adaptation. 504 00:19:09,000 --> 00:19:10,200 Right, always be learning. 505 00:19:10,200 --> 00:19:12,200 Exactly, and this conversation, it 506 00:19:12,200 --> 00:19:15,520 needs to include everyone, academia, government, 507 00:19:15,520 --> 00:19:17,400 industry, civil society. 508 00:19:17,400 --> 00:19:18,960 We all need to be working together 509 00:19:18,960 --> 00:19:21,720 to make sure that AI is used for good. 510 00:19:21,720 --> 00:19:23,360 Right, that it benefits everyone. 511 00:19:23,360 --> 00:19:24,120 Exactly. 512 00:19:24,120 --> 00:19:26,000 Well said, and I think that's a great place 513 00:19:26,000 --> 00:19:27,440 to wrap up our discussion today. 514 00:19:27,440 --> 00:19:29,200 I mean, we've covered a lot of ground. 515 00:19:29,200 --> 00:19:30,000 We have. 516 00:19:30,000 --> 00:19:35,160 From the incredible potential of generative AI to the importance 517 00:19:35,160 --> 00:19:37,880 of responsible development and governance. 518 00:19:37,880 --> 00:19:38,840 Yeah, we've gone deep. 519 00:19:38,840 --> 00:19:39,240 We have. 520 00:19:39,240 --> 00:19:39,880 We've gone deep. 521 00:19:39,880 --> 00:19:41,080 I think it's clear that we're still 522 00:19:41,080 --> 00:19:43,760 in the early stages of this AI revolution. 523 00:19:43,760 --> 00:19:44,920 Oh, yeah, for sure. 524 00:19:44,920 --> 00:19:46,640 There's a lot that we're still figuring out. 525 00:19:46,640 --> 00:19:48,080 But it's exciting to see that we are 526 00:19:48,080 --> 00:19:50,640 starting to have these important conversations about how 527 00:19:50,640 --> 00:19:53,800 to make sure that AI is used for good. 528 00:19:53,800 --> 00:19:55,680 Absolutely, and I think the companies that 529 00:19:55,680 --> 00:20:00,600 are really going to succeed in this new era are the ones 530 00:20:00,600 --> 00:20:04,960 that embrace the power of AI, while also being realistic 531 00:20:04,960 --> 00:20:07,280 about its limitations and prioritizing 532 00:20:07,280 --> 00:20:10,480 ethical development and responsible use. 533 00:20:10,480 --> 00:20:13,760 And those that can build that culture of trust 534 00:20:13,760 --> 00:20:16,440 and transparency and collaboration, 535 00:20:16,440 --> 00:20:20,600 both within their own company, and then the wider AI ecosystem. 536 00:20:20,600 --> 00:20:21,840 Yeah, that's key. 537 00:20:21,840 --> 00:20:23,680 Well, thank you for joining us for this deep dive 538 00:20:23,680 --> 00:20:24,800 into generative AI. 539 00:20:24,800 --> 00:20:28,520 I hope you found it insightful and thought-provoking. 540 00:20:28,520 --> 00:20:56,080 And until next time.