Making Sense of Sustainability Data With Blockchain and AI

Transcript

So, we have quite a diverse panel for you as well. It's a mix of a variety of topics. I'm really glad that we have some experts here that are going to tell you all about sustainability, AI, and blockchain. There's a lot of mix, but we have some great experts that I hope in 40 minutes are going to be able to crunch all these numbers for you and uh yeah, we'll take it uh from there. So, Mik, I think you're our moderator, so I'll let you bring the panel together. Let's give them a round of applause. Thank you. Thanks everybody for for being here. We will do you know a bit of energy. We have our first panelist Alvaro on the second chair and then Ivon on the third one Daniel on

the yeah second to last Roman. Thank you. So I'm Mikuel. I'm the chair of the sustainability working group at crypto valley association and I wanted to put together you know a panel at the intersection of blockchain AI and sustainability taking the perspective that on the one hand there's always an increasing need you know particularly when it comes from an European perspective of sustainability data for sustainability reg uh reporting regulations uh green green green directive uh nature restoration law. So an increasing number

of regulations which are really sustainability data intensive. On the other hand also taking the perspective of understanding and being aware that there is a significant business value also around sustainability data. So yeah, for this reason uh we came we put together this nice uh range of difference experiences. So I'll be happy for you to maybe starting with Alvaro just respecting the order to quickly introduce yourself and uh from this perspective starting from the issue perspective I mean uh what are the biggest challenges and difficulties

when it comes to data independently you know from the blockchain and the sustainability perspective in your respective industry. All right, great. So, I'm Al. So, I I'm senior innovation lead at what we call Switzerland for HR. Switzerland HR. Think of it as a kind of a joint venture um which is a Swiss foundation part of the UN refugee agency. But this status actually allows us to test things and uh and then that will be later scaled up. So in terms of data uh UNHR is a big provider of data and um collector of data. our perspective as Switzerland for HR is really to connect with the private sector in Switzerland and working on innovation on that. So our our

perspective is a lot into philanthropy uh a lot into like uh really how to find fundraising solutions for today and for us this data that is available for us today is at least on the projects I'm working on is uh is we see a lot of potential in the gathering data for everything that has to do with traceability reporting that uh if you think about it the way we do reporting today hasn't changed much since the 70s and it's very um I not really costefficient. So, we're really excited in in in seeing how the you know this data collection and how we can use this data to actually improve and uh uh become like a kind of a gateway to make uh the way we work more efficient.

Yeah, about your side Ian. So, I'm Ian, director of research at Definity, the foundation that is the main contributor to the internet computer. And for those of you that are not familiar with the internet computer, it's a decentralized uh cloud platform and actually at 3:30 there will be an exciting demo. Um if you want to see more of what is possible, what kind of compute you can do on the internet computer come then um with respect to um data what uh I see at Definity but also what I've seen in uh previous roles uh in industrial research. I think it's super important

to be in control and to know exactly who can access what kind of data, who can do what to your data. So it's not just them lying around. We want to do something with it. So who can do what? Um and uh yeah, how to do this uh efficiently and in a privacy preserving uh way is dear to my heart. Yeah. Daniel, what's your take on this? myself. I'm Danielle uh Jo uh CEO of NHSH. Uh we are an energy tech company.

So I'm coming from a little bit different uh part of the data management. We are working with a lot of large large scale utilities and oil and gas companies to um implement uh different uh energy solutions. We develop flexible energy systems, computing infrastructure and um basically flare gas recovery systems uh for uh oil and gas companies. My part at the data is that we work directly with the largest emitters of the world and we see how much data they actually don't have. So uh when they don't have the data you know there there is no problem.

So we are working on it like how to reduce the notexisting emissions let's say and uh and make it accountable and and how to you know scale it on a larger scale. Mhm. Thank you. Good afternoon everybody. Hi uh Roman Castillo. So I've been working around 20 years in the the financial industries in various roles and most recently I'm a director at Swiss sustainable finance which is a large association in Switz in Switzerland trying to help the financial services sector transitioning into more sustainable practices so sustainable financing or sustainable investing more specifically. And when it comes to your question uh Mik about about data and the

challenges that we see my answer will not relate that much to my own organization because it's a small organization on the data challenges are pretty simple but more to what we see in the financial services industry are the typical issues of market participants and here I think there are like three or four things um and especially in relation to sustainability data sustainable finance data the first thing is in that industry you have more and more reporting requirements. So you are supposed more and more to report higher volumes of data which are supposed to be more granular your scope one scope two scope three etc. So there's a challenge in finding the data and providing it to

your to your clients and your stakeholders. In order to do that you need to to get the data you have a dependency on what people provide you. And here there's a lot of issues on availability of data. So that's basically echoing what you were saying. uh very often you just don't find the data you need to forward to your own stakeholders or investors. There is a big issue in availability of data. There's a big issue in the quality of data because sometimes you have data but it's basically a a rough estimate on one or two years down the road once you have refined your assumptions you realize suddenly that the 200 you had reported two years before actually it was like 120. It's too bad you have to restate it. So you have the availability issue, you have the quality issue, you have the

sourcing issue because you have lots of different sources of data which are heterogeneous and very often the reports and the sources you use do not use the same format. They are inconsistent with each other. So you cannot really compare you need to you know rebrand restate the whole stuff that's mostly external challenges and the last challenge I have in mind is an internal one which is I think specific to large FS organizations which is basically that your own systems are very fragmented. You have lots of databases and very often the data you need is scattered in different databases that don't connect to each other. That's something I've seen quite often. That's specifically the the typical changes I can see and I can guess you you guys have seen something similar. Yeah, that's interesting. So to see somehow

some teams which are recurring, you know, availability of data, quality of data, efficiency in obtaining data and management of data, you know, who can access it, you know, the different layers, privacy layers. So I mean now let's get somehow to the meat of the discussion you know so from your perspective starting from Alvaro what what's like your take on uh you know blockchain and and AI to be helpful in you know overcoming these difficulties we've just mentioned so with blockchain and AI I mean as as I was mentioning we're a very big provider

of data data that we use for example placement data uh data about like uh the way we manage for example refugee camps the number every actually in June every year um all the media is actually waiting for the official like statement of how many refugees you have around the world so it's really around June 20th that you get like this global report on the number of internally displaced number of people were actually displaced as a refugee around the world so we we are a big provider of data the thing that that happens is that and in in our role of connecting with the private sector, finding philanthropy solution, financial, you know, like impact investment is that how do we uh get this data to match uh to be good enough uh to

be used uh in financial and other things. So I would say that blockchain and AI in two different ways helps us uh focus on the right data and how do we bring it. So we're developing solutions like okay predictive analysis of where for example help will be needed the most uh in some areas of the world to have like for example an emergency preparedness fund to be set up to know that something you know cool happened there also for example uh accompanying our donors like really this predictive do I mean they've been funding this they'll be interested you know with this by this as well you know on the blockchain uh for us and that's really We started with blockchain actually

before actually AI for blockchain for us is really is this transparency. A lot of donors are asking for transparency. They're asking okay how do you spend my money? How do you spend the funds? I know I don't have to remind you the latest news all around. They asking for more and more scrutiny more transparency from everywhere. And even if uh not all countries and not all private sector uh um actors are asking for that kind of transparency, blockchain is a good way and efficient way uh in some cases to provide this and at a costefficient way because you have to imagine if sometimes for some donations of some funds we receive if we uh spend like 30% of the total amount in reporting what we did with the funds that doesn't really make sense. So yeah and I'm curious so from

that perspective I guess you need to work on a lot of assumptions. Is it better to have bad data or no data at all? I don't know whether that's maybe a recurring question on your side for you or Yeah. Yeah. Yeah. Yeah. Follow up on you if that's a that's a that's a that's a really good question actually. wrong data definitely not I mean not enough data let's see again if technology today helps us to use it but I think like even like for questions of I mean engagement which is a very big part of uh what we're trying to do at least from Switzerland you know in the global north

with audiences in the global north even like some data we can use it and make things out of it I think of uh for example we did a a fundraiser with Christies last here and we used some of the asylum seekers like data from the biggest I mean the biggest economy in the world and we did even an artwork with hands opening and closing depending on the percentage of asylum requests that were accepted and if the uh percentage was increasing the hand will open and then it will actually close down. That was a very successful example of how you can actually use data in a different way. But I would say that uh more than no data a little I mean more I mean I would say that wrong data is what

we don't want even a little bit is we can always find ways to use it. Yeah. Yeah. Yeah. Can I jump in here maybe Mike because your your question I I I like your question because it made me think of a French joke which I'm going to try on translate which is um do you prefer to have a 10 meter long arm or a leg made out of foam? Right. And basically you don't want neither nor right so do you prefer to have bad data on the data it's like it's an impossible choice my guess or what I've seen in the industry in most cases is there are specific points you have to report it's like required to report data in these cases if you don't have an exact data you need to make your best estimate it's

like you know your best will basically and at some point you have to restate it you have no choice you just do that in other cases where you don't have a best estimate available I would tend to think that it's better not to report something than to report something that you know almost with certainty is probably BS. So that's probably where I would draw the line, you know. Yeah. Yeah. Yeah. And from a data perspective then I was wondering when it comes to assumption if you in many chance in many occasions you slightly change some parameters and you get a huge range. How do you deal with that? So when it come for example you know to climate measures just a

small changes in assumptions in the model can dramatically change you know the the yeah the range basically of data. So I was wondering how would you deal with that um it's it's a tough one but I can sorry say is so there's no right solution. What I what what I see is this modeling is being refined over time and for example if you take the climate modeling as an example you know what what we look at in our industry the financial services industry is um when you set up investment products or when you um package specific portfolios to which scenario to which warming scenario do you align so you use the well-known

1.5 scenario the two degree scenario or other scenario right and when you look at the data over the years It's quite interesting, maybe not positive, but interesting to see that the trend is not to have more and more responses in the 1.5 bucket. And that's because people might have the intention to align to 1.5 target, but over the years they simply realize when the data gets more accurate, more complete, more fine that actually it's not as positive as I thought. So you see a small trend going from 1.5 to two not because people per se want less commitment but simply because the data they have gives them a different evidence. So I think there is no way around it that actually you refine your modeling over time and maybe

it's going to tell you something that's less optimistic that you thought at first and that's what I'm see I'm seeing happening in the past five six years. I see I was mentioning at the beginning part that they actually don't have data. So we work with quite a few oil companies uh mostly governmental loaned state oil companies in order to reduce the emission. And when we get to a field you know you can see this huge flare stacks burning there uh which is like one of the largest emitters uh basically globally it's more than 500 million tons of CO2 emissions uh globally on a yearly base. And when we asked them, okay, how much do you flare? And they were like, we don't know. We we we, you know, like

it seems like a lot. And you know, 30,000 cubic meter a day. And when you actually make measurements, sometimes it's 200,000 instead of 30,000. And and then you ask like, okay, but what's the composition? How you measure it? And how you declare your emissions is self declaration, you know, like but they don't even know the data how much they're emitting. And then they come like then the ministry ask how much is the emission? Uh it's five. Okay, who cares? Like because nobody's checking there's no measurement on the fields. Nobody goes there. Nobody's measuring. There were a couple of U systems now because you can actually measure it if you want to but nobody wants it because once you measure

it then you are accountable for it. So if you measure it then they can show like this is your emission, this is how much you reduce. If you're not doing anything then it's your fault. And you know it turned out sometimes the emissions were 16 times higher compared to what they declared but on average it's two three times higher when they actually measure it from helicopters and so on and they can actually measure the meta emission. So when we come to the data in order to create the data there is a big lack of like real information and measurements of the systems like how much is the real emissions of of productions for example but I think there will also be a change and with the regulation that is coming there will be

also more pressure to actually measure more in certain areas we see that already. So for example um cities and governments are trying to be um more accountable at least in some parts of the world and also there are a lot of uh creative ways of uh measuring some data. So I've just um yesterday heard from a startup that uh builds on the internet computer that uh they are working with um cities that collect um CO2 uh values around the city with buses. So the buses drive around anyways. So they equip them with the measurement stations and measure where in the city um there are yeah different

microclimates and uh different emissions. So um I I think um the this uh this challenge that that we are facing it can also be uh um yeah can with with creative um solutions um maybe we we can tackle at least some part of it. I really believe in it but uh yeah it's going to be a tough one. It's it's really tough one because when we are dealing with these companies who are the regulators in the given country is the owners of the oil companies who are like running the country in regulate and and then comes to like why you penalize yourself and you are producing and you have the emissions and you don't

want to declare and then it's like a circle and and all the oil companies and 54 governments signed all the uh zero routine flaring decree for example there is some regulation that until 2030 they need to stop to zero. Last year it increased 7%. So in 15 years they didn't decrease anything and they don't even know the data yet. So they have a couple of years left. I hope I really hope because that would be a high pressure on them and they would really need to make something to stop uh you know emission. So if there is some regulation that's always really good because maybe they will actually want to do something then but I really hope so. That being said, I mean, yeah, we are still looking quite a

lot into the issue part. I would be interested to hear like the solution part, the blockchain, the AI part. I I you see that can somehow help you know for example I was connecting Daniel to the fact you mentioned before of people actually probably not being willing to put data in the blockchain because once it's there you know you are starting to become transparent and maybe you don't want to be transparent but then what what are the use cases I mean not just for you Daniel for whomever where actually blockchain and AI is already helping from this perspective I can try to jump in on on that one. Um I'm going to focus more on the AI side

because that's what I'm working on on on the use cases I've seen in the industry. First thing to be really clear, we've been discussing here a few issues right around availability quality etc. Clearly when there is no data AI or blockchain is not going to invent it for you. So if it's not there it's not there. So there won't be a solution to that issue of availability except once it's being produced. But when we take the next step where we say okay we have some data right now we need to work on it and we need to assess it we need to make use of it for our estimate scenarios computation I think the best use I I've seen the industry maybe I take a step back we are at SSF working on a big publication on AI on sustainable finance

and among others we have been surveying the whole market the whole financial services industry on what the use cases are so what I'm trying to describe here is a bit of a summary of what we observe across the market. It's not one specific company. It's more an industry thing. And what I've seen are solutions around extraction relatively obviously, right? So you can automate extraction and make it much more efficient. That's relatively useful in the world where you have to deal with always higher volumes of data, right? So it's not assessing the quality, but it's managing the workload that you have to extract your data from the sources and put in your own report. So here you have more and more solutions being used by the industry to extract. The second thing and that's a super important one. I

think that's even the the most important one in my view. Any AI system or any report will only be as good as the data you feed to it. And as we were saying earlier, we have lots of issues in the quality of that data or the comp comparability or the consistency of that data. And there are a few solutions out there, not many but a few solutions AI based which do plausibility checks on the data you extract from various sources. Um by doing these plausibility checks saying for example okay um we have uh these emissions uh data from that company. If I know I compare this data to the previous years of the same company or I compare to other companies in the same sector or to other companies

with the same number of employees or same capex or same industry or same same country is it plausible right by doing that you can feed a huge number of of data into a plausibility check and you can select the data source which seems the most reliable one. So again not helping with availability but helping with cleaning the quality and then of obviously in the later stage and I'm going to stop there otherwise I'm too long but in the later stage we have then the next step which is about getting insights out of it and putting them in reports and maybe we'll have time to talk about it but when it comes about the issues we are speaking about earlier you know on the quality etc. I think that's the two main things I see dealing with the volumes and then doing quality checks making sure then in your own

database you have stuff you can really use and you can start building on that for your own reports and your own deliverables. I would also like to report on on some examples. So I and I fully agree with the problem if the data that you base your computation on doesn't make sense then uh good luck with the result. So but yeah there I think AI can can really help with plausibility checks. So um I would like to mention two examples one more from the environmental side and the other one um more on yeah economic or societal. So on the um environmental side um in addition to the the bus example there or there's also um a

startup that um um builds a charging infrastructure um that is powered uh by blockchain um driven incentives. So everything is super transparent. you um can earn um rewards for um offering charging um possibilities and um also being uh involved in the the revenue that is being generated. So it's really it really helps to um move from towards more um e-mobility which I think as a whole while we we on on this planet we are going in this direction but there's still quite a long way. So I I think having the right incentives can

definitely um help and u making sure that the yeah the the the batteries are being used uh in the in an efficient way. Um so that's one use case and then the other one um is about um um small and microloans in developing countries. So there um we are actually working with UNDP um on a on a project that uh will make it much easier for um small enterprises or individuals to get loans by streamlining um processes involving um blockchain technology that will make sure that things are more transparent that there's less um possibility for

corruption and that also makes things much more efficient and um I think there are a ton of use cases out there where we can really benefit um from blockchain but also from uh AI to um yeah make this world a better place. Mhm. Yeah. Yeah. Yeah. Yeah. I like the part of rewards you know because I believe from that perspective blockchain infrastructure can be really helpful in creating a reward system you know with tokconomics and similar elements. Therefore, I was wondering whether you have some examples or some comments on the question of somehow getting people rewarded to assure the

quality of the data or to somehow create data or if you find any interesting connection feel free to jump on this. We're actually working uh if I may say you were talking about UNDP sister agency um we actually for us data and uh I mean blockchain through blockchain and AI is an enabler. So we really opening like new ways of of working being more efficient or new ways of funding in our case the first like straightforward first solution that came out with blockchain for example was payments. I mean we're talking about crisis situations. tell you send money somewhere in the middle of a crisis of war. So when uh the Ukraine invasion uh took place uh we uh partner up with a

with a blockchain uh I mean Stellar and USDC and to be able to uh for refugees to cash out uh if they were in Poland in zlotties any monograms or in Germany in euros. So this kind of data instead of shipping because we still like for example nowadays in the humanitarian sector you still have like planes flying in and out with pallets full of cash which in terms of security can can be questionable right so this kind of data is super important for us to see for example if everyone cashes out everything on the same spot well something is weird right so you can actually investigate what's happening in that area for example but then we're also working on offsetting like impact certificates to show again this

transparency And last but not least, we're working also on a on a project of uh changing the way people do fundraising as well, like donating uh by incentivizing donation and actions and rewarding them with tokens, but they don't know it's tokens. They just get it's like a loyalty card. Yeah. Yeah. Many times that's the solution which is somehow suggested to the business industry in the perspective of you know using web three but without even you know telling the user that is using web three and uh I'm a I'm a bit aware of time if you just have maybe we we can take a couple of questions but if each of you can conclude you know with a thought you

know or a burning element they wanted to raise before you know closing the panel you are the Of course it's on me. That's a tough one. Um I need to say something smart. Uh I think um two things first because we are here at an event where we're talking about AI on blockchain, right? So my first thought would be that the typical interaction I can imagine between the two technologies would be leveraging on the first one which um main benefit is around safe secure non-penetrable storage I would say on the second one about driving on providing insights out of large data

sets and here if I try to put that together and link it to the topic of sustainability which is sustainability on sustainability data I think there are a couple of use cases around the setup of specific products like sustainability linking clones which could leverage you know AI to monitor the progress of certain conditions or then could feed for example a smart contract to then execute on the conditions on the loan and provide the discount the discount payment to the to the to the to the borrower. So that I think there are a couple of use cases here which uh which could maybe emerge over time under a number of conditions on maybe more broadly I think when it comes to new

technologies and now I'm speaking specifically about the industry I know the financial services industry right that that's what I know most um I think there's still a lot of way ahead right uh most of these new technologies are used still in very early stage AI for example is used mostly for natural language processing, a bit of sentiment analysis and reports. There's a lot of stuff that can still be done when it comes to assessing and monitoring client preferences, client interactions, designing products by mixing genai on predi.

So I think we are really just at the beginning on that days under conditions on using the right data etc. A huge potential for these technologies to be used much more massively in the next few years. That would be my forwardlooking view on on my end. Any other concluding point or let's go step by step. So on my part, I really hope that there will be more regulation in the near future because that would really help on the quality of the data that they would actually make some measurement on the emissions and that would help us really develop a lot more because when there is not a pain they don't really want to do anything like in real cases. So but if if you have data there is a pain because then you can see

the the actual data and the more blockchain there is in the system because we operate a lot of blockchain and AI uh infrastructure but hardware infrastructure that can be utilized also for like basically um mitigating the emissions and reduce the wasted resources and connect to the to where it it's it's not used. So I hope we can run a lot more data uh basically on the infrastructure in the future because that will be really helpful for um every industry basically. Yeah.

And I I would also like to to say a few words uh regarding it's not just blockchain and AI. I think we can also use blockchain for AI because the transparency that uh we've heard uh mentioned quite a few times today. I think also for AI this will be super important um to that we can really trust that the things that the the AI tells us is the truth is really um computed correctly. So I hope that the blockchain transparency benefits can be used on AI as well.

Yeah, as a as a note I would say um that yes in a in our case in a you know humanitarian sector I mean technology in general new technologies can is disruptive can be scary um giving given the situation we can't not afford not to explore new ways but one thing that uh we often see is that we trying by trying to do too many things actually you were talking about pain points we've got only pain points we only have challenges So, so and actually that's our role also to look out there in the private sector in academia to see okay what solutions already exist and how can we adapt it to solve the pain points that we have and uh yeah if you want more pain points

just come to see me yeah interesting one I don't know if we have any questions from the audience please thank you very much foring So you know whatever gets in in terms of data considering the current

um investment and standardizing. So what's the the question exactly? How do we see what in relation to the collection of system related data the evolution or okay whether it's worth it you mean or Um okay I I will give it a try and maybe my my colleagues can can jump in. Um there's a lot of talk about you know the ESG backlash on the ESG fatigue on you cannot open your newspaper every day

without having the the face of Donald Trump everywhere and of course you think of baby drill baby drill on all sustainability is dead but basically very interestingly when you look at the development of the um requirements on sustainability reporting across the world they actually tighten absolutely everywhere US is the exception it's the huge market I don't deny that it's a topic but I I think it to be probably um a misperception due to large media coverage to think that it's something that's happening everywhere and that it's just going backward everywhere.

There is a generic trend and that will I thinking continue with bumps on the road to get that transparency simply we because we need it not because people really want it because it's constraining and it's also economically constraining because you need it simply to maintain your economy over the longer term. So over time these requirements to get more data to get more transparency and to overcome these challenges that we've discussed on availability they will be there. So investing in solutions to get that data to process it and to deal with that is worth it because despite the impression we might have right now it's not going to get away anytime soon. It's just going to have maybe a few hiccups but it's going to continue going in that

direction. So investment is definitely worth it. We're not going to go away with that. That's my personal view. I also think standardization will help um the economies in general because the then they won't have to reinvent the wheel but can really communicate and exchange data with each other. So I think pushing into pushing for standardization um also is something that we should definitely do and not just hoping that some big leaders somewhere will have the good solutions but all of us who we vote for what the kind of things we buy this has an influence. It's not just the big um government somewhere. So we should not forget that.

Yeah, I like it as I believe as somehow a concluding message and uh I mean before we thank the our beautiful panelist I wanted to let you know at the sustainability working group we are pretty much interested on questions around decentralized AI how the two can work together. So reach out if you're interested in doing something together and prepare your hands for a big round of applause for our panelists.