- il y a 1 semaine
How Can Tech Help Us Preserve the Natural Environment?
Catégorie
🤖
TechnologieTranscription
00:00Hi everyone, thank you for joining us. I'm Freya, I'm a reporter at Sifted. I write about climate tech and
00:08we're going to be talking about the role of tech in preserving nature.
00:12So I know there's a lot of discussion about the harmful effects of technology on nature from e-waste to
00:18the energy demands of AI.
00:21We're going to talk about the flip side of that and the role that tech can play in preserving it.
00:25So I'll run down the panel for a quick intro and then we can get into the questions.
00:28Emily, do you want to start off?
00:30Thank you, thanks for having us. My name is Emily Charitissier. I'm the CEO and co-founder of Whale Seeker,
00:36a Canadian company that uses AI to detect marine mezzo and megafauna from imagery.
00:43Carl?
00:44Great, good morning everyone. My name is Carl Hamilton. I'm the head of the digital department in the European Environment
00:50Agency and we're a decentralized agency of the European Union, which I'll talk more about in a few minutes.
00:56Hi everyone. My name is Tom Elliott from a startup called Restore. It's a spin-off from the Swiss Federal
01:03Institute of Technology and it is democratizing access to a suite of environmental and biodiversity data to a global network
01:13of local communities and indigenous populations.
01:17Hi everyone. My name is Stuart Minchin. I'm the director general of SPC. SPC is the intergovernmental organization in the
01:26Pacific for science and technology.
01:29We work on behalf of 27 countries across the Pacific providing science and technology to support development.
01:37Let's start off kind of setting the scene then. Why is monitoring nature using technology so important? Carl, should we
01:44start with you first?
01:46Yeah, thanks Freya. So I think the first thing to start with is looking a little bit at the context
01:50of how we operate.
01:51So as I mentioned, we're a decentralized agency of the European Union, but we work with 38 countries, not just
01:57EU countries.
01:57And our role is really at this science policy interface. So we collect data across many different areas and we
02:05perform assessments on that to inform the state of the environment and the impact of certain measures.
02:10So the role of technology and actually helping us to generate that data, and of course the data is provided
02:16to us by countries and industry, is really fundamental.
02:20In terms of the need for the data, we have to make sure that it's a quality check, that it's
02:26accurate, that it's reliable.
02:28This is absolutely crucial because our research and work has to be based on evidence.
02:34And we collect over 140 regular data flows from marine ecosystems, the state of greenhouse gas emissions, noise, pollution, etc.
02:45So it's really important we have the technology in place and countries have the technology in place to be able
02:49to actually provide this data.
02:53You asked about the gaps as well.
02:55When we talk about gaps, I think what's really important to note is that the legislation that is being complied
03:03with by countries can mean that the data is very often reported every six years.
03:08It can sometimes be near real time when we talk about air quality data.
03:11So when we talk about gaps, there's more and more a need for much more real time, near real time
03:18data so that we can have much more regular information on the state of the environment to be able to
03:23take action and inform policymakers and citizens.
03:27So that's a really key point when we look at gaps.
03:30We also see there's a need for more socio-economic data and risk data so we can actually see the
03:35impact on health, on insurance claims, on economies.
03:40But then there's also gaps in how we provide access to that data.
03:43Data can be very fragmented.
03:45The previous panel talked a lot about this, I heard.
03:48So it's about providing access to that data in a way that can be used and actually inform decision making.
03:52And I think the third gap, just to flag briefly, is also, I think, on the capacity.
03:57We need to make sure that people actually are trained and able to use the data in ways that actually
04:02inform the type of work that the colleagues here on the panel are also undertaking.
04:07Great. I want to get into all of those points in more detail.
04:09First, Stuart, perhaps similar question to you.
04:12What's the role of tech-enabled data in the work you're doing and gaps and problems you see with it
04:17at present?
04:19Well, firstly, to give you some context to understand the Pacific, the Pacific is about a third of the planet
04:25and only 2% is land and islands.
04:30So these islands are very spread out with great distances between them, small populations, so a very small capability to
04:40actually monitor and measure what's happening in their jurisdictions.
04:45So we need to use technology to cross that gap.
04:49We need to use Earth observations from space.
04:52We need to use remote buoys, for example, for oceanography and other things that can allow these communities to manage
05:02the huge areas that they're responsible for.
05:06This is also the area where the biggest challenges are facing us at the moment, the frontline of climate change.
05:13So an example, and you can probably see this if you visit the Pacific Tech Village outside, is actually describing
05:23and identifying for decision makers globally the impacts and showing them tangibly how it's making a difference.
05:33So there's a 3D printed model of Tuvalu, which is going to be the first country to disappear off the
05:41map due to rising sea levels.
05:44And we've done detailed modelling with Tuvalu to identify how different scenarios of climate change will impact their country.
05:52And you can go and experience that yourself outside.
05:54So I think tech can really help us not just to monitor these changes, but to actually change decision makers'
06:04minds to actually take action now.
06:09Tom, when we talked before this, you brought up the opposite side of the question, which is perhaps having too
06:15much data and not knowing how to differentiate or which to use.
06:19Talk us through that point more.
06:21Yeah, I wouldn't maybe say there's too much data, I think, coming from a lab of scientists that just devour
06:29data.
06:30I think they are always welcoming more and more information that's going to help us really understand and refine the
06:36models and refine land management strategies.
06:40But what I would say is I think we have enough data to be able to make decisions that sometimes
06:46aren't being made today.
06:50You know, Restore started five years ago with, you know, NGOs and local communities coming to us to say, how
06:58can we get access to the data that we understand is now out there?
07:02We're seeing it reported and talked about in terms of the ability to tell me what the potential tree cover
07:07is for my site, the carbon storage in my location and providing access to all of that has allowed better
07:17land management strategy on the ground.
07:18The next thing we're starting to see is hesitancy from investors and from governments and companies on saying, well, we
07:26don't think the data is good enough and these businesses are too nascent and they don't have a big enough
07:31track record.
07:31Whereas I think we know that biodiversity is degrading around the planet and we know the land management strategies that
07:39are causing that degradation and the ones that are improving it.
07:42And so I just think there is enough data there to allow things like investment to flow into the communities
07:47that are protecting and restoring the planet for their own livelihoods as well as ours.
07:51Stuart, do you want to come in?
07:52Just to follow on, I think the issue of too much data is an interesting one. And if I could
08:00use an example from the Pacific. The Pacific has these challenges of climate change, as I said, sea level rise,
08:08but Pacific was not making use of all the satellite imagery that the world collects over the region because it
08:15was just too hard to actually process all of that information.
08:19We have 750,000 satellite images across the Pacific from the last 25 years, but that's about 800 terabytes worth
08:28of data. It's too much for people to use as raw data. So you've got to have services that transform
08:34that into decision ready products that people can make decisions on.
08:38And that's what we've done with a tool called Digital Earth Pacific. So we can now track every island, every
08:45coastline, every beach in the entire Pacific continent and see the coastal change over the last 25 years, the health
08:52of mangroves over the last 25 years, the water presence or absence, bathymetry.
08:58We're producing these products because the countries themselves and the people in the administrations don't have the expertise or the
09:07capability or the computing to handle that kind of analysis. So we're doing that in the cloud, making it accessible
09:13so that they don't have to be remote sensing specialists to do that.
09:18Emily, I want to dig into the specifics of tech in different natural environments and perhaps innovation that's coming through
09:26and how it's progressing. Can you talk us through the marine environment, which is where you work, and particular recent
09:32innovations that you see coming through that are going to change how we measure?
09:36Yeah, this is very timely. I've just arrived last night from the United Nations Oceans Conference. And this is what
09:43everyone is talking about. We need more data because I think having enough data is probably true for terrestrial ecosystems
09:52and perhaps coastal ecosystems that are populated, but we do live on the blue planet and there is this big
10:00question of how do we have data acquisition at that scale?
10:04And so satellite imagery is a really good example. And then how do we make it? What is the tech
10:12aspect of making those data interpretable for decision makers? But also, how do we convince the private companies that own
10:21those technologies not to throw away those data that we can use them?
10:27And so, you know, if they're not selling them at whatever, whatever is the commercial price, that the world still
10:36needs those data. So I think the market needs to go quite a ways in terms of sort of breaking
10:41down silos and understanding the value of data that can be used cross sectors.
10:49And so you spoke about how good our data, how they need to be trustworthy. And I think that's another
10:57big impediment because most people who are, I'm assuming most people in this room are AI savvy.
11:05They don't see it maybe as a black box, but I can promise you that the majority of the world
11:12does. And how do we enable them to be critical thinkers about the tools?
11:18Because we absolutely need AI as a tool to get through this amount of data and to pull out the
11:25important decision making tools or important decisions for each use case. And AI absolutely has that capability.
11:34But how do we, how do we, how do we make sure that we have that, that standard of quality?
11:41Yeah, I just briefly wanted to chip in a little bit there because this, um, you, you mentioned, uh, Thomas
11:47about, uh, we have lots of data and it's actually true. We, we've never had more data effectively, but is
11:53this a fragmentation of the data?
11:55Or as you mentioned, Emily, you know, there's a lot of data out there, but it's actually making it accessible
12:00and combinable and cross sector analysis ready. Uh, and that's what we're seeing as well.
12:05And I think initiatives like at the European level, like the, uh, the data spaces is part of that and
12:11making sure that the data is open, uh, that it's actually usable and it can be used for data as
12:16a public good.
12:17And also for companies and investors to be able to take that data and, and, uh, generate solutions from it.
12:22Just wanted to briefly come in on that point. I think it's a really important one.
12:26Yeah. Maybe I would also add around, uh, measuring biodiversity, you know, was five years ago when we were talking
12:33to, to boardrooms, it was, uh, and we're trying to tell them about the importance of biodiversity.
12:38The message was coming back that can you please communicate this in the context of carbon because my CEO isn't
12:44going to respond to this unless it's, it's framed in that context.
12:48And now we're seeing far and far more that the, the, the importance of biodiversity is being recognized because value
12:54chains depend on it, et cetera, and so forth.
12:57But this, this thing about how do you understand biodiversity as a concept is really hard.
13:02And that's something we've been working on with the science team, um, uh, uh, restore is to try and, uh,
13:09create a framework to ingest a lot of the new data that's coming online.
13:13So previously we've seen indicators of biodiversity being used by banks and financial institutions, basically how close are you to
13:21a protected area?
13:22How many threatened species are there or what's the abundance of species?
13:27And the reality is that biodiversity is far more complex than that.
13:31And, uh, we, it's a shame I don't have a visual, but we've created a kind of radar chart across
13:35genetic species and ecosystem level diversity.
13:39And it can, to ingest all these different data types across these nine different axes, three within each of those
13:45three I mentioned,
13:45to then try and give a better perspective of what the state of nature is in any location.
13:50And that will just evolve over time.
13:52But if we can agree on some broad principle of how to interpret biodiversity with a reference ecosystem, we think
13:58that's got great protection.
13:59And are you finding, uh, those board members, CEOs you're talking to, or people using your data to talk to,
14:06are they more receptive?
14:07Are they moving beyond carbon as this measurement now?
14:09Like, is the data working in that way, do you think?
14:12Um, uh, I think we're still, there's still a lag.
14:15I think there's definitely more talk, uh, of action.
14:18And there's initiatives like the Task Force for Nature Related Financial Disclosures and the Science Based Targets Initiative.
14:24Um, but I think there still is hesitancy.
14:27Um, and companies are waiting for others to lead the way.
14:30Uh, and they're also wanting clarity on what is this framework to measure biodiversity.
14:35So we can all be on the same page because we don't want to lead and then find out it's
14:39not good.
14:39You know, that's what I'm finding is that.
14:41Yeah.
14:42I've seen too that there's a big emphasis on standardizing data, but everyone wants, uh, people to standardize it to
14:49their method.
14:49So I think initiatives like yours that, that actually ingest everything and then give a decision-making tool is really
14:56what's needed on a greater scale.
14:58And I know the oceans need that desperately too.
15:01Because I know we're creating marine protected areas for species that don't even exist in those areas anymore because their,
15:07their ranges have shifted.
15:08And so how do we move away from carbon for carbon sake, but also we need, if we're looking at
15:15carbon and we are just taking a carbon credit, a monoculture is not, is not good carbon credit either.
15:23So we need to protect that biodiversity if we're looking at these assets and how do we ensure those assets
15:29and how do we recognize those assets as, as collateral that countries can have too.
15:36Do you have it?
15:38Thanks.
15:39Two comments.
15:40Um, one on this issue of, uh, um, you know, access to data.
15:45Um, I think it's, uh, been a complete game changer that what decision that happened back in 2008 to make
15:51Landsat data freely available for the world.
15:54That led ultimately to the Copernicus program from the EU being, making free data as well.
16:00The 750,000 photographs that I talked about from space across the Pacific would have cost you nearly a billion
16:08dollars to, to access prior to that change.
16:11So it would have put out of reach any of these tools for, for the region.
16:16So it's incredibly important that we keep, uh, reviewing, um, the value that we can create from this data, not
16:24just its commercial value, but the value of it being used.
16:28Second comment, um, is about this carbon credits, uh, issue.
16:32And, and I'll share an example from the Pacific, um, that's trying to tackle, uh, this, um, combination of carbon,
16:40biodiversity and, and other things.
16:42And it's, uh, uh, a little, little country called Nui in, in a little small island in the Pacific.
16:48And they've taken their exclusive economic zone and they've, um, worked out how much it's gonna cost them to patrol
16:55that, to protect it, um, to monitor it over the next 30 years.
17:00And then they've worked out their capital investment that would be required to, um, to, um, set up a trust
17:07to, to pay for that work.
17:09And they're now selling that as, um, what they call, uh, ocean conservation credits.
17:15So it has a blue, it has a carbon credit potential, but it's also a biodiversity potential.
17:20It's whale sanctuary.
17:22It's, uh, it's, uh, um, you know, incredible coral reef.
17:25And, and, and you can buy those credits today if you look up the Nui Now Trust.
17:30Um, you can buy that for your family and, uh, and, and, and elsewhere.
17:34But they're trying to capitalize that as a mechanism to not just be focused on carbon, but focused on, uh,
17:41the broader values of biodiversity, ocean protection, and carbon.
17:46And I think this is gonna be more of a model that we see as we go down, um, into
17:51the future where people aren't just focused only on the one issue of carbon, but on the broader, uh, protection
17:58of ecosystems.
17:59The other side of this I want to bring in is using tech and data for adaptation resources.
18:06Carl, perhaps you want to talk about some of the work you're doing on that?
18:09Yeah, I think that's a really important point.
18:11You know, we obviously climate neutrality and, uh, decarbonization is really important, but we're seeing an increasing focus on climate
18:17adaptation.
18:18Um, and this is a big focus for many regions, including Europe, um, and it's about, uh, autonomy, um, resilience,
18:25et cetera, across sectors.
18:26Um, for example, the European Environment Agency, we released the first EU climate risk assessment last year, and some of
18:33the, uh, findings from that are really, really stark.
18:36You know, we know the planet is heating, uh, but Europe is heating twice as fast as any other continent
18:41on planet Earth.
18:43Um, and this is leading, as we know, to, um, more extreme weather events, um, heat stresses in populations, et
18:49cetera.
18:49So, it's really important that we take action, not just to decarbonize, but actually to deal with the impact of
18:55the inevitability of climate, um, temperature increases.
18:59There are a few areas in terms of how tech can help with that that I think are really important
19:02to flag.
19:03One is the role of technology in solutions and sharing solutions.
19:07So, we run a climate adaptation, um, portal with the European Commission, and that's effectively looking at good examples of,
19:15um, climate adaptation measures across cities, regions, et cetera.
19:18So, by providing access to tools and solutions, including the types of solutions we see here today that are being,
19:24um, developed by innovators, that's really important.
19:28Um, the second is, um, artificial intelligence.
19:31You know, there's so much information and data out there, um, we can use artificial intelligence to help navigate that
19:37vast trove of data.
19:38So, helping, you know, if you're working in a municipal office, having thousands and thousands of pages of data and,
19:44and, and knowledge is difficult to navigate.
19:47So, AI can actually help with navigating that and also developing, uh, climate adaptation plans is another area.
19:54Um, digital twins is another area we're seeing growth in.
19:57Um, I know there's a digital twin for ocean that's been launched and is being developed.
20:01Um, we also worked with, uh, the European Space Agency, ECMWF, and the Commission to develop a digital twin for
20:08climate adaptation.
20:09And what's really powerful about that is we can, uh, look to model different climate scenarios
20:14and then see how that could have a regional, local, national impact and how effective those measures might be in
20:21different climate scenario outlooks.
20:22So, it's a really powerful tool and we hope to build the next European Union climate risk assessment on digital
20:28twins.
20:29Um, another area is Earth observation that Stuart also touched on.
20:32We have a role in delivering part of the Copernicus program at the EEA.
20:36Um, and this is a massively important tool because it can give us much more near real-time data, um,
20:42on different, uh, phenomena, water, uh, drought, wildfire, extreme weather events, um, land use, uh, bioeconomy, food production, etc.
20:52So, we've got this tool in our arsenal as well, which I think is really important.
20:56And then, of course, for climate adaptation is the solutions that businesses and innovators are developing.
21:01You know, smart, smart grids, um, smarter, uh, use of, uh, transport, encouraging transport decarbonisation technologies,
21:08water reuse, smart technology to actually, um, deal with leaks and water wastage.
21:14And the EU launched a climate, a water resilience strategy, I think it was last week.
21:18And this is a big part of it, is actually using technology to help us to save water.
21:22So, there's lots of areas I think we can explore.
21:24I could chime in too.
21:26I think the, the digital twin is a really good example of data visualization that, that we need in order
21:32to communicate the output from these massive amounts of data.
21:35Because reading a 400-page, uh, spreadsheet is no longer the way to interpret these data.
21:41And so, how do we pull, pull that out and illustrate the important parts to the, the different decision makers
21:49that we're trying to get that, that decision, those, those data to?
21:52I think it's such an important tool.
21:55If I could make another plug for, uh, a visit to the, uh, the Pacific Tech Village and have a
22:01look at the Tuvalu model.
22:03Because this is a real example of how countries are having to adapt.
22:08That, by 2050, Tuvalu is expected, at the best case scenario, to have 34 centimetres of sea level rise.
22:17Now, it doesn't sound like a lot. It's just longer than, you know, school kids ruler, right?
22:21But what that 34 centimetre does, of sea level rise does, is turns what was a one in a hundred
22:28year event to something that happens every year, right?
22:32For that country.
22:34And the one in a hundred year event impacts hospitals, schools, energy infrastructure, the, um, the farming land.
22:43And it completely, uh, and, and there's a huge challenge to overcome that.
22:47Imagine in your homes, if every year your home washed away and you had to rebuild it.
22:54How could you afford to do that, right?
22:57So, this information, this detailed modelling, which is demonstrated on the model outside, allowed Tuvalu to identify that they really
23:06needed to, to, uh, seriously adapt.
23:09And they've actually reclaimed a whole bunch of land by, uh, you know, um, pumping sand out of the, the
23:17sea and creating a new block of land that they're going to move a lot of that infrastructure to that
23:22will be higher than, um, than this flooding.
23:25So, this is a really active case going on right now to, to, to, to allow Tuvalu to survive as
23:31a country, uh, in what's going to be a truly existential challenge for them within the next 25 years.
23:38And what do you think, in that scenario, the most impactful innovation that could come along could be?
23:45Like, where's the gap for adaptation in that particular setting?
23:49Oh, well, look, it was a mix of things.
23:50It was getting the data to actually demonstrate the real impacts.
23:54This was using LIDAR and, uh, drone information and satellite information, detailed oceanographic modelling.
24:01But that told them what the problem was.
24:04Um, but getting the action to release the climate financing to allow the adaptation measure to be done was about
24:13convincing people of the need.
24:15And this is where the visualisation has been hugely powerful.
24:20We've taken it to the UN, to the COP, and it's, it makes it real for people.
24:25It's no longer numbers.
24:26They can see, this is the hospital.
24:29Okay.
24:29See the hospital?
24:30It's now underwater.
24:31Right?
24:31Um, so these scenarios are very real.
24:35And it's also important that the, uh, the community of Tuvalu are now, uh, they're owning this and presenting this,
24:43uh, themselves.
24:44Hearing from a presenter, um, that's showing you these scenarios.
24:49And they can say, my sister lives in this house.
24:51My brother lives in this house.
24:53And look, they're gone.
24:54Right?
24:55It makes it very real for decision makers.
24:57So I think the, uh, the combination of getting the data, doing the modelling, but also, uh, delivering the visualisation
25:05in ways that people can really understand is critical in getting people to, uh, engage with climate change.
25:12Yeah.
25:12And, and so just talk about adaptation to climate change.
25:15Climate change is, is disrupting things hugely.
25:19These are really great examples, but we don't have to look that far back to see what does a huge
25:24disruption look like with, with how the world responded to COVID when all of a sudden all their supply chains
25:30were, were disrupted.
25:33And, and even a little block in that supply chain and all of those really complex marine logistics.
25:40What do we do then?
25:42We absolutely need AI in order to understand how the supply chain is.
25:47What are the, what are the alternatives?
25:50Um, and then also in quantifying that the effect on biodiversity, because, because right now there's a lot of terrestrial
25:57information, but in terms of, of the high seas and quantifying biodiversity.
26:04And we can look at our scope three emissions, but we don't know what our scope three marine biodiversity impact.
26:11And we can talk about it qualitatively.
26:13So how much money have we put into saving the oceans, but we can't talk about it quantitatively.
26:20And so how do we measure that?
26:22Do we start with indicator species?
26:24Do we use indexes?
26:25Are the data that we're pulling from trustworthy?
26:28How are they verified?
26:30Is it transparent?
26:31Is it understandable?
26:32So all of those things, we tend to talk about AI as just the model.
26:38And so what are the model weights and what model are you using?
26:41But really we're all here.
26:42We're talking about AI tools and AI tools mean where did the data, where the training data come from?
26:48Where were they labeled?
26:49How is it using?
26:50What's the interface?
26:51Who's in charge of the data?
26:53Where are the data being stored?
26:54And then how is it being used?
26:56And then we don't want it to be just someone's PhD project that has something really good.
27:03It gets flashy in the news.
27:04And then it just, it doesn't go anywhere because it's not scalable.
27:08So we need to make sure that we have buy-in from industry, from research, from government to make sure
27:13that these systems are robust.
27:15And that all of the investment that we've put into creating these solutions, we have a long-term investment on.
27:24And we have a long-term return on too.
27:28Thomas, you were nodding along.
27:29Do you want to come in on that?
27:30Yeah.
27:31No, I just totally agree.
27:32It was the sign of my nodding.
27:33But yeah, I think not only unlocking the climate finance funding, it's also about unlocking the funding from companies.
27:41Because that transparency of what is going to happen to that island is equally relevant to what is going to
27:47happen to the communities that are working in my supply chain.
27:50And what's very surprising for me a lot of the time is so many companies have no idea where they
27:57are sourcing a lot of their products.
27:59And that's the EU deforestation legislation was meant to come in last year at the end of the year, which
28:06we thought would be transformational in terms of companies now tracking the impact of their supply chain.
28:10It looks like it's going to be delayed another year.
28:12So not coming in at the end of this year, but maybe even next year, because of this challenge of
28:17companies figuring out where everything is.
28:19And that's holding us back.
28:21And you're right.
28:21Once you can start to visualize it, we've got the data to be able to show and track the deforestation.
28:26Once companies know it, once consumers know where the products they are buying, whether they are contributing to deforestation and
28:33biodiversity loss, that's when I believe we can start to sort of get this flywheel.
28:37I spoke to someone recently who's building tech to oversee supply chains.
28:41And she was saying the tariffs stuff has meant businesses now want to know everything.
28:47And she thinks that she can kind of piggyback in sustainability metrics on the back of just this need for
28:53a forensic oversight of supply chains, which I thought was, yeah, a curious kind of ramification of this crazy situation.
29:02Before we run out of time, the other thing I wanted to bring in, Carl, was you were talking previously
29:06about using tech to combat climate misinformation.
29:11Do you want to talk about that side of things?
29:13Yeah, I think when we talk about all this data and having it available, I think that's a really important
29:18theme that's been brought up is making it understandable and making it trustworthy and traceable.
29:24So as we're telling a story in a way, we get the data and the facts across.
29:27So it's not someone being given an interpretation, but they actually have access to the verifiable source of that data.
29:34And that can also extend to product lifecycle and also the downstream impacts of that.
29:40I think also when we talk about artificial intelligence, that's a really important one as well.
29:45And making sure that the models that we use are open and that they can be assessed and that we
29:51know where the data is coming from.
29:53But also what we're seeing is, well, yeah, it's great to have an assistant that can help us to navigate
29:57knowledge.
29:57But we need to be able to provide people with the ability to go back and see what the source
30:02of that is, to go to where the source material is, to actually do a claims verification on that side
30:07of things.
30:08So this ability to have data open and accessible that you can go and actually see what it's based on,
30:14see where it came from, see where the knowledge that you're getting from the AI models is actually being traced
30:19back to, I think is really, really important.
30:21And something we're wrestling with a bit at the moment is, we see the rise in LLMs and Gemini and
30:28ChatGPT, etc.
30:29So what's it going to look like in three, five years' time? Are people still going to go to organisations
30:35like us to get their data?
30:36Or are they going to go to an interface and actually type in a query? And then how do we
30:41actually know what's going to happen with that data?
30:43How do we know it's going to be interpreted? So I think this is a really important point when we
30:47talk about democratisation of data,
30:49making sure it's available and accessible and knowledge, but it's also about being really responsible with the way that we
30:55actually provide evidence
30:57and the verifiable sources that people can go back to this really critical area, especially in the field that we
31:03work in.
31:05Stuart?
31:05Yeah, just to follow on with that one, because I think it's a really key point.
31:10I think there's, if I could use an example from the weather area, when you have an interface that's collecting
31:21all of this information and being the public face,
31:24you can make mistakes about not valuing where the data comes from in the first place.
31:31So this has been the experience with NOAA in the US, where there are many weather companies, the Weather Channel
31:37and other things that repackage that data and make it available.
31:41And sometimes politicians make the mistake of thinking that, well, why do we need NOAA? Because we've got the Weather
31:48Channel.
31:49The Weather Channel is just repackaging NOAA's data, right? So the fundamental data has to continue to be collected.
31:56And this is really critical for people to understand. I think it's, you know, really important that those fundamental data
32:06providers continue to be funded.
32:08Even if AI is packaging that and making it accessible, you need to recognise that the fundamental collection has to
32:15be there.
32:16And it has to be operational. And if I could use the example, what I mean by operational is actually
32:22the weather.
32:23So we know we can pick out our mobile phone any place in the world today and we can get
32:28a weather report that tells us it's going to be rainy or sunny, what the top temperature is going to
32:34be.
32:34And we know tomorrow it will be updated and the next day it will be updated.
32:38But businesses need that operational reliability with any service.
32:42If we just do one-offs and it's not clear when the next update is going to be or, you
32:49know, we're creating a one-off map of environmental factors and we're not producing a regular feed of that at
32:58a regular cadence that is known by the customers, then they won't change their business model to incorporate that information.
33:06So we've got to learn that lesson from the weather services and build that into our climate data, our environment
33:13data, and have operational services that just routinely deliver the same tools so that people can rely upon them and
33:21build it into their businesses.
33:24I wondered if we – so when we started the discussion, Emily, a point you made to me backstage was
33:30that there's a lot of – we're talking in the context of a wider discussion about tech's negative impact on
33:36the environment.
33:37Talk us through how you weigh up the benefits these AI tools can bring with the – because I know
33:43it's the kind of – it's the ever-present discussion, but I think it's important context to bring in when
33:48we're talking about the good they can do.
33:49How do we balance that with – yeah, absolutely, and there's another misconception that AI is just AI, and it's
33:57not all generative AI, so the types of AI that we use are – it's for analyzing data.
34:04And so we looked at the energy and GHG use for manually annotating all of the images that we go
34:12through, then those data are not standardized, it's very slow, or you can use our AI, and we're actually 98
34:19% less energy hungry.
34:23So there's – so it's a huge – it's a win-win in all the scenarios.
34:27So we're also based in Quebec, so we have hydropower, so I think people need to know where's the energy
34:34coming from, what's the tradeoff between, you know, doing it manually, using AI.
34:40I think AI is used for a lot of things that we maybe don't need to use AI for.
34:45We can – we're sort of using a sledgehammer when we need, you know, just a normal spoon maybe for
34:51something.
34:51But really understanding how crucial are the insights that we're getting from these data that are analyzed and how are
35:00they being used, and really understanding the impact of the data.
35:04It's not data just for data's sake.
35:06And so I think those are the conversations that we need to be having.
35:10And I want to come back to trustworthiness.
35:14And so absolutely the consumer needs to know, and then we need to know where are the data coming from.
35:21But the average consumer is not going to be able to backtrace, you know, how exactly, you know, which model
35:29do we use, and then what is the source data.
35:33I'm going back to imagery that that's the data I deal in.
35:37But I think there needs to be other metrics for trustworthiness.
35:42And so I know that there's a million different platforms out there or frameworks, but we're a signatory of the
35:50Montreal Declaration for Ethical AI.
35:52We're the first certified B Corp in the world using AI in service of wildlife.
35:57And so there needs to be – there need to be business metrics for your ethics and how robust and
36:04how honest and good you are.
36:06And I think there need to be those equivalencies for technology as well.
36:11And that we need to start seeing those signals from the large companies so that people know who to base
36:18their data on.
36:21Paul, does what Emily said about weighing up the impacts, the negative impacts, the energy consumption of tools like that
36:28use AI with the benefit we can bring, does that tally with how the EEA is thinking about it?
36:32Yeah, I think the words that you used, Emily, that really resonate are the trade-off, yeah.
36:37Because, I mean, this is a big discussion, and we're all coming from different regions of the world, and this
36:40is a cross-border issue.
36:42We're all representing organizations that work internationally.
36:45And I think these two sides of the coin, you know, digitalization can help us with sustainability, but we have
36:51to digitalize sustainably as well.
36:53And it's a really important balance that we actually strike there.
36:57Something interesting I've seen recently is, like, in terms of supply chains and textile use.
37:02We've actually peaked in terms of textile use in the EU, which really blew my mind, because I thought we
37:08were becoming a lot more conscious, et cetera.
37:10So, you know, AI, and I'm not linking these two directly, but it's interesting how, you know, technology may actually
37:16be changing consumer habits,
37:17and actually leading us to actually consume differently than we did before.
37:21In terms of technology, of course, we talk about AI data centers.
37:26So, of course, this needs more energy, water, minerals, et cetera.
37:31But everything we do, I mean, us sitting here today is also consuming energy.
37:35So it's not to say that we can't do anything, but it's also a major part of the solution.
37:39I mean, what's fantastic walking through the conference here is you see so many innovations,
37:44which are powered by digital and data, that are actually providing solutions,
37:48whether it's generating water, clean water from air, whether it's actually using technology to educate people in the solutions,
37:57smart grids, smart energy use, transport systems.
38:00So I think it's always going to be a trade-off.
38:04There's also, very briefly, that element of supply chain and making sure it's transparent,
38:07not only at the point of sale, but also the downstream effects.
38:11You know, recycling is part of the solution, but actually what happens to those materials afterwards?
38:16How do we promote a circular economy, which can also be tied into competitiveness?
38:20So I'd just like to finish on that, just to say, I think we have to look at it systemically,
38:23and it's not an either or, but just a really delicate balance we need to strike.
38:28Great. We can...
38:30Just a very quick comment.
38:31Yes, quick comment.
38:32An example, again, on the 750,000 images, Digital Earth Pacific is done in the cloud.
38:39That was able to be done for roughly 30,000 worth of cloud fees, all of that analysis.
38:45To do that at our office, we would have had to build a large data centre with cooling, heavy computation.
38:53We would have had to keep that running continuously.
38:57You know, the...and this is how we had to do it before the cloud came along.
39:02To collect that imagery, to collect that data in our vast continent,
39:07we'd have to go in boats burning fuel and, you know, dip things over the side,
39:12and the cost would be prohibitive, the energy use would be prohibitive.
39:16So you've just got to balance the two things.
39:18Yes, it's using a lot of compute resource, but it's amortising that across many uses
39:24and making that more efficient through using cloud resources than it ever was before.
39:30Thank you. That's a really good, tangible example to end on.
39:33We are out of time, sadly, because it's a massive topic.
39:36There's loads more to talk about. But thank you. That was really interesting.
39:39Thank you very much. Thanks very much for having us.
39:41Thank you. Cheers.
Commentaires