Cha-Ly KOH, Founder and CEO, Urbanmetry Shaun KOO, CEO and Co-founder, H3 Zoom.AI Joe XIA, CEO, Jidu Moderator: Nicholas GORDON, Editor, Asia, FORTUNE; Co-chair, Fortune Brainstorm AI Singapore
Category
🤖
TechTranscript
00:00So, Charlie, first question to you.
00:02You know, a lot of places talked about being smart cities.
00:05And the benefits from them are sometimes a little hard to recognize.
00:10There have been some pretty high-profile failures.
00:13As ai really develops and we get this new technology, is the idea
00:17Of the smart city now maybe coming into fruition?
00:21Thank you so much for having me today.
00:23And i think i'm really excited about the potential of ai
00:27Generally because it levels up the playing field.
00:30So for the longest time, smart city was pitched as this
00:34Innovative, very, very effective way of governing cities.
00:38But it was afforded only by developed nations or developed cities.
00:42So the four processes to achieve smart city, data collection,
00:46Data analytics, decision-making, and action.
00:50For the longest time, the folks who can afford data collection
00:54Was really developed cities like singapore, many parts of china, korea.
01:01Today, data collection can be done at a fraction of the cost
01:05Using ai for developing cities.
01:07So that makes it bringing it forward in terms of leveling up
01:11The playing field for these cities.
01:13But i think to also want to curb some of the enthusiasm.
01:17So if you think about data collection, that levels it up.
01:20Data analytics, great.
01:22You can use some of the ai to do that.
01:24But let's talk about a little bit in the data decision-making
01:27And action. Is that purely able to be done
01:30By ai? in this part of the region, i
01:33Think we are still quite far away from that because
01:37Fundamentally cities are political.
01:39And there are political parts of the decision-making.
01:42Unfortunately, we don't trust ai.
01:44I don't foresee that we trust ai to do that completely for us at
01:47This point.
01:49I do want to get back to this point about politics and trust.
01:52But before i do, i want to bring in shawn.
01:54As someone who is doing this data collection, could you tell
01:57Us a bit more about zoom ai and what you're trying to do?
02:01Right. So at zoom ai, we see ourselves
02:05As the palanter of the built environment.
02:08So we provide intelligent insights and actionable
02:12Outcomes for property developers, fortune 500
02:16And engineering consultants on the entire building condition
02:19Assessments. Essentially maximizing the
02:23Life cycle value of these assets.
02:26So that's essentially what we do. We work with a myriad of data
02:32Sources, drones, off-the-shelf cameras, and we ingest all of
02:36These data sources to provide these outcomes in a very
02:39Intuitive form. I want to bring in joe now.
02:44Autonomous driving is another one of these terms that's been
02:46Talked about for a long time. You're with gdo.
02:48Could you explain to us kind of what exactly is gdo and how it
02:52Relates to baidu, its ai initiatives with ernie, and the
02:56Apollo robo taxis. Where does gdo fit in this network?
03:00Yeah. Thanks for having me.
03:03I think gdo, first of all, gdo, we're kind of a start-up
03:07Company. We're only three years.
03:10So when we founded it, we're aiming to actually developing
03:14What we call smart ev. It's not ev, it's smart ev.
03:18So we call our product robocop. So we have a joint shareholder.
03:26The goal for them to come together is, you know, for the
03:29Past ten years, baidu has been, like, invested over 100 billion
03:36Dollars in their own ai technology.
03:38So it's just the right time right now for us to bring those
03:41Ai technology. It used to be run on the cloud.
03:44Now we can bring that to the cloud.
03:46So that's why it brings us here. And i think it's the right time.
03:50Because the computing power on the car side and both on the
03:54Computing side, on the car side, right now is big enough to
03:58Support those ai to running. For example, al4 used to be,
04:02Like, you know, very heavy on computing, on gpus.
04:06And we don't have a vehicle level chipset in order to run
04:10That on consumer vehicles. Not until 2023, we have orinx
04:16Come from nvidia. So that's why we're doing that.
04:21We develop cars, and we integrate baidu ai technology,
04:27Like, you know, ai program and apollo l4 technology.
04:33But not exactly l4. But how are we going to firstly,
04:38You know, kind of downgrade the l4 and having customers to be
04:41Able to use a high-level, you know, kind of autonomous
04:44Driving in the city. So that's the purpose and the
04:48Goal of baidu. Yeah.
04:50All right. I want to stay on this politics
04:52Point now. Maybe first question to you,
04:54Charlie. You know, i've read some of
04:56Your commentary on this, and you talk about things like
05:00Housing affordability, big political topic.
05:02I saw your linkedin post about diesel subsidies in malaysia,
05:05Another big political topic. How do you see data analytics
05:10And ai and data collection interacting with the political
05:14Discussions in these cities, particularly in asian cities?
05:18No, i think for the longest time, because data was so hard
05:22To collect, and things become a little bit more opaque, it was
05:28Very hard for data-driven decision to enter into the
05:31Conversation. I'm saying enter the
05:34Conversation, because things are moving very quickly.
05:37By the time you collect this information, it's too late.
05:42Right? but today, things are so realtime
05:45That the conversations and the disparity between some decision
05:49Making and the data that's available becomes a little bit
05:53Hard to reconcile. So i think the availability of
05:57Data gives a lot of transparency in terms of what
06:01Are people really being motivated by doing certain decisions.
06:06Right? so the diesel point that i made
06:10On linkedin was that we are consciously thinking about how
06:15We are building our city very far away, and that in itself is
06:19A conscious decision on how much we're spending to transport our
06:23Logistics. So while we're solving it with
06:27Ev, while we're solving it with efficiency of ai, shouldn't we
06:31Also use this to plan in the future and reduce that today?
06:35So those are a part of using the data collection using ai will
06:40Help us bring that point forward to be part of the
06:44Conversation, not really to dictate it, but at least
06:47Participate in it.
06:49People can interpret the data how they wish in ways that
06:53Support certain conclusions. I mean, what's the risk of
06:56People then misinterpreting that data and thus worsening the
06:59Political debate over a certain issue?
07:01I think that's always the case with any kind of data and
07:05However you collect it. And i think a lot of times people
07:11Are really afraid of using data or ai in analytics because of
07:15Biases, right? but i think it is even worse if
07:19You don't have the data available for people to not be
07:23Able to actually drill down to even question the data.
07:27Data transparency is very, very important, and the ability to
07:32Have general folks actually criticize that data i think is
07:35Step one. I want to go back to shawn now.
07:38Remember when drones were the next big thing in tech?
07:41They were cheap, ubiquitous, they were everywhere, and then
07:44We all got distracted by something else.
07:49What else can drones help us do in terms of urban planning?
07:52And then how do you manage people's perception of those
07:55Drones? they may worry about them being
07:58Invasive, personal privacy. How do you chart that balance
08:02In terms of taking advantage of what drones can do, but also
08:05Making sure urban populations aren't scared by fleets of
08:07Drones collecting data on them?
08:09So that's actually a very good question.
08:12So obviously i think there's a lot of fixation on drones as
08:17Being the next big thing, but obviously through sort of
08:23Conversations with partners, customers, as well as industry
08:28Stakeholders, at the end of the day, the data that's being
08:33Collected by drones, whether it's for construction,
08:38Monitoring, for asset inspection, for precision
08:42Agriculture, what the industry and the customers really do care
08:47About are the insights. What are actionable outcomes and
08:53What can a.I. Algorithms that have been
08:57Developed, how can they actually interpret some of this
09:01Unstructured data? so i think the perception
09:06Definitely has changed over time, and from my perspective,
09:13Being in the business for close to ten years now, the sort of
09:18Evolution of acquisition has also been extended to ground
09:22Robotics, wall robotics, because the end client, what
09:27They really value and care about are outcomes that would drive
09:31The decision-making processes. So i think in terms of
09:36Governance around the data collection and security aspect,
09:41I mean, there are existing guardrails that are in place.
09:46So obviously there are algorithms out there that could
09:50Ask faces, windows, number plates, et cetera, to sort of
09:56Appease some of those regulatory requirements.
10:01On the other hand, there are iso standards that can be built
10:09Within the system level, right, in terms of data encryption,
10:14Sovereignty, et cetera. So there are definitely best
10:18Practices out there. So i mean, in short, i guess
10:22To drive home the point is, you know, the end outcome of being
10:29Able to interpret the unstructured data to drive the
10:34Business decisions are the true sort of testament of what the
10:38Industry and the clients are actually looking out for.
10:41Thank you, joe. One more politics question,
10:43Then we'll move back to safer territory.
10:45So apollo's robo taxis, investors have been very
10:47Excited about those. But i know that's led to some
10:52Controversy in wuhan from people worried about how they will
10:55Affect jobs. And, you know, how do you, how
10:59Should cities manage that transition as more stuff gets
11:02Automated, how do they manage the human labor that gets
11:06Affected from that new technology?
11:08Yeah, i think just quickly answer the question.
11:12It's still far, it's still, you know, too early to think about
11:16That, right, because, you know, baidu has, like, started
11:20Developing autonomous driving since 2013, right, like 11 years
11:24Ago. So the government under actually
11:27The supervision of the government, right, so, you
11:30Know, started from a single vehicle, a few vehicles on a
11:34Single road, you know, all the way to now, you know, just to
11:40Be clear, not until last year, you know, 40 l4 robo taxi is not
11:47Allowed in china to operate. The government just opened that
11:51Up last year. And really with a small scale,
11:55Not a large scale, talk about wuhan, wuhan has over, like,
11:5940,000, You know, taxi, and the l4 robo taxi right now is only
12:03Like 400, 500. It's very small amount of robo
12:07Taxi. I think it's still under trial.
12:10The government, it's not like the next day there are going to
12:14Be, like, thousands of, hundreds of thousands of, you know, taxi
12:17On the road, right? it's still, you know, too early
12:20To worry about that, right? i mean, still, it's still a trial.
12:24And most importantly, i think every time there's an
12:27Evolution of a technology, it's going to open another window up, right?
12:31And you think about years ago when we talk about, you know,
12:35Those robots on car manufacturing lines, right?
12:38People thought, like, we're going to lose our jobs, right?
12:41But turns out there's more jobs that are created because those
12:44Robots, when those robots come in, and it's actually increased
12:47The efficiency of the manufacturing.
12:49And in that case, actually, it makes the sales of the car
12:53Volume even bigger, and it creates more job, right?
12:56So i don't think we have to really worry about that at this
12:59Stage, yeah.
13:01So automation will create more labor.
13:03Yeah, yeah. End it up, we'll create more jobs.
13:07You know, similarly, on this labor point, i mean, by
13:11Replacing, you know, data collection, like, people, you
13:14Know, guys on the street with clickers, monitoring the
13:17Traffic with sensors, yes, i mean, you will be replacing
13:22Jobs that would normally go to people.
13:24How do you see that transition, especially in most of asia
13:28Where, you know, labor is probably still cheaper than
13:30Capital. How do you kind of, again,
13:33Balance between labor and technology?
13:36So i think most of asian is still thinking of the current
13:41Situation, but ai is going to be built for the next, you know,
13:4550 Years. It's not going to be for today.
13:48So if you are thinking of your population as the next, of
13:52Course, everybody needs a job, but is it for the job today or
13:55The job for the next 30 years? do you want to be a guy
13:58Standing there doing the clicker or the guy who is actually
14:01Designing the ai system who is counting the cars?
14:05So governments are actually in the position right now to make
14:08A call around the transitioning towards the higher value jobs
14:12In which you can actually create the value in the next 20, 30
14:16Years and not having to worry about stagnation.
14:19So stagnation is something that we should worry about today.
14:24And taking the first step of actually using ai as an
14:28Augmented, you know, help for you to do that is going to be a
14:32Good first step, i think.
14:34Sean, i do want to bring in the sustainability point, you
14:37Know, making cities more climate friendly.
14:39Obviously we're in a region of the world that's going to be
14:42Affected by climate change. How will these technologies
14:46Make cities more sustainable?
14:49So obviously basically the environment is kind of like the
14:54Bedrock of urban city centers. Obviously, you know, if you're
14:59Using robotics and drone platforms, et cetera, there are
15:03A lot of direct benefits in terms of replacement of having
15:07To ferry workers on site, specifically in southeast asia.
15:11Because we still rely on a lot of low-skilled foreign workers
15:15That are, you know, essentially building the buildings and the
15:18City centers from ground up. So obviously in terms of
15:21Carbon emissions, there are indirect impacts on
15:24Transportation. That's one.
15:26In terms of the data aspect of things, by being able to plug
15:31Into various data sources within building management systems,
15:36Hvac, there is a potential to also reduce the carbon emissions
15:40From buildings. So 30 of carbon emissions
15:44Today, you know, they come from buildings.
15:48So if we can reduce sort of like the footprint in terms of how
15:53Buildings function today through some of these insights, then it
15:57Would, you know, empower architects and designers of
16:00Tomorrow to design more sustainable buildings and assets.
16:04So that's how we, you know, how i see sustainability.
16:09I want to squeeze in a couple more quick questions.
16:12Joe, you were also previously, you were the cofounder of
16:16Mobike, bike sharing app, bike sharing service.
16:19What's a lesson from that experience that you think
16:22Applies in terms of data and urban planning?
16:25Yeah, we actually, so people just think mobike is bike sharing
16:30And we just dump bicycle on the road without thinking.
16:35And actually not. We're utilizing a lot of data.
16:39So we actually work with a lot of transportation research
16:42Organizations in china. So what we're doing is we're
16:46Actually using the bicycle riding hotspot, like map, and
16:52Combine that with bus and taxi and those like, you know, ride
16:58Hiring, you know, see how these, those three different type of
17:02Transportation is going to make up the whole transportation
17:06System for the city. So we actually help the
17:10Government to re-planning, re-plan their bus stations
17:14Because of the bike sharing. Right?
17:17Because sometimes the last mile can be covered by bike sharing.
17:21And we actually use those data, well, it's anonymous data, right?
17:25Work with government, like combine those three systems,
17:29Like data from different systems and use that for urban, you
17:33Know, kind of planning, you know, to make people's, like,
17:36Life in the city more easier, yeah.
17:38Super last word, charlie, how much is this like the game sim
17:42City? so, like, the game sim city,
17:45When you interact with the city, there will be an effect, right?
17:48Yeah. I think the most of the time
17:52People forget that the people protest, too.
17:55So i think if you remove, if you increase taxes and you start
17:59Monitoring everybody, there is a danger, as well, of pushing
18:02Them too far and they start going on to the streets, right?
18:06So i think as much as we love ai and data and data collection, we
18:11Have to be wary that being in a city and being in an urban
18:15Setting, it is a political thing.
18:18And decision makers can use it to assist in collection of data
18:22And data analytics, but beyond that, i think we do have to be
18:25Wary about it.
18:27Okay. That's the time we got.
18:29Please join me in thanking all of our speakers.