A conversation with Microsoft Canada’s John Weigelt and Natural Resources Canada’s Vik Pant.
Click to read the full transcript of this episode.
[00:00:00] Alistair Croll: Hi, and welcome to another episode of Industry [00:00:10] Innovations, where we sit down with some of the companies and public sector, individuals who are pushing forward, public private collaboration, and changing the face of what’s possible with technology [00:00:20] in government. Today, I am thrilled to welcome to the industry innovation stage. A longtime friend of the conference, John Weigelt from Microsoft, [00:00:30] along with Vik Pant, who is the chief scientist and chief science advisor at Natural Resources Canada, for a conversation on how AI machine learning and data science [00:00:40] are changing the environment, resource management, and plenty more about society and government. So without further ado, please welcome John and Vik. Hello.
[00:00:50] Hi there. Hi. So Vic, why don’t you tell us a little about yourself and the work you’re doing?
[00:00:57] Vik Pant: Yes. Absolutely. Thank you, Alistair for having me [00:01:00] always a pleasure to join my, my friend and colleague John Weigelton these on these events. And before I talk about myself at the start, I do want to thank yourself and Rebecca and the [00:01:10] phenomenal FWD50 team that is doing great work and bringing together thought leaders and subject matter experts and domain specialists in a safe, inclusive place to have that much [00:01:20] needed a creative collision of ideas. So thank you for creating this platform for us to co-create knowledge and, and value in terms of, for myself. Yes, indeed. As you pointed out kindly I’m the chief [00:01:30] scientist and chief science advisor at natural resources, Canada. We are a federal government department and our mandate focuses on energy, forestry and mining. And our mission is to [00:01:40] help our industries develop Canada’s abundant, natural resource endowments in a way that is responsible in a way that has benefited. Inclusive and sustainable. And what [00:01:50] that involves, what that means at the heart of it is to take in insights from science. And we have a very strong science compliment in our department and infuse them into our [00:02:00] policy making and program design and program operationalization elements so that we can actually achieve the, the goals that the, that the Canadian government has set for.
[00:02:09] Alistair Croll: [00:02:10] For sure. And John why don’t you remind us who you are? You’re you’ve been a part of FWD50 in the past, and maybe tell us a little bit about what Microsoft AI for earth. [00:02:20]
[00:02:20] John Weigelt: Sure thing. And I’d like to echo fix, thanks to the FWD50 group and Vik, it’s great to see you again here in the virtual space. [00:02:30] So I’m the national technology officer for Microsoft here in Canada, and I help organizations using technology while avoiding the unintended consequences. And [00:02:40] recently I’ve also been able to lead Microsoft Canada’s responsible AI team so that we ensure that AI is put in place in responsible ways. Now, [00:02:50] you know, the question of what is Microsoft AI for earth? You know, we see AI as a game changer, a really fundamentally changing and disrupting industries in the way that [00:03:00] we look at the world around us. And so because of that, we created what we call AI for good projects and AI for good looks across a wide variety of the [00:03:10] themes. We have AI for health AI for earth. And they’ll talk to that in the moment. AI for accessibility, AI for humanitarian aid and AI [00:03:20] for cultural heritage preservation. Now, what AI for earth does is it put Microsoft tools in the hands of individuals. It really democratizes [00:03:30] AI for individuals. So we provide grants, technical resources, open source tools, models, infrastructures, and API, to focus on things like [00:03:40] climate, agriculture, biodiversity, and water. Now what’s really exciting cause we have over 700 grants in over 80 countries around the world. And we’re now [00:03:50] building that out to what we’re calling the planetary computer, which is going to help with data sharing around the world.
[00:03:57] Alistair Croll: That’s a pretty tall order. I mean, it [00:04:00] sounds great, but that’s a lot of moving parts. How did the pandemic accelerate or delay some of those projects or like reallocate resources across different projects? [00:04:10]
[00:04:10] John Weigelt: The project kept going. And so it was really great to see that people continue to keep their head down and work hard at the environment because we [00:04:20] see that, you know, while there is this a blip in the CO2 emissions because of the ability or reduced economic output we [00:04:30] continue to create more waste than ever before we continue not to provide those supports so that our natural resources could bounce back. You know, when we start to look at [00:04:40] some of the studies while we pushed down on one side things pop up on the other side, when we think about things like personal protective equipment that is used, or [00:04:50] people are suggesting that consumer waste, you know, when you think about all that packaging that’s gone out there that it has increased. And so it’s, there’s a need more than ever before to [00:05:00] start to look at what these impacts are. And these AI for earth projects help us look at those impacts.
[00:05:06] Alistair Croll: All right. So the NRC collaboration that you’ve got going with the federal [00:05:10] government. It’s supposed to provide technology resources and expertise to solve some of the world’s hardest issues. Vik, can you talk a little bit in a more concrete terms about what’s involved [00:05:20] in the.
[00:05:21] Vik Pant: Yes, absolutely. Alastar happy to do so. And again, always a pleasure to see my good friend, John Weigelt on these conversations. So the main thing here is I want to take [00:05:30] perhaps a slightly more strategic look at this, the partnership we have with Microsoft. So what. When we started looking into a very deep [00:05:40] strategic intertidal at Microsoft, was that typically in the public sector, when, when somebody uses the term partnership they may be referring more to a sequence of repeated [00:05:50] transactions as opposed to a true complimentarity harnessing synergy, leveraging kind of a relationship where truly you, you have allies are working [00:06:00] together to achieve common objectives. So I think for us, the first thing was when John and I started collaborating at the design stages was to say, look, there is no template for this. There is no blueprint for the kind of [00:06:10] relationship that we wan we want to, or to operationalize in the government. So it took us a lot of time to get all the right sort of mechanisms in place from an intellectual property perspective, [00:06:20] from a legal clearance perspective, from a procurement and transactional perspective. So Alistair once that rubric was set, once that foundation was set, then John and I and our respective teams [00:06:30] could come together to value co-create in terms of some of our projects that we’re working on.
So one of the projects we’re working on, which is very interesting, is a project. What we call the [00:06:40] Evie grid project, the electric vehicle readiness grid project, which we’ll talk about we’re also working on projects related to mining safety and mining risk mitigation. [00:06:50] We’re also working on some projects related to something called the energy star detective project, which is a very interesting and useful application of machine vision computer and [00:07:00] also some very advanced information, retrieval information, search and cataloging functionality that Microsoft AI brings to the table. So for us to get to those specific projects, which themselves are, [00:07:10] are very notable and very remarkable I think the thing we do, the accomplishment we don’t want to lose sight of is the fact that Microsoft and NRCan can enter into this partnership as a first of its kind [00:07:20] collaboration between the federal government. And if in any industry thought leaders such as Microsoft, because fundamentally Alastar, you know, what we focus on here is not just completing [00:07:30] individual products, but the collective impact, the cumulative effect that the successful delivery of those projects have. So Microsoft has goals on sustainability as do we, Microsoft has goals [00:07:40] on net zero as do we, Microsoft has goals on carbon footprint reduction as do we. So when we, when we started with that broad strategic macro vision, Lock that foundation down in place. [00:07:50] Then we started to see that doing these types of projects was a lot more viable and feasible compared to had we just started with a traditional model of attempting to partner on these deep immersive [00:08:00] project we would have run into all types of hurdles throughout the way. I think that’s, that’s a pretty common insight for successful partnerships is that you can’t just have transactional [00:08:10] insurances. You need to have. Alignment of, of directions.
So you probably know that at FWD50, we tend
[00:08:17] Alistair Croll: to ask some of the questions that, that are [00:08:20] obvious and then some unusual ones. So I’m going to start with a, an adjacent one. If we’re going to fix climate change I mean, with forests on fire and [00:08:30] subways in New York, flooding and glaciers ripping off seawalls and floating away and so on, we’re going to have to do lots of things. Some of those things are [00:08:40] austerity things, meaning, you know, we’re going to reduce the, the consumption of high carbon footprint, fool foods, for example, or we’re going to switch to electric vehicles. [00:08:50] And some of those things are innovation things like we’re going to use carbon sequestering concrete, or you know, find other ways to innovate [00:09:00] there’s reports of a seaweed that reduces methane emissions from cow burps by like 98%. So there’s this weird, and I know you didn’t wake up thinking, [00:09:10] I’m going to talk about cow burps today, but there is this, this equilibrium with this sort of yin and yang of austerity and innovation when we deal with [00:09:20] natural resource issues and sustainability issues, can you talk a little bit John, about where we were computers can help with [00:09:30] austerity or reduction of harm and where Computers can help with innovation or sort of introduction of new goodness into the system?[00:09:40]
[00:09:40] John Weigelt: Well, you’re absolutely right. I did not think I’d be talking about cow burps this morning. I thought I typically hear about the methane that they produce, but you know, I think it’s a, it’s a great [00:09:50] question that we can put our we put our mind to, you know sometimes we think that we’re doing the right things by perhaps reducing a particular behavior. And what we might [00:10:00] find is that, Hey, what’s actually like a big squishy balloon. I call it, we pushed down on one place only to see that things pop up in another place. And so having the data to be [00:10:10] able to understand what’s happening and be able to reason over that data is critical that. And so when we think about those austerity activities, you know, being [00:10:20] able to visualize and predict weather patterns or ocean patterns, or be able to track and predict forest fires, you know, AI for earth [00:10:30] projects can do those types of things. Perhaps we can look at a hyper-local area where, you know, we have a one project called Terra fuse, which is looking at a very, very [00:10:40] localized environmental impact for organizations. We have our AI project in Calgary with evergreen, that’s looking at the heating of the cities. And so we [00:10:50] see that we can have a sense of what’s happening around the world, around us. And then we can model that to see, well, what happens if we, excuse me, what happens if we [00:11:00] put this particular input or reduce that piece? So that’s one piece I think, on the innovation. We have the ability to look at new materials. There’s exciting advances in [00:11:10] looking at new energy sources the work that NRCan and is doing around the grid for electric vehicles to be able to understand where the traffic flows to be able to enable [00:11:20] and empower commuters or the transportation grits is going to be able to help us predic put those pieces into place. So a wide variety of of [00:11:30] elements that go into helping keep those subways empty well out of water, let’s put it, keep it safe for people and be able to then [00:11:40] really help our global earth be more.
[00:11:45] Alistair Croll: And Vik, what are your thoughts about once you’ve got this data and you’re [00:11:50] able to, to ask sort of, what if questions out of the data as John described, what kinds of questions would you like to ask the data?
[00:11:59] Vik Pant: [00:12:00] Absolutely. I’ll start and just want to completely agree with what John said and just to build on what John said. I want to use some concrete examples. So if we look for instance at the Azula for agriculture product, the [00:12:10] formerly known as farm beets product, and you look at the thought leadership that Dr. Ranveer Chandra who’s the chief scientist at Microsoft responsible for the product lines innovation is responsible for you. [00:12:20] Look at the partnership that now Microsoft, that team has with our agriculture and Agrifoods department under the leadership in this case of Vidya ShankarNarayan, [00:12:30] CIO and assistant deputy minister think about, you know, just what it takes to do to manage and to monitor you. You really can’t manage what you can’t measure and you can’t measure what you can monitor [00:12:40] when you think about farms, which are massive. What does it even take for you? Not even for the actual production. But of the, of the, of the things you’re are growing on the farm, but just to even monitor how your farm is [00:12:50] performing, imagine how much energy it takes to drive those vehicles to, and from project different sports and do this regularly. But now using technology, including drones, using [00:13:00] technologies like smart sensors and Iot and 5g enabled low latency, low energy requiring sensors that Microsoft is really advancing the state of the art with and partnering up with an [00:13:10] organization such as agriculture, Canada, which has subject matter expertise and deep domain knowledge they’re able to build these digital. To enable this precision agriculture and smart farming of the [00:13:20] future. So just think about Aliistar that, you know, you can now build these computational analogs of physical farms where you can do all kinds of what if analysis? What if a pestilence breaks out? What if [00:13:30] this kind of, for this aspect of the farm gets hit by the drought or something like that? So I think that’s a very concrete example where our partnership is unlocking insights between the federal [00:13:40] government, scientists and also Microsoft, I want a quick example if I may, because this is a, we talked about this very briefly earlier is can make mining, which is a part of Enercan led by our [00:13:50] director general Dr. McGee Habib. He’s the overall in charge of this safer mining initiative, the risk mitigation initiative on mining that we’re partnering [00:14:00] up with Microsoft. And one of the things we’re looking at is Alistair. You know, Canada has around 10,000. Mine’s defined sort of in a, in a broad way. And it’s not possible, especially [00:14:10] now in a COVID context to go and physically monitor every mine, abandoned or active to see what its, what its condition is. And certainly you mines have rock piles. They have [00:14:20] emissions bonds, they have tailings lakes. And so if you don’t have. On a very regular basis and you have some mind or a few minds that start leaching into the environment, they can have [00:14:30] devastating impact on the ecology, on the floor of the fauna. But now partnering up with Microsoft on location intelligence and geospatial analysis what we’re able to do is piloting this project. [00:14:40] We’re using satellite imagery and other types of aerial photography imagery. We can look at different mining structures and analyze their health from a distance. And then of course it doesn’t mean [00:14:50] that humans aren’t involved anymore, but now we can prioritize. We can triage, which are those mining sites, which are at more risk of failure in some way. And then we can send our targeted teams [00:15:00] to assess, to manage, to mitigate and to remedy. So really concrete examples, I’ll start aware Microsoft and NRCan and agriculture, Canada partnership. We can talk about environment and climate change too, but I would just [00:15:10] want to say lots of great, exciting partnerships that are coming through the.
[00:15:14] Alistair Croll: Yeah. And I think from the current headlines, I’m thrilled to say all 39 of the miners that were [00:15:20] trapped in Sudbury in the last few days are recovered. But clearly, you know, there are, there are real risks to resource extraction and the more that algorithms and [00:15:30] machines can help us to mitigate and identify those the better I want to pick up on one thing you just said, Vik. I had the incredible pleasure of talking to Dr. Elena Berman the other day. [00:15:40] She’s the chief scientist of a company called Kairos aerospace. And they have sensors that look for methane leaks because as you know, methane is far more impactful on the [00:15:50] planet. But Dr. Berman told me that they use Sesnas because the transportation regulations, this is in the states, the transportation regulations in the US [00:16:00] mean that you have to have a man to drone with a pilot who certified there’s no automation. It seems to me like what you just described, which is drones, looking at farmer’s fields [00:16:10] or pipelines to look for problems predictively and collect data requires coordination between many government departments in that case, transport Canada have to certify certain kind of drone for [00:16:20] use in a certain area. Maybe there’s some kind of safety mechanism to ensure doesn’t start fires. What are your thoughts about the collaboration between [00:16:30] departments required to unlock many of the sort of promises that, that technology can have?
[00:16:36] Vik Pant: Yes, absolutely. I’ll completely agree. You know, when we look at these [00:16:40] systemic changes, when we look at these infrastructure wide initiatives, absolutely one department can do it. I mean, certainly as you, as you just mentioned right now, there’s there needs to be closed interplay, but I think [00:16:50] what I can notice from my perspective is the government of Canada, when it comes to that that level of top leadership in these departments have have professionals such as for instance, Raj, to pull [00:17:00] over a transport or surge visual mean over a transport really from a technology and policy perspective who are not only really well-connected within their department, but are in fact connected with NRCan [00:17:10] and agriculture and eater policy look, you know, regulations certainly, and by design are, are meant to be stable. They are meant to be continuous and bring that sense of constancy and certainty. But also in the [00:17:20] government of Canada, there is a stated commitment to regulatory modernization. How do we streamline regulations? How do we make sure that regulations can be collectively co-developed [00:17:30] across departments. I think where it starts is in two ways. One is within the federal family we have lots of tables. So I, I sit and represent Enercan on various tables where exactly the type of question [00:17:40] which you raised. I think about underground mining vehicles, autonomous vehicles, same kind of who’s the regulations come into play. It’s not just NRCan , who can regulate them. Then we also have to work with transport and [00:17:50] provinces, but I think the other. The partnership with a firm like Microsoft, because Microsoft is a boundary spanner. While we are working with Microsoft Alistair you know, working closely with [00:18:00] John on a variety of initiatives. They’re also working similarly with other federal departments that actually completes this feedback feed forward loop in a very synergetic and complementary way where [00:18:10] not only are departments talking to each other, I’m working with Raj and Vidia and Serge and other colleagues on our regular basis, but also working with John. John is able to sort of be that [00:18:20] boundary spanner who helps to accumulate and collect vies a lot of our shared insights and drive the, the, the, the machine forward
[00:18:27] Alistair Croll: that’s a really interesting point. I mean, I’ve had lots [00:18:30] of conversations around public private partnerships and the need for open source and all these other. But I that’s the first time I’ve heard someone refer to the [00:18:40] private sector as the sort of connective tissue between departments that might otherwise seem isolated. John, have you found that to be the case where you’re able to sort. [00:18:50] Meet three different departments who aren’t aware that they’re collaborating on similar things and then be able to connect them.
[00:18:55] John Weigelt: Absolutely. And it’s been a really rewarding to have that partnership with the [00:19:00] federal government across departments to be able to have these meaningful outputs. And so having a champion like, like Vik, that’s leading the charge, you know, champion, I don’t [00:19:10] think is even the right word. He’s, he’s really an AI superhero for the federal government and really helping us then find those connective tissue that could, that those connections. [00:19:20] And I think as well, what happens is when we look at regulations while they’ve been put in place to last many years, there’s purposeful ambiguity. That’s written into them. I recall my days at a [00:19:30] treasury board secretary at writing regulations and really taking that thoughtful approach to make sure that, Hey, we could have some flexibility. And so sometimes coming from the [00:19:40] outside in helps then. Talk about, helps them have interpretations of the legislation to do these projects. Your observations are absolutely well made around [00:19:50] that connective tissue between departments, where transport, Canada, agriculture, Canada, and our Canada communications research all would need to come together. And these [00:20:00] projects Doing just that. And it’s really, really quite rewarding. And we’re seeing innovations across Canada with the drone flights all over the North. I’m part of a Canary, [00:20:10] the national research and education networks and they have projects, but whereby they have drones flying over the north and knitting together with satellite imagery. And you [00:20:20] mentioned a Sudbury with those poor miners. Great to hear that they’re safe. The first time I encountered a some industries in, in Sudbury was when those otherminers [00:20:30] got stuck underneath the earth in Chile. And it turns out that we have world leading innovation around autonomous mining vehicles built in Sudbury that are being [00:20:40] used in Chile. And so really quite amazing partnerships that start off. These ideas that we land with our government departments and working [00:20:50] with amazing people like Vick and then bring that through to commercialization so that we can maintain our prosperity or Canada’s prosperity on the global scene.
[00:20:59] Alistair Croll: So years [00:21:00] ago I had a fascinating conversation with Daniel Goroff. He’s the director or professor emeritus of the Sloan foundation. And he worked for the [00:21:10] US government as part of their energy nudge program in the department of energy. And if you’re not familiar with it, the energy nudge program sends messages to people to try and improve their energy consumption. [00:21:20] Now if you stay in a hotel and I asked you to reuse your towel. You have a certain likelihood of doing so, but if I say other guests in this hotel reuse their [00:21:30] towel, you, or something like 36% more likely to do so and weirdly, if I say other people who stayed in this room, re use their towel, you’re even more likely to do so, which is very strange, but [00:21:40] these processes are used in energy nudges to change consumer behavior. So for example, if you’re consuming more energy in your neighbors, I’ll send you a thing saying, Hey, you’re [00:21:50] not as efficient as you could be, and you’ll reduce your energy consumption. But if I say, Hey, good job, you’re doing better than your neighbors that reinforces your behavior and you continue to be efficient. [00:22:00] So Goroff did this research and they sent out messages to people telling them what their energy consumption was. And then weirdly they would [00:22:10] take all of the nudges sent to Republican households that said they were consuming more than their neighbors and not send them. [00:22:20] And the reason was that their studies showed that when you send a Democrat a thing saying they’re consuming less more energy than their neighbors they will reduce their [00:22:30] consumption, but when you send her a public and a message saying, good job, you’re consuming less than your neighbors, they will increase their consumption. And so in order to get a net reduction in [00:22:40] energy, which was the program’s goal, they actually had to filter the nudges based on political alignment. To me, this is a fantastic story that just [00:22:50] underscores the nuanced sort of human motivations that come about in trying to accomplish energy reduction goals. [00:23:00] And I know Vik, one of the things that you have going is this energy star detective. Maybe you can explain the energy star detective program and then talk about the sort of psychological [00:23:10] aspects of. Not just having the energy star certification, but getting that, getting the consumers to change behavior in product selection around. [00:23:20]
[00:23:20] Vik Pant: Absolutely. Thank you. That’s a great setup. And I’m reminded of the work also by Bob CLD, chronicled in the book, persuasion, they talk about leavers of influence and one of the influencers they talk about [00:23:30] is social proof. Indeed. You’re right. And I remember also to the point you made around sort of different segments behaving differently to these these nudges. I remember the great economist, Ronald cos I [00:23:40] used to say Nobel prize, winning economist used to say that, you know, the only place where this creature of homo economics exists is in the microeconomics textbooks because the underpinnings of neoclassical economic [00:23:50] theory are that humans are rational, utility maximizing self-interest seeking beings. But in fact, that’s not the case as Dan Ariely and have shown right with, with [00:24:00] behavioral economics that certainly emotions play a big part in our, in our decision process. And many times. Decide and then rationalize expos, but coming to something, you mentioned Anissa, actually, I want to use some very concrete [00:24:10] examples, which are quite timely. So in the PCO department, within our federal government, the clerk of the privy council, we have a fantastic assistant secretary responsible for innovation and [00:24:20] impact called Rodney galley and he’s leading this team that does exactly what you described as behavioral. And they, they build a sign. So they actually partner up with federal department. So [00:24:30] theypartnered up with health, Canada and public health Canada during the early days of the pandemic to understand vaccine hesitancy, to understand the adoption and in some cases, sort of [00:24:40] the spreading of misinformation when it comes to COVID and now actually they’re working with us at NRCan and also eat triple C on the use case that you just described right now, which is when it [00:24:50] comes to energy efficiency, driving best practices among the. Homeowners and people that are using electricity on a private basis, but also remember a big source of energy [00:25:00] consumption and thss emissions are large corporations, not just large corporations, but anybody who occupies larger buildings where offices are housed. So I think what’s very interesting here is that those, [00:25:10] that kind of thinking I’ll start exactly. The thinking that you described is now present in Canada. We are in the process of getting this behavioral science project off the ground to collaboration with. [00:25:20] PCO Microsoft or what we’d like to get Microsoft in the partnership to enter can eat triple C. And as you described as a behavioral science angle to it, you need domain scientists and subject matter experts [00:25:30] from the energy consumption behavioral side, but you also need data scientists because at the end of the day, if we want this to be insights that are inferred or elicited from [00:25:40] information, as opposed to on intuition and gut feelings and anecdotes, we need data scientists, and then we need the compute and we need the data and the platform. So happy to report that, the use case that you described now, [00:25:50] in the case of energy star, that’s also a very important initiative. And what that has to do with is. You know when you buy a large electronic appliance or you see a [00:26:00] dishwasher washing machine dryer, storage, et cetera, you see that energy star logo. Well, that logo is a very specific thing. And what that logo means is that the consumer can [00:26:10] have the peace of mind that that device is going to be energy efficient. That the input to output ratio is, is very calibrated and optimized towards energy saving. Now, what we [00:26:20] find is that there are certain actors in the marketplace as unsurprisingly, whoplay a bit fast and loose with that energy star logo. So this is a [00:26:30] certification you can’t just sort of print it off and put on a device that you want. You have to comply with the rules and regulations. You have to meet the standards. But what some players do is they will take that logo slightly [00:26:40] distorted to have plausible deniability put it on some kind of an online website where they may be retailing, some items, which they are claiming or portraying to be energy efficient, but they are not. [00:26:50] So in the past, what used to happen is a team of human inspectors would basically look at the internet and try to find within the Canadian jurisdiction, where are these [00:27:00] appliances being sold on websites, where they are being advertised as energy star compliant, looking at the logo over a browser, and then figuring out does it look real? Does it look fake? You can imagine doing that at scale [00:27:10] this is not humanly possible. So of course, if you catch a few, you don’t know how many are getting away. So what we partnered with Microsoft on is this very interesting initiative that combined state-of-the-art search [00:27:20] engine technology with state-of-the-art machine learning technology. So now what happens is goes out in our prototyping phase that’s where we’re at, goes out scans websites [00:27:30] that are catering to the Canadian retail space. They find those listings and catalogs online where energy star devices are being advertised, and then they can [00:27:40] download the image automatic. Like a search engine, retreats content, and they can then score it against a benchmark of approved golden records that John’s [00:27:50] team. And my team have actually collectively trained the machine learning models on. So now what it does is you’re managing by exception as opposed to managing by sort of random sampling. And population-based sort of, you know, okay, [00:28:00] let me look here. Let me look there. So what this tool now does energy star, detective it flags possible misuses or abuses of the energy star logo online. And now our [00:28:10] we can actually have a much more targeted, focused in scope intervention where now they can actually investigate that matter much more in detail. John.
[00:28:18] Alistair Croll: A use case I [00:28:20] never would’ve thought of yeah. John, John.
[00:28:23] John Weigelt: Yeah. Well, I think, you know, it’s a, it’s a fantastic use case and. I could imagine that, you know, that [00:28:30] energy star would be a forged online. But, I think the value of the program is, is just so tremendous. The program partnership that NRCan has and, [00:28:40] and the extent of which was to me, quite amazing, not only for. Those appliances and those other energy consuming devices by going off to buildings as well. So it’s a huge, [00:28:50] great program, I think to your point Alistair around these gentle nudges, the gamification I think is is a really great thing to help people be aware of what’s happening in [00:29:00] their environments and what’s happening around them. You know, when you were speaking about being able to look at political affiliation you go, I, I always think about a polite dinner conversation and you never want to [00:29:10] talk about politics around the dinner table. But it also raises this question around what data are you going to use to make your [00:29:20] decisions or to read. And some data uses are quite benign and don’t deal with personal information than others might start to then creep into that environment. And so in [00:29:30] addition to these tools, it’s important to put in place processes and practices to ensure that AI is used responding. We’re responsibly. And you know, a government of Canada is one of the [00:29:40] leading first runners, to put in place algorithmic impact assessment guidance. And so when you’re using these tools for reasoning [00:29:50] over data, you must go through an algorithmic impact assessment to talk about fairness, to talk about or to, to make sure that your system is fair to make sure it’s reliable and [00:30:00] safe to make sure that it’s a secure. To make sure it’s inclusive for everybody that needs to access the tool be accountable and be transparent. [00:30:10] And so responsible AI is key. And so that’s where some of the, the tools and the theme that I have is able to then review these uses that [00:30:20] could be called to question. So if Canadians were to get those gentle nudges, we want to make sure that we’re doing everything properly and we have a tool [00:30:30] that aren’t infringing upon people’s rights that we have a way that make sure that these tools are used for the betterment of the community. And then we’re seeing, you [00:30:40] know, the betterment of the community come through with these first initial projects that NRCan and Microsoft are working together. But really then to [00:30:50] champion them to go that much further.
[00:30:52] Alistair Croll: So I, I like the idea that it’s this sort of collaboration between human and machine that the machine is saying, Hey, there might be thinlike. [00:31:00] It’s similar to what I’ve seen done in radiology, where an algorithm can tell the doctor, Hey, you might want to check out these spots on the lung x-ray, but then the AI misses that someone has painted [00:31:10] a picture of a gorilla on the x-ray. And so the human has to go, no, that’s a gorilla. And I think that that, that human machine kind of hybrid collaboration. Produces [00:31:20] the best results as cast are showing these Camaro chess games and so on. And John you’ve been involved in putting together some of the AI ethics and AI policies. Can you talk a little bit [00:31:30] about how you’ve built in sort of human involvement and, and sort of this, this recourse into this.
[00:31:37] John Weigelt: A hundred percent. And that’s a, that’s a great question. [00:31:40] Because when we start to lose our awareness of what’s happening behind the scenes, then all of a sudden things can go awry. And we’ve seen that with some of the automated vehicles, for example, you [00:31:50] know, you’re intended to have your hands on the, on the steering wheel. But, it turns out that your reaction time is particularly slow. And so by the time that you are back and [00:32:00] cognizant of what’s happening, then. It’s too late. And so this all fits under the accountability principle and making sure that people are accountable, that you [00:32:10] keep those humans in the loop for those safety, critical activities. And I think your point’s well made around this idea of, you know, mixing things up for the radiologist, for [00:32:20] example, because it’s human behavior, we’re talking about human behavior a little bit earlier but it’s human behavior to say. You know what it’s been okay the per previous, 10,000 times, this [00:32:30] 10001st time it’s going to be okay as well. And so, you know, we have teams that look at user experience from our years of working with just the commonly used [00:32:40] productivity tools. But now they’re looking that next step to make sure that Hey, people are really aware and understanding of the decision that they’re making. And that’s, that’s [00:32:50] the decision that’s being made is actually reasoning over the the right information. We’ve all heard about the AI models that built, that we’re trying to reason over [00:33:00] whether or not a picture was a Wolf or a dog. And it turned out that it was actually reasoning over the background. It turned out that the pictures that were used to train them on were [00:33:10] all wolves in the winter and the dogs were all with green backgrounds. And so we didn’t get the right to reasoning over that. And so accountability and transparency [00:33:20] really then help us make sure that we’re making the right decision to your point. It’s not AI that’s going to be the best decision maker. It’s not people, that’s the [00:33:30] best decision maker. It’s the people and AI, that reason the best income to the best conclsuions
[00:33:36] Alistair Croll: So Vik I, you mentioned earlier, you know, making [00:33:40] decisions based on data instead of shooting from the hip, I’m reminded of bruno’s hit and came study from, it came out a few years ago that showed across 179 [00:33:50] large publicly traded organizations companies that use data-driven decision-making do five to 6% better per year in terms of [00:34:00] productivity and profits than their competitors that shoot from the hip and that obviously compounds. So can you talk a little about how the shift has gone from, [00:34:10] you know, once upon a time, the leader was the person who convinced others to act in the absence of information to now the leader is the person knows what questions to ask of the data to make the [00:34:20] best decisions. Can you talk a little bit about how that.
[00:34:23] Vik Pant: Absolutely. Alistiar. That’s a, that’s an excellent question. And one, I think that is not often enough spoken about, and this is [00:34:30] something John and I have talked about at length, and this is another very concrete collaboration that Microsoft and NRCan are working on Alistair, you know, whether it comes to corporate strategy in a large organization, [00:34:40] whether it comes to innovation, as you touched on at the heart of any kind of economic activity, progress, prosperity is innovation is the notion of value creation, value [00:34:50] allocation and value distribution. Now I think a lot of times, because AI is so much in the zeitgeists and everybody talks about AI all the time.People presuppose that if I put AI into it, it’s going to [00:35:00] automatically translate into value. So one of the research areas that John and I are looking at and our teams are collaborating on is to say, hold on, stop for a second. Does it even make sense to use AI and [00:35:10] data science in this context or not? That doesn’t mean that doing some kind of information analysis or data sciences and running the degrees to which you can apply AI to a problem [00:35:20] space needs to be calibrated sort of on a spectrum as opposed to zero and one. So one of the things I want to talk about very quickly here at the start is for instance you know, something that John and I talk about [00:35:30] is when we talk about AI, very often, people think about, okay, here’s an AI product or project that I’m going to do, it meets this specific, immediate objective of mine in the enterprise. And [00:35:40] that’s it. And that’s on now that we move to the next one, what we’re looking at. Can there be a framework that you can use to assess the worst and the value added and the added value [00:35:50] of data and digital solutions, AI, and otherwise to meeting a broader set of organizational objectives. So I think to your question at the start, when you are a leader if [00:36:00] you look at your enterprise AI interventions as isolated, one off siloed experimental exploratory enterprises, that’s fine, but you’re missing a lot of [00:36:10] complimentarity you’re missing that synergy across the table. Of course there is some work done on this kind of a value assessment framework, but there isn’t a generally accepted one [00:36:20] out there as that. So happy to also partner with the FWD50 team on this initiative. But John and I already quite well on our way to put our thoughts together to say, okay, how do we go from the data sets to the [00:36:30] algorithms, machine learning models, et cetera. But that’s the one, where’s the soul. What the soul, what is when you map it to the financial and non-financial organizational objectives. [00:36:40] And I think there’s a lot of great partnership opportunities there.
[00:36:44] Alistair Croll: I like that analogy of you’re you’re teaching people how to fish rather than getting them an AI fish. [00:36:50] So we are already over time, but if it’s okay with you, I’d like to keep going for a couple minutes. Cause I have a couple more questions I want to ask. First of all Microsoft has been a supporter [00:37:00] of the open AI project. And I think open AI was created for those that aren’t familiar because many of its founders felt that AI technology should not be in the hands of a private [00:37:10] few, but rather should be open-sourced and publicly available. And that this stuff should be licensable and, and sort of the work done in public. When GPT three [00:37:20] came out as a result of open AI, obviously companies like Microsoft have access to it and the power of that stuff is incredible. But at the same time, it’s very experimental. Like we [00:37:30] look at opening a GPT three, and the other products that open AI are producing on the one hand we see tremendous power. On the other hand, it’s trained on a data set that [00:37:40] maybe doesn’t represent the things that we want it to represent. So you can, for example, ask GPT three, why vaccines are a hoax and it will tell you because it’s trained on [00:37:50] humans, writing about hooks vaccines. At the same time you mentioned enterprise AI, Vik. Today I have PowerPoint and I have buttons like copy [00:38:00] and paste and undo that I’m very familiar with. It’s entirely feasible that in a few years, we’ll have a button called go where I write a paragraph in Microsoft word and hit go, [00:38:10] and it finishes the document for me. And maybe I’m going to go edit that document because it’s wrong. But it’s certainly an interesting function to have in an enterprise productivity suite. [00:38:20] We’re already seeing sort of AI powered design and PowerPoint and other forms of assistance. I think that there’s an interesting question [00:38:30] here about whether the baby steps of information productivity assisted bay by AI can sort of [00:38:40] raise the tide of AI adoption by everyone. So that as opposed to waiting for the day, when a magic robot walks into our house and solves all our problems, we get AI [00:38:50] one button at a time and productivity suites and start to learn how that works and change the way we work. John, can you talk a little maybe open AI and how you see [00:39:00] more democratize the adoption of AI happening in the everyday life as, as something that would affect the average worker or a citizen.
[00:39:06] John Weigelt: Certainly. So you know, one of the things that we’ve seen [00:39:10] is that there is far more at Microsoft than ever before our contributions to open source and open projects and collaborations with even organizations [00:39:20] that you would characterize as our competitors you know, have increased dramatically over the last over the last few years. And so open AI is yet another example of that [00:39:30] collaboration to see, we know what’s happening in this space and to contribute to this space, we have fantastic infrastructure components that can support these large. Tens of [00:39:40] thousands of CPU cores that can support the investigation and analysis. And, you know, there’s recently a project it’s open AI for energy. That’s now [00:39:50] moved out and so expanding into almost industry focused type activities. One of the things that to talk about the democratization and one of the things tha people don’t [00:40:00] recognize is that, Hey, we use AI in our, on a regular basis. When we think about machine translation, when we go to a foreign website and look to. It make sense of the language that [00:40:10] there, because we don’t understand that, you know, we use that into this transparent, or if we’re using a wayfinding application, you know, that’s helping us using AI. And I think everybody should [00:40:20] pay to have a quick look at what the tools that they use on a regular basis. I was at a chamber of commerce conversation at a one of Canada’s smaller cities [00:40:30] and the chamber of commerce said, so what is AI going to do for my constituents? You know, my carpenter, my plumber because they’re not data scientists. And then when you start to explain the [00:40:40] wayfinding applications or the time management applications or some of the fraud detection tools that are built right within some of the productivity suites, [00:40:50] you know, that then helps put a face on to AI for those, casual data scientists. Think about what Excel spreadsheets did for. You know, the [00:41:00] financial community, and then expand that out. We all have some familiarity with using spreadsheets and being able to, you know, understand how to work with the columns. AI is going to [00:41:10] has become the same thing with low code, no code applications that harness the power of AI that’s in the background. And so I think that’s a particularly exciting [00:41:20] in this domain.
[00:41:22] Alistair Croll: Yeah. I think I was showing a bunch of people in the Canadian federal government, the PowerPoint live translation where you speak in English and the text [00:41:30] appears in French or vice versa and they were blown away. And, you know, my point was, why isn’t this a default if we’re a bilingual country, just every meeting you do turn that on. And it’s all machine learning and it’s all [00:41:40] pretty straightforward. I think I want to wrap up with this quick question and John, you’re welcome to chime in here. Vik, you mentioned earlier about the cross pollination between departments and the [00:41:50] role that the private sector has to play in that. Climate change doesn’t no borders, pollution from the states blows up here. You know, we’re seeing forest fire smoke from [00:42:00] California, make its way to New York. Obviously it’s a global problem across all of its consequences. What do you think about the role of the private sector in, in encouraging [00:42:10] collaboration across countries?Not just across departments within government or across Europe?
[00:42:16] Vik Pant: Absolutely. Yeah, let’s start. I think that’s an excellent question. If you think about [00:42:20] it, I think it starts with an acknowledgement that nobody knows everything and everybody, you know, something, especially when it comes to these massive institutional systemic type issues you’re going to [00:42:30] have to collaborate if you want to meaningfully. In finding a solution to that problem. Now, when we look at a firm like Microsoft, one of the things we appreciate is that Microsoft is a [00:42:40] transnational, multinational, global, truly international firm. And so their expertise are decentralized and federated all over the world. They have connections just as John has a respect and [00:42:50] authority within circles within Canada, but certainly John has colleagues in other parts of the world as well that then collectively brings those networks of knowledge and networks of resources together in a [00:43:00] meaningful way where we can advance these dialogues in a, in a, in a complimentary manner. I think the second part of it is that technology is something that is not just an enabler. Often we hear the word [00:43:10] enabler or support a technology is actually a very important driver. It’s not just a catalyst to start the conversation with it’s something that is going to be crucial before. I see this through [00:43:20] Alistair, you talked about concrete and there’s lots of other interesting, you know, building products. That’s a whole area unto itself. Transportation, that’s an area. Agriculture is an area resource development is an area. [00:43:30] So here, each of these areas have very specific technologies that can be either retrofitted or upgraded or in some cases completely designed. They Novo to with [00:43:40] those design characteristics in mind that lead to reduced reduced emissions and enhanced efficiency, but also then more broader kinds of of technology plays such as, as John [00:43:50] talked about the planetary computer and the Microsoft data trust model data mesh model, where you can have players in a cooperative frame, as John mentioned, not just pure competitors or pure cooperators, [00:44:00] but organizations that are bound together by a larger objective, which is to fight global climate change but at the same time also have their private interests, which they require to fulfill [00:44:10] their fiduciary obligations. They can play a great role in indeed. I would know from my partnership with John, that they are playing a great role.
[00:44:17] Alistair Croll: So we’ve covered a lot of stuff today. And [00:44:20] Vik, I think we touched on the energy star detective program, the Evie grid readiness program, the mining risk mitigation program. It sounds like this is just the tip of the iceberg. [00:44:30] Although maybe invoking icebergs is a bad idea for global warming. But it does sound like this is the start of, of a sort of, you know, a thousand points of light that eventually change [00:44:40] how governments work. Thank you so much both of you for spending some time and teaching me so much about this stuff. I really, this is a fascinating topic and I think we touched on a number of different things [00:44:50] that I love this concept of sort of cross-organizational collaboration and cross departmental collaboration. That can be cross-pollinated by. The private [00:45:00] sector. Vik, if folks want to find out more about the various things you’ve touched on here, what’s the best way for them.
[00:45:06] Vik Pant: Absolutely. So use your search engine of preference, Bing or [00:45:10] otherwise, and you can look for natural resources Canada, digital accelerator, and there’s a website. We list our projects. We have a contact us form, reach out to us. We’re very responsive. And again, Alistair, I do want to [00:45:20] thank FWD50 and the phenomenal work you do. I mean, indeed FWD50 is one of the organizations that brings thought leaders together that brings practitioners, policy leaders, program experts together [00:45:30] in that safe, inclusive space where we can have a creative collision of ideas. So thank you again for the great complementarity and synergy that you create and infuse into the, into the global system.[00:45:40]
[00:45:40] Alistair Croll: Thanks, Vik. John, what, where can people learn more about the stuff that we’ve covered today?
[00:45:45] John Weigelt: So a quick thanks for this pointer Vik around BIng [00:45:50] a quick bank search of Microsoft AI for earth. We’ll show you all those resources and, you know, it’s it’s always a pleasure to be here at FWD50 with yourself [00:46:00] Alistar and you know, I’m always invigorated by, by Vik with the his great ideas. And, you know, he really is an AI superhero. And you know, when we talk about these collaborations [00:46:10] like globally, you know, we punch above our weight when we are able to share the examples of what we’re doing in Canada and to your point I’ll star, these are just the initial points of light. You know, this, [00:46:20] they were there to help us build this foundation for projects going forward. And so look for bigger and brighter things coming from Vik and his team at NRCan. And then [00:46:30] we hope to be part of all their innovation as we go.
[00:46:34] Alistair Croll: Awesome. Well, thank you both for being here. I really enjoyed the conversation and looking forward to seeing more, both of you in the coming [00:46:40] months.[00:46:41]
John Weigelt: Thank you.
The rapid pace of innovation has given us a world yesterday’s sci-fi authors could scarcely predict: The sum of human knowledge in our pockets; instant video communication around the world, basically for free; robots and algorithms taking on work that was once thought impossible. But this rapid progress has also dramatically impacted the world in which we live. Resource extraction, carbon footprints, congestion and crowding—today’s crises are the direct results of our achievements.
The solution is likely to blend austerity—wiser, more efficient use of what we have—with innovation. And in both cases, the tools that brought us here can help us move forward.
Vik Pant and John Weigelt are no strangers to innovation. Weigelt, the National Technology Officer at Microsoft Canada, spoke at FWD50 in 2017, and took the stage with Pant—Chief Scientist and Chief Science Advisor at Natural Resources Canada—in 2020.
The two have been behind a number of applied AI efforts to improve traffic congestion, mine safety, and energy consumption. In this episode of Industry Innovations, we delve not only into the tradeoffs of technology, but also the role of the private sector in encouraging cross-department and cross-jurisdiction collaboration.
Learn more about Microsoft AI for Good, which provides technology and resources to empower governments and organizations working to solve global challenges to the environment, humanitarian issues, accessibility, health, and cultural heritage: www.microsoft.com/en-us/ai/ai-for-good