AI in Smart Cities: Real-World Uses, Risks and Investment Shifts

00:03.880 --> 00:05.360 Unless you've been living 00:05.360 --> 00:08.960 under a rock, you've probably heard the commotion surrounding AI. 00:09.120 --> 00:14.160 From predictions of catastrophic risk to increasingly autonomous AI agents. 00:14.280 --> 00:17.640 The conversation has taken a dramatic turn. 00:17.680 --> 00:21.840 But beyond the headlines, AI is already transforming entire 00:21.840 --> 00:26.120 industries, and smart cities and utilities are no exception. 00:26.160 --> 00:29.800 It's changing how solutions are developed, how cities operate, 00:29.840 --> 00:33.600 and even how investors choose which companies to back. 00:33.960 --> 00:37.640 Municipalities are not trying to automate everything at once, 00:37.680 --> 00:41.520 but rather trying to restructure a little bit their core workflows around 00:41.560 --> 00:46.840 targeted, well understood technologies and real capabilities which are proven. 00:47.040 --> 00:51.200 And that's the key distinction in the industry: AI hype vs AI 00:51.320 --> 00:54.320 that works for city and utility management. 00:54.400 --> 00:58.160 In smart cities, we can identify four major AI applications: 00:58.240 --> 01:00.200 computer vision, predictive AI, 01:00.200 --> 01:03.000 optimization AI, and generative AI. 01:03.000 --> 01:06.400 The first three are already delivering measurable value in areas 01:06.400 --> 01:10.720 such as traffic management, energy forecasting, and waste collection. 01:10.760 --> 01:14.000 We’re routing, let's say a few thousand pickup 01:14.000 --> 01:17.000 points daily for our clients. 01:17.040 --> 01:20.840 Then AI can help a lot in making the right routes 01:20.840 --> 01:23.840 and taking into account a lot of 01:24.640 --> 01:26.560 variables and parameters. 01:26.560 --> 01:29.040 But efficiency isn't the only consideration. 01:29.080 --> 01:32.760 AI promises to reduce costs and automate work. 01:32.880 --> 01:36.080 Yet what makes the technology attractive to private 01:36.080 --> 01:39.440 companies can make it difficult to sell to governments. 01:39.640 --> 01:42.640 Cities aren't necessarily looking to eliminate jobs. 01:42.640 --> 01:46.320 They need solutions that address real operational problems. 01:46.480 --> 01:49.480 Solutions have to be implemented at the right time. 01:49.720 --> 01:53.280 Ten years ago, we started with a solution for optimizing 01:53.520 --> 01:58.120 waste collection, which would save 20-30% of truck drivers. 01:58.600 --> 02:02.400 That only became accepted because the market for truck 02:02.400 --> 02:05.600 drivers was very difficult and they couldn't find any new personnel. 02:05.880 --> 02:08.080 And when it comes to generative AI, 02:08.080 --> 02:11.680 not a chatbot, but rather an autonomous AI system, 02:11.760 --> 02:16.000 it's not just about how the tech can help solve city issues. 02:16.120 --> 02:20.600 Hallucinations, cybersecurity, data sovereignty, and accountability 02:20.640 --> 02:24.040 remain major concerns when installing gen AI, 02:24.400 --> 02:28.120 especially when AI starts influencing critical infrastructure. 02:28.280 --> 02:32.240 The part of public sector management that demands absolute 02:32.240 --> 02:36.640 predictability, repeatability, and strict regulatory compliance, 02:37.640 --> 02:38.240 that's where 02:38.240 --> 02:42.080 probably generative AI and agents will have a hard time to come. 02:42.120 --> 02:47.240 So even managing the traffic, for example, that could have impact on human safety. 02:47.240 --> 02:50.760 And that is not acceptable for city operators. 02:50.800 --> 02:53.040 But cities aren't rejecting generative AI. 02:53.040 --> 02:56.920 They're experimenting with it, testing where it can deliver value 02:57.040 --> 03:00.360 without compromising reliability or public safety. 03:00.360 --> 03:02.360 Cities are allocating some budget 03:02.360 --> 03:03.600 to create pilots. 03:03.600 --> 03:07.520 Because although the technology is really powerful, 03:07.520 --> 03:11.400 we still need to apply it to different use cases. 03:11.400 --> 03:14.440 So the use cases are still to be refined 03:14.800 --> 03:17.920 because the technology is super new and we are starting to adopt it. 03:18.000 --> 03:22.320 But AI isn't just changing the solutions cities buy, it's changing the companies 03:22.320 --> 03:23.160 building them. 03:23.160 --> 03:24.080 At Waste Vision, 03:24.080 --> 03:28.160 for example, the development team has gone from eight people to two, 03:28.360 --> 03:31.880 something the company attributes to AI-enabled efficiencies. 03:32.040 --> 03:35.960 And as AI makes software development faster and cheaper, 03:36.080 --> 03:39.880 investors are rethinking what makes a startup worth backing. 03:39.920 --> 03:46.200 I would say that for us, it's not about the tech or the use case, 03:46.200 --> 03:50.080 although obviously we see it because we need to understand the opportunity. 03:50.280 --> 03:53.520 It's about the people, like the team who is behind the technology 03:54.040 --> 03:58.160 because as I told you, everything changes every week. 03:58.200 --> 04:02.320 So we need to invest in teams that we know that they can adapt. 04:02.440 --> 04:05.720 AI is already transforming the smart city industry 04:05.760 --> 04:09.920 from how solutions are built to how they're funded and deployed. 04:10.040 --> 04:13.320 But there's a growing gap between what AI promises 04:13.360 --> 04:16.880 and what municipalities can realistically implement. 04:17.040 --> 04:18.400 The winners in smart cities 04:18.400 --> 04:22.160 won't necessarily be those developing the most advanced AI. 04:22.400 --> 04:27.400 There'll be those proving it can deliver real value safely and reliably.