Latest / Second Nature / E194 New York Climate Week Live: The Uncomfortable Truth About AI, Climate Change & Nature

Transcript

E194 New York Climate Week Live: The Uncomfortable Truth About AI, Climate Change & Nature · Second Nature

• 0:06
  SPEAKER_03: Good morning, all.
• 0:08
  SPEAKER_03: Welcome to the 1970s.
• 0:10
  SPEAKER_03: What a fantastic room, right?
• 0:12
  SPEAKER_03: Just unbelievable.
• 0:13
  SPEAKER_03: We did um four or five podcasts in New York Climate Week last year, and it was me, it was incredible guests in a very small, very warm room.
• 0:23
  SPEAKER_03: And it just felt selfish and unfair.
• 0:26
  SPEAKER_03: There are incredible conversations that people got to listen to afterwards.
• 0:29
  SPEAKER_03: And we just thought, there's an opportunity now.
• 0:31
  SPEAKER_03: Why wouldn't we bring a group of people together, very intimate setting, and let you be part of the discussion, the conversation, throw in your questions, interject from the floor if you would like to.
• 0:41
  SPEAKER_03: We did the same thing in New York with a much bigger auditorium, but got such a range of questions in there.
• 0:46
  SPEAKER_03: It was wonderful.
• 0:47
  SPEAKER_03: So we want you to be involved, we want you to ask things, debate, discuss, be part of the conversation.
• 0:52
  SPEAKER_03: Um so we are going to start as we mean to go on.
• 0:55
  SPEAKER_03: Um I'm gonna be pretty direct, we're gonna move at pace, and I'm gonna let the panelists tell you their stories.
• 1:01
  SPEAKER_03: Um keep coming in if you wish.
• 1:03
  SPEAKER_03: Guests keep coming through.
• 1:04
  SPEAKER_03: Um the panelists will explain who they are, what they do, and their take on AI, on technology, on data, on the world around us, and on capital.
• 1:13
  SPEAKER_03: But my opening question is is AI, and I want a direct yes-no, um, or better worse, um, will AI be positive or negative for nature?
• 1:25
  unknown: Yes.
• 1:26
  SPEAKER_03: Excellent.
• 1:26
  SPEAKER_03: Anna?
• 1:27
  SPEAKER_00: That'll be what we do.
• 1:28
  SPEAKER_03: Ooh.
• 1:29
  SPEAKER_03: Sheila?
• 1:30
  SPEAKER_00: Yes.
• 1:31
  SPEAKER_03: Excellent.
• 1:32
  SPEAKER_03: Okay, what I would love is for each of you to give a 90-second intro, say who you are, what you do, and what you're most excited about in terms of technology right now as we sit here today.
• 1:43
  SPEAKER_03: So if we start with Dan at the end, yeah.
• 1:44
  SPEAKER_05: Uh Dana Margison, CEO of Art and AI.
• 1:47
  SPEAKER_05: So we're market intelligence for the land markets.
• 1:50
  SPEAKER_05: And land markets are obviously very dynamic right now.
• 1:52
  SPEAKER_05: You have all of this data center build-out, infrastructure build-out, commercial industrial, but it's also where we eat, it's where we live.
• 1:58
  SPEAKER_05: And so it's really interesting asset class to be part of.
• 2:01
  SPEAKER_05: And basically what we do is we take a bunch of geospatial data, market data, we track developer activity and other things of that nature and try and identify where there's interesting things happen and help investors make good allocation decisions and also identify for landowners, especially large institutional landowners, what should you do with your land based on the core characteristics of it?
• 2:19
  SPEAKER_05: Should it continue to be farmland?
• 2:20
  SPEAKER_05: Should it be a data center?
• 2:21
  SPEAKER_05: Should it be conservation land?
• 2:23
  SPEAKER_05: And so that's fundamentally what we do.
• 2:26
  unknown: Thank you.
• 2:39
  SPEAKER_01: What does this mean in this context?
• 2:41
  SPEAKER_01: Well, we do a couple of things.
• 2:42
  SPEAKER_01: We try to work with leading organizations and experts to direct AI and AI solutions towards some of the sticky problems or the opportunities we can see in the space.
• 2:53
  SPEAKER_01: Then we build AI capabilities, AI fluency we call it, amongst frontline organizations, climate nonprofits, and other institutions that regulate these markets to make sure that they actually understand what AI is and could do for nature and climate.
• 3:06
  SPEAKER_01: And then lastly, we try to convene this growing community and make sure that we direct uh the direction of travel for the technology.
• 3:14
  SPEAKER_01: So I guess kind of in weaving together all these actors and trying to kind of make sense and push us towards a shared positive journey.
• 3:23
  SPEAKER_00: Sheila?
• 3:24
  SPEAKER_00: Yes, so my name is Sheila Oviedo.
• 3:26
  SPEAKER_00: I head up the um the product for the climate and nature products at S ⁇ P Global.
• 3:33
  SPEAKER_00: So in the market, you probably know SP Global as the ratings agency, but SP Global is a big provider of data, um, the financial and non-financial, and that includes climate and nature.
• 3:45
  SPEAKER_00: So my lens, I come here with the lens of product and technology because I own the products, um, the delivery channels of all of this physical risk products, the nature products that go to mostly financial services clients.
• 3:59
  SPEAKER_03: So the first bit of audience participation for the morning.
• 4:03
  SPEAKER_03: So I'm overwhelmingly positive about AI.
• 4:07
  SPEAKER_03: I think it is the answer to resolving climate change and our problems with nature in in every which way.
• 4:12
  SPEAKER_03: I would like a show of hands first.
• 4:14
  SPEAKER_03: Who is positive about AI and its impact on the world in general, please?
• 4:18
  SPEAKER_03: Show of hands from the audience?
• 4:21
  SPEAKER_03: Okay, you can vote yes on both questions here.
• 4:23
  SPEAKER_03: Who is also worried, concerned, or has some big questions about the impact AI will have on the planet?
• 4:29
  SPEAKER_03: Show of hands, please.
• 4:31
  SPEAKER_03: Interesting.
• 4:33
  SPEAKER_03: Um What are the things that you are most excited about and optimistic about from AI and tech at the moment?
• 4:39
  SPEAKER_03: And we'll let's make this circular.
• 4:41
  SPEAKER_03: Whoever wants to go first, jump at it.
• 4:44
  SPEAKER_01: Um I can jump in also just to reflect on this audience assessment.
• 4:47
  SPEAKER_01: Like this is what we see all the time when we work with organizations, like sustainability teams and businesses or frontline uh climate conservation organizations or or community orgs around the world.
• 4:58
  SPEAKER_01: Like there's this excitement, this sense of hope, and you know, this is a possibility and also like real concern.
• 5:04
  SPEAKER_01: And I think, I don't know, from from our programming, we always think that taking the concern seriously and actually like pulling it apart and looking at data and trying to figure out what's going on and look at the systemic issues, that helps to move the conversation to kind of a constructive and productive one.
• 5:19
  SPEAKER_01: Um yeah, so just kind of a reflection on like this is what I see all the time, and this is really great.
• 5:24
  SPEAKER_03: Yeah, both hands was the answer I largely anticipated getting.
• 5:27
  SPEAKER_03: And I think it's a realistic one for the world.
• 5:29
  SPEAKER_03: Like we've never seen anything like this before in terms of pace of change.
• 5:33
  SPEAKER_03: All the positives, things that we're optimistic about.
• 5:35
  SPEAKER_00: Yeah, I think I'm very optimistic about AI as an accelerator.
• 5:40
  SPEAKER_00: In my world, I deal with climate science, climate scientists.
• 5:44
  SPEAKER_00: And what we could not do five years ago when we started, we couldn't do today.
• 5:49
  SPEAKER_00: I mean, granted, we have to work with hyperscalers, we work with you know, frontier model um providers.
• 5:56
  SPEAKER_00: But what was very, very expensive or almost um unattainable for even for a large company in like five years ago is very much attainable or very much possible today, because you have AI help uh accelerating um climate science, for example, or climate forecasting, weather forecasting.
• 6:18
  SPEAKER_00: That's just the I think that measure the the measurement of exposure is really accelerated by AI.
• 6:26
  SPEAKER_00: Now, to Anna's point, to the audience's point, there is it it's not yet, I think we haven't reached the point where it makes decisions, it makes the the right decisions, it makes the right decisions and interventions.
• 6:40
  SPEAKER_00: I think we're not yet there yet.
• 6:41
  SPEAKER_00: But in terms of accelerating the science, accelerating what we could like the modeling, that's very, very exciting.
• 6:49
  SPEAKER_00: And that's what, you know, in my world, that's what makes me very excited about AI.
• 6:54
  SPEAKER_05: Yeah, I mean, there's a lot of scary things about AI, the pace of change, it's true.
• 6:59
  SPEAKER_05: I mean, impact on jobs, but also there's gonna be this enormous industrial build out.
• 7:03
  SPEAKER_05: We're already seeing it.
• 7:04
  SPEAKER_05: And that build-out won't just stop with data centers and energy.
• 7:07
  SPEAKER_05: It'll go into robotics and all kinds of other things to support large GDP growth if these projections are true, which you know, I think Jerry's still out on whether AI is really gonna bring the productivity that we hope it will.
• 7:18
  SPEAKER_05: Um, so there's lots to be scared of.
• 7:19
  SPEAKER_05: I think on the more positive, optimistic side, you have to fundamentally believe that humans make the right decisions for nature and for the world when given the right information.
• 7:28
  SPEAKER_05: And one of the big challenges with development so far, which is something we think a lot about, is that you're myopically kind of looking at the returns from a specific project, but those returns often almost by the nature of the data you have, can't include biodiversity, water, other things that many in this room care about.
• 7:44
  SPEAKER_05: And so by being able to surface that as part of any decision-making process and sort of building it into the core operating model, uh, you hopefully will get much more positive outcomes uh overall.
• 7:54
  SPEAKER_05: And that's something that I think AI has completely unlocked, um, both in terms of how we process that information and build it into return models, but also our ability to collect that information at a scale that we've never seen before.
• 8:05
  SPEAKER_05: So I think that's the exciting part of it.
• 8:09
  SPEAKER_01: I just wanted to add also, because you mentioned kind of the people.
• 8:12
  SPEAKER_01: I what we are seeing is also small, very small organizations are all of a sudden able to do a lot more without having to expand headcount and and define new resources.
• 8:22
  SPEAKER_01: So this idea of like doing more with less, let's make sure the more is the right more, but but doing more with less is quite inspiring, particularly from a field that's like, you know, overworked, stretched, low budgets, whatever.
• 8:32
  SPEAKER_01: So I think that I'm like, I'm really excited about that part.
• 8:34
  SPEAKER_03: Often when I've sat in audiences of these types of sessions, I've I've heard people talk at a high level about how great AI is and things.
• 8:40
  SPEAKER_03: And I've gone, yeah, but give me the real example.
• 8:43
  SPEAKER_03: What are you doing to actually find productivity?
• 8:45
  SPEAKER_03: How are you using it on a day-to-day basis?
• 8:47
  SPEAKER_03: So that's the thread I'd like to pull next.
• 8:49
  SPEAKER_03: Can you give us some very specific things that you're actually doing that might be actionable for the audience to take away?
• 8:54
  SPEAKER_03: And then I'm going to ask you guys as well.
• 8:56
  SPEAKER_05: I I have a very clear example with our clients.
• 8:59
  SPEAKER_05: So we work at a lot of large uh industrial foresters.
• 9:02
  SPEAKER_05: A lot of the privately held land in the US is forest land.
• 9:05
  SPEAKER_05: And all of that is under enormous pressure right now because the timber markets are down and they have been down for a number of years.
• 9:11
  SPEAKER_05: And so many folks are struggling to stay in business to raise capital.
• 9:15
  SPEAKER_05: Uh, the way that they stay in business and are able to purchase more forests, which generally drive really positive outcomes, right?
• 9:21
  SPEAKER_05: You want these forests to be managed and that allows people to recreate in them.
• 9:25
  SPEAKER_05: Obviously, wood products are really great to build with relative to other building materials.
• 9:28
  SPEAKER_05: It's lots of positive to having working forests.
• 9:30
  SPEAKER_05: Sometimes the only way they can make those returns work is if they can take some chunk of that large forest land, say 200,000 acres, and use 10 of those acres for a data center or for a couple of solar projects or maybe some housing development.
• 9:42
  SPEAKER_05: And so being able to allow people to make those decisions allows us to continue to protect essentially, you know, hundreds of thousands of acres of forests in a way that that we wouldn't before.
• 9:52
  SPEAKER_05: And I think AI and our ability to just look across all these different use cases is is fundamentally changing the investment case for uh forest landowners.
• 10:00
  SPEAKER_05: Love it.
• 10:03
  SPEAKER_00: Yeah, so two things.
• 10:04
  SPEAKER_00: So personally, of course, I have my own agents.
• 10:07
  SPEAKER_00: I think everyone has to do just do my job.
• 10:10
  SPEAKER_00: I'm a product manager by trade, so all my requirements to the technology team, I use AI heavily.
• 10:17
  SPEAKER_00: But from an org perspective, what we're doing, which is kind of exciting, is one of the problems we're trying to solve for our clients is correlation of physical hazards.
• 10:27
  SPEAKER_00: So what happens if let's say you have your bank, you have mortgages across, you know, a big land area, let's just say in the US, you have mortgages across the Midwest and Northeast, and suddenly you have extreme heat, you know, uh across Chicago, up to New York, and then Florida, and then you have at the same time drought, and you have strong winds.
• 10:54
  SPEAKER_00: What happens to your portfolios then?
• 10:56
  SPEAKER_00: You have three, basically three physical hazards at the same time.
• 11:00
  SPEAKER_00: What happens to your portfolio?
• 11:02
  SPEAKER_00: What are the losses that are expected that you can expect from that, from those three hazards that are just happening at the same time?
• 11:10
  SPEAKER_00: That's been a problem that's very difficult to solve, very frankly, very difficult to solve five years ago, almost a lot of computing power, but now we're using AI-driven weather forecasts, we're leveraging that to come up with kind of correlation of hazards.
• 11:30
  SPEAKER_00: So using weather data for climate, um to for climate science, which is again very difficult to do, was almost practically impossible five years ago, but now happening because of AI.
• 11:45
  SPEAKER_00: So that's that's something that it's in progress.
• 11:48
  SPEAKER_00: It's not going to be available in two days.
• 11:52
  SPEAKER_00: But we're using AI for it.
• 11:57
  SPEAKER_00: Yeah.
• 11:59
  SPEAKER_01: Um I think from our perspective, so Climate Collector, we do these AI fluency programs and we partner with impact investors or or philanthropic organizations, and then they send their portfolio of investment companies or startups or or nonprofits our way and ask us to kind of build our AI literacy and support them in building out AI tools for their organization.
• 12:20
  SPEAKER_01: So last year we partnered with uh Klarner, which is Swedish FinTech, and MilkyWire, which is an impact investing norm, um, and took on a number of frontline nature conservation organizations.
• 12:31
  SPEAKER_01: Some of these are tiny, like two, three people teams solving like glacier melting issues as an IPCC scientific uh breakthroughs.
• 12:41
  SPEAKER_01: And others, uh like Bug Life International, tiny little team in Europe trying to fight for pollinators and raise awareness around the importance of pollinators for food and food systems.
• 12:51
  SPEAKER_01: Bug Life International, for example, builds in their capstone projects essentially like a support agent that can help them with their advocacy campaigns and their comms material and their marketing.
• 13:01
  SPEAKER_01: So this tiny little team of two can now expand their work massively and run much more effective campaigns and be more successful in their campaigns.
• 13:10
  SPEAKER_01: Similarly, a woman who's the head of research for coral reef restoration for Blue Alliance, she was able to build a capstone project that essentially when they're out doing coral reef restoration, the data capturing that they're doing, they were able to use voice and drop in data information in a structure and skills system that we had built for her, where information goes straight into the donor report or the ME portal or whatever else she needs.
• 13:37
  SPEAKER_01: So instead of keeping her scientific team like out of the water, like filling in the data between dives, like they just kind of dumped the information, went back in to work on Coral Reefs.
• 13:47
  SPEAKER_01: And that makes me really excited.
• 13:49
  SPEAKER_01: Like seeing them get time back and actually being able to expand the impact they have is really what matters.
• 13:55
  SPEAKER_03: How do you um how do you balance the use of frontier models, the the anthropics, the open AIs, the grox, the Gemini's of this world with things you build yourself?
• 14:06
  SPEAKER_03: And how do you blend those two things?
• 14:09
  SPEAKER_00: So we do use I I can only speak for my little world in in climate.
• 14:14
  SPEAKER_00: There is use of those tools, there's use of uh um different models, um, but we do have an overlay to it.
• 14:22
  SPEAKER_00: Um I think um SP is a regulated entity, so you know it comes with a lot of guardrails.
• 14:30
  SPEAKER_03: Um I think the assumption would be certainly from me in the audience that it'd be very difficult for you to quickly access new tools.
• 14:36
  SPEAKER_03: Whereas those of us are more of a start or startup, you can use something new tomorrow.
• 14:40
  SPEAKER_00: Yeah.
• 14:41
  SPEAKER_00: So I uh before I was SP, I was in a startup.
• 14:45
  SPEAKER_00: So I always think, you know, if we had these tools five years ago, we could have done so much and just you know built the product.
• 14:52
  SPEAKER_00: I could have just built them, built the product.
• 14:55
  SPEAKER_00: But um today, of course, we're part of a bigger company.
• 14:58
  SPEAKER_00: It's very difficult.
• 15:00
  SPEAKER_00: But so we do use the the frontier models, but there is a we white label it, then there are a lot of guardrails around it.
• 15:07
  SPEAKER_00: Um make sure that, you know, the for example, because we use it, we use LLMs, for example, to help make the data digestible for for clients, um, to to help with interpretation of the data and make sure that you know the guardrails are there so that it's based on the vetted sources, it's based on the our data.
• 15:30
  SPEAKER_00: It doesn't grab data from you know just the web and kind of start interpreting it um outside of that ecosystem.
• 15:38
  SPEAKER_00: So we there is that layer, multiple layers actually, of like and guardrails around that from a big company perspective.
• 15:48
  SPEAKER_05: Yeah, I mean, the frontier models are incredible.
• 15:51
  SPEAKER_05: It's like I use uh probably too many race car analogies, but if you have this unbelievable engine in your car, then obviously, you know, you can go faster.
• 15:59
  SPEAKER_05: And so anytime we get a new engine, we're gonna put it in.
• 16:01
  SPEAKER_05: Um, at the same time, you have to build the car around it.
• 16:04
  SPEAKER_05: And so that's I think where we differentiate.
• 16:06
  SPEAKER_05: Not just us, but any startup, any company building with frontier models has to differentiate by building their own, you know, harness you'd use the term harnesses, you could uh use the term agents, but there's lots of different ways to kind of wrangle this beast.
• 16:19
  SPEAKER_05: And ultimately we have to build our own IP, uh, assuming that the engine will continue to get stronger and faster and better while also kind of focusing on the specific use cases of our clients and being confident that we're gonna stay in front of uh any frontier model developments.
• 16:35
  SPEAKER_05: So it's a really awesome time to build.
• 16:38
  SPEAKER_05: And I'm so grateful for them because, you know, we have we're only a team of 10, but I've led pretty large software organizations in my career, up to 450 people.
• 16:46
  SPEAKER_05: And I mean, we're developing not quite at 450 person pace, but probably at 50, 60 person pace with 10 people.
• 16:52
  SPEAKER_05: And we're having the most fun we've ever had because even on the business side, like, you know, I can go uh add to the product and get something shipped overnight.
• 17:00
  SPEAKER_05: And it's just this like super fast feedback loop.
• 17:03
  SPEAKER_05: So um it's truly awesome and uh yeah, an exciting time.
• 17:07
  SPEAKER_03: You you mentioned about staying ahead of the front end models there.
• 17:11
  SPEAKER_03: Do you think about the risk of them deciding they want to copy or essentially take on your business model because they have access to elements of your data and your approach?
• 17:20
  SPEAKER_05: Totally, yeah.
• 17:20
  SPEAKER_05: Um and they probably should try because our market is awesome and there's lots of cool things to build around land.
• 17:26
  SPEAKER_05: Um but yeah, I mean, again, at the end of the day, software has always, at least in my career, been about identifying the right customer and building something that that matters to them.
• 17:34
  SPEAKER_05: And it's not just the functionality that's an important piece, but it's understanding them as humans and and you know what they really need to get out of it, the relationship so they have somebody to call.
• 17:43
  SPEAKER_05: So there's a lot more to building a business than just building a better model.
• 17:47
  SPEAKER_05: Um, we hope that they continue to build better models.
• 17:49
  SPEAKER_05: I think that makes our our uh position in the industry stronger, especially since we're kind of a market maker between developers, allocators, and landowners.
• 17:57
  SPEAKER_05: Um, you know, that's gonna be hard for I'd love for Dario to try going to a forestry conference in Birmingham, Alabama.
• 18:03
  SPEAKER_05: But um, but I think, yeah, we'll we'll always have an edge there.
• 18:06
  SPEAKER_03: So for those of you who haven't been to a forestry conference, that is an incredibly good line bravo.
• 18:12
  SPEAKER_03: Um I I often get asked this a lot.
• 18:14
  SPEAKER_03: Are you wedded to one frontier model and you build and and try and work with that?
• 18:18
  SPEAKER_03: Or when your your teaching skills around it, are you focused on one or do you try and stay across a spectrum and and do you do you hold on to that one model strongly or do you jump when the next iteration comes?
• 18:28
  SPEAKER_01: We're total prostitutes.
• 18:30
  SPEAKER_03: Um that is the line of the podcast.
• 18:34
  SPEAKER_03: I don't say that.
• 18:36
  SPEAKER_01: No, I think uh I actually, I mean, I when we teach, this is like one of our biggest recommendations that to be tech agnostic.
• 18:42
  SPEAKER_01: So like don't like lock yourself in too early in one ecosystem for lots of reasons.
• 18:47
  SPEAKER_01: I, you know, I when I worked at Meta and my job was essentially to match Meta's um products and and AI capabilities and data sets with social impact problems, like the sustainable development goals is what we called it at the time.
• 18:59
  SPEAKER_01: And um like there were decisions that were well, well, well beyond the social impact team in terms of product and depreciation of uh APIs and codes and whatnot.
• 19:09
  SPEAKER_01: So my recommendation then today for always will be like you can't control these companies.
• 19:15
  SPEAKER_01: You don't, you can't so if you put all your data and all your organization and everything in one place, like you're gonna have to live with that dependency.
• 19:22
  SPEAKER_01: And that you might be okay with that.
• 19:24
  SPEAKER_01: But for for a lot of organizations, at least like impact organizations, uh like entrepreneurs, et cetera, that could be a bit tricky.
• 19:30
  SPEAKER_01: So I think that's one piece.
• 19:31
  SPEAKER_01: And then I think also there's like it I'm a little concerned as like a responsible technologist advocate, if you will, of the increasing black box boxness of it.
• 19:41
  SPEAKER_01: I think there's uh there's a whole community of kind of green software developers and uh climate or nature organizations that are excited about these technologies, but would like to make sure that they're not being wasteful.
• 19:53
  SPEAKER_01: Like they don't want to take the Ferrari to the grocery store.
• 19:55
  SPEAKER_01: They want to kind of match task to tool and make sure that they're using like efficient models and responsible and safe and ethical models.
• 20:03
  SPEAKER_01: And I think the the more there's a tension here between convenience, like even just like Anthropic the other day, when they're bundling all their tools, and now you kind of like they will make the choice for you.
• 20:13
  SPEAKER_01: What what what's happening under the hood?
• 20:16
  SPEAKER_01: I think it's powerful to make that choice myself.
• 20:19
  SPEAKER_01: And like I want to understand the emissions of these options and the data safety of the options, and then kind of decide where I what I share and how I engage with them.
• 20:26
  SPEAKER_01: And I think that's getting a little harder right now.
• 20:28
  SPEAKER_01: So that's uh I don't know, it's a concern I'm having.
• 20:30
  SPEAKER_03: I remember listening to an awesome thing from Boris Chernsky, who created Claude Code originally, said every time they release a new model, they start again.
• 20:39
  SPEAKER_03: Anything they built before, anything they work with, they completely start from fresh.
• 20:43
  SPEAKER_03: Because having a new model run something is like having a completely different employee.
• 20:47
  SPEAKER_03: You have to think about it in different ways.
• 20:49
  SPEAKER_03: Have you got examples where you've something's been working really well and you've had to pivot to take the next step in the iteration?
• 20:56
  SPEAKER_05: Every time a new model comes out for us, we test it a lot.
• 20:59
  SPEAKER_05: Um, our team calls it weird mode because we just don't know what it's gonna do.
• 21:02
  SPEAKER_05: Uh, and then eventually we release that to clients.
• 21:04
  SPEAKER_05: I think the the most interesting stuff on model choice to me is around speed versus processing power versus certainty and understanding and you know, trying to balance all those things.
• 21:15
  SPEAKER_05: There's certain parts of any application where you just want it to go fast and kind of do simple things quickly, other parts where you really want it to think, but it's gonna take a while, it's gonna cost more, uh, and finding the right balance across those different areas.
• 21:28
  SPEAKER_05: So when there's a new model, generally it's gonna have um better reasoning and and sort of better deep processing capabilities.
• 21:35
  SPEAKER_05: It's also going to be slower and more expensive and weirder and uh more unpredictable.
• 21:40
  SPEAKER_05: And so um, I think we're always trying to find that edge of what's useful to clients because you don't want to hold back too much until it's perfect, it'll never be perfect.
• 21:47
  SPEAKER_05: And you also have to trust in your clients and their abilities to decipher kind of crap that comes from AI.
• 21:53
  SPEAKER_05: Uh, but but also um they're paying you to do to bring you beyond what cloud can do.
• 21:57
  SPEAKER_05: And so if you're just kind of shipping what cloud can do.
• 22:00
  SPEAKER_05: Cloud is just shipped or OpenAI or whomever.
• 22:02
  SPEAKER_05: Open weight stuff is really interesting, exciting right now, too.
• 22:05
  SPEAKER_05: Um, then yeah, what are they really paying for?
• 22:07
  SPEAKER_05: So it's a really delicate balance.
• 22:09
  SPEAKER_05: Um, and I'd say probably one of the biggest strategic choices we make is kind of model choice across the application.
• 22:16
  SPEAKER_03: Um we will get on to open weight stuff, I I hope.
• 22:19
  SPEAKER_03: But i interestingly, you you've each had different chapters of working with technology and different style of organizations and and shapes of organizations.
• 22:26
  SPEAKER_03: How does this period of technological change change from those you've or differ from those you've observed previously?
• 22:34
  SPEAKER_00: I think this uh pace uh is very disruptive.
• 22:38
  SPEAKER_00: It's it's incredibly disruptive in a in a good and bad way.
• 22:41
  SPEAKER_00: In a good way, again, going back, it's just uh this pace has made has uh and I I point to my startup experience uh five, six years ago, where there were so many things we couldn't do.
• 22:55
  SPEAKER_00: Um so and in in in in climate science, there's so much modeling, there's so much um simulations that need to be run, we just couldn't do it.
• 23:04
  SPEAKER_00: Um now uh we can do it, not because we're a big part of a big company, but because of AI.
• 23:11
  SPEAKER_00: So the pace of change has really made us uh in a way uh get to a point where we could do the things, we could run simulations, we could we could explore expansion of the product um uh because of AI, not because we're part of a bigger org, but because we have that technology.
• 23:32
  SPEAKER_00: And so it's it's it's making product development faster, efficient, and you know, scalable um for us.
• 23:41
  SPEAKER_00: And um but yeah, so I think that just goes back to what I'm very excited about with AI.
• 23:49
  SPEAKER_00: But again, of course, there's with with a lot of a lot of tools comes the responsibility of having to have the guardrails, having to having to really safeguard having the governance for the use of that technology.
• 24:04
  SPEAKER_03: You mentioned the speed of change.
• 24:06
  SPEAKER_03: Is it the breadth as well?
• 24:07
  SPEAKER_03: Like my sense is that it's affecting every part of the audience.
• 24:10
  SPEAKER_00: Every every part, yeah.
• 24:11
  SPEAKER_00: So from our day-to-day, you know, just emails, having agents go through your inbox and just not having to read, you know, 100 emails per day, up to product actual product development, uh, up to coding of the engineers.
• 24:30
  SPEAKER_00: Um it's not coding everything for sure, but you know, there's the starting point.
• 24:35
  SPEAKER_00: There's that so it's across everything.
• 24:38
  SPEAKER_00: And I'm not sure about the executives, but for sure there's it it has impacted everyone.
• 24:44
  SPEAKER_03: Do you think the speed of change is slower at the executive level than it is at the junior?
• 24:48
  SPEAKER_00: Probably.
• 24:49
  SPEAKER_03: Probably.
• 24:51
  SPEAKER_00: Probably.
• 24:51
  SPEAKER_03: Anna, what's your sense on that?
• 24:52
  SPEAKER_03: Do you think we're seeing adoption from young people earlier than those experienced in their career?
• 24:57
  SPEAKER_01: Uh yeah, but I think we've we've always seen that.
• 24:59
  SPEAKER_01: Like, I think one way for for me to get into the big projects when I was younger and worked in Africa was like with all these big German engineers.
• 25:05
  SPEAKER_01: Like, okay, I was like, well, do you know that cell phones have sensors?
• 25:08
  SPEAKER_01: We can actually measure things out.
• 25:09
  SPEAKER_01: And they're like, really?
• 25:10
  SPEAKER_01: And then, you know, so I think young people are always pretty quick at picking up technology as a way to like get get heard or get, you know, get a seat at the table.
• 25:17
  SPEAKER_03: This may shock people, but I was young at some point.
• 25:19
  SPEAKER_03: Um I I worked for the government for nine years at the start of my career.
• 25:22
  SPEAKER_03: Please don't hold it against me.
• 25:23
  SPEAKER_03: Um I remember being the youngest director in the room.
• 25:26
  SPEAKER_03: And when new phones got issued, the senior directors, if they rang in a meeting, they didn't know how to silence them.
• 25:32
  SPEAKER_03: So they'd just like pass me the phone to try and turn it off because I was the tech guy, because I was under 30 at the time.
• 25:37
  SPEAKER_03: So yeah, maybe it's always been true that way.
• 25:40
  SPEAKER_01: Yeah.
• 25:40
  SPEAKER_01: But but I do think what like a big difference today is uh the unlike it is going so fast that I think like there's a massive shadow AI use in in most organizations.
• 25:52
  SPEAKER_01: I mean, clearly these are examples of very technology native organizations here, but most of the groups that I'm working with, like, you know, the large climate organizations or governments, or you know, that like that's not what's really happening.
• 26:04
  SPEAKER_03: We are in a very small purple bubble.
• 26:08
  SPEAKER_01: So yes, we should acknowledge that.
• 26:10
  SPEAKER_01: Um and I think there's uh, you know, when when cell phones came out and and we were, I worked a lot with um like farmers in Africa and with when I was with the World Bank and trying to figure out, okay, now they can have information about what the tomato prices are in that village, so that actually that these women who are not like making it out to the markets, they're not dependent on some, you know, sketchy guy who tells them what the price is, they can actually get information for themselves.
• 26:32
  SPEAKER_01: Like there were some of those like kind of steps steps stepwise changes in technology that empowered people.
• 26:40
  SPEAKER_01: What I think is happening now, which I don't I don't think we're thinking about this well enough, is we can actually change the makeup of organizations, like to change the way we are are thinking about this talent across the field or knowledge and expertise that you need, expertise that you can complement with, like this kind of human AI, human sandwich of like making sure that you have humans in the beginning and then you use technology to kind of advance or augment what you do, and then you bring it back to the human aspect.
• 27:10
  SPEAKER_01: Like I'm really excited to nudge, and it's probably not next year, but in a couple of New York climate weeks from now, maybe there'll be much more like kind of remolding of what an organization is and how we are like, oh, I'm fighting for my budget, my team, my mandate, versus like, no, actually, like maybe like agents, we can agree on certain goals.
• 27:31
  SPEAKER_01: I mean, in fact, we have goals, like we have kind of clear climate goals, we have clear nature goals.
• 27:35
  SPEAKER_01: What if we would think a little more like agents and be like, okay, here's the goal?
• 27:38
  SPEAKER_01: What is the rubric cubes solutions that exist and which one should we choose together as a sector and drive towards?
• 27:44
  SPEAKER_01: And I that could potentially be a lot more effective than what we're doing today.
• 27:48
  SPEAKER_05: I I totally agree with that.
• 27:49
  SPEAKER_05: And I think that's actually one of the biggest challenges I see with AI adoption is, and this is where I would say executives are maybe leaning in too hard.
• 27:57
  SPEAKER_05: People assume that AI is going to solve a lot of problems, but AI is just a tool.
• 28:01
  SPEAKER_05: You have to figure out the problem first.
• 28:02
  SPEAKER_05: And you see these business transformation teams saying, what's our AI strategy?
• 28:07
  SPEAKER_05: And organizations are spending enormous amounts of money on AI adoption and tools.
• 28:11
  SPEAKER_05: It's really unclear if it's actually producing much.
• 28:14
  SPEAKER_05: And I think, you know, there will be a shift back.
• 28:16
  SPEAKER_05: It probably backlash will happen quickly, actually, uh, where people say, okay, our budgets have been blown out of the water, but are our clients getting better services?
• 28:24
  SPEAKER_05: Is our team more efficient?
• 28:25
  SPEAKER_05: And so it'll go back to goals.
• 28:27
  SPEAKER_05: And that's when I think kind of the beautiful stuff starts to happen is people say, okay, I just want to solve this problem.
• 28:32
  SPEAKER_05: But the problem can maybe be a bit more ambitious and bigger than the one that I was focused on three years ago.
• 28:36
  SPEAKER_05: And um, and then yeah, using the best capabilities of AI to do that.
• 28:41
  SPEAKER_05: But um, but yeah, I think many folks are just sort of in this uh, like for us, it was the opus moment when this really powerful model came out.
• 28:47
  SPEAKER_05: Oh, it can do everything.
• 28:48
  SPEAKER_05: And then as you actually get into the details and talk to the people, the younger people using it, you realize actually, yeah, it can do a lot, but it's broken in many ways and it hallucinates, and I don't trust it, and it doesn't sound like new and I write and all these other problems.
• 29:00
  SPEAKER_05: And so then you start to get into the minutiae of you know delivering something powerful, and that's it's hard.
• 29:05
  SPEAKER_03: Anna mentioned um talent there in terms within a business.
• 29:08
  SPEAKER_03: Do you think AI has fundamentally changed what we need to look for in talented people entering organizations?
• 29:14
  SPEAKER_03: Are those skill sets different than they were five or ten years ago?
• 29:18
  SPEAKER_05: I was just talking, I'll go first because I was just talking to somebody about this uh last night.
• 29:22
  SPEAKER_05: Um yes, uh we no longer need to hire people to process and just do tasks.
• 29:28
  SPEAKER_05: And that was a big part of large organizations before.
• 29:31
  SPEAKER_05: You just needed capacity.
• 29:32
  SPEAKER_05: And I think capacity is almost becoming a commodity.
• 29:35
  SPEAKER_05: And now it's about creativity, taste, drive, and vision, and ability to kind of connect with the the market and the clients and bring it back into the organization and communicate that.
• 29:44
  SPEAKER_03: Do you make that taste one an example?
• 29:46
  SPEAKER_03: Because people say it a lot, but you rarely see an actual example.
• 29:50
  SPEAKER_03: Can you give me one where a person has expressed a taste or a preference for something that's been valuable?
• 29:55
  SPEAKER_05: I think it's a lot of for me, I think about taste in the micro sense of all the daily decisions.
• 30:01
  SPEAKER_05: So for example, it was about six months ago, we started to get on this really extraordinary cadence of shipping stuff to our product.
• 30:08
  SPEAKER_05: And every day we were adding 15, 20 new features and we're just totally leaned in and AI pilled.
• 30:13
  SPEAKER_05: And and I would get on client demos and realize like, I don't even know what the product's gonna look like day over day.
• 30:19
  SPEAKER_05: And our clients who are going in much less frequently than me certainly don't.
• 30:22
  SPEAKER_05: And so, how do you rein that in and really focus on the biggest problems uh and and you know, bring some perspective to that?
• 30:29
  SPEAKER_05: And then also really think through like how the person is working with it and processing it and give them an experience that they'll remember, not just give them more compute.
• 30:39
  SPEAKER_05: Um, and so there's lots of things around that.
• 30:42
  SPEAKER_05: Um, one example I often bring up around not AI related, but very closely tied, is when I was at booking.com, um, I was leading our front-end product team and we had extraordinary data, 60 billion in transaction volume and a huge number of users.
• 30:57
  SPEAKER_05: And uh, you know, we could predict with very high confidence what hotel somebody was going to book after they'd already booked two or three on the site.
• 31:04
  SPEAKER_05: So they'd go in and put in their dates and their and where they wanted to stay, and we would pretty much know with 90 plus percent confidence which hotel they would end up booking.
• 31:12
  SPEAKER_05: But if you showed that in the first couple of results, they almost never booked it and they would exit the session and maybe they would come back later, maybe they wouldn't.
• 31:19
  SPEAKER_05: But if you put it on page two or three, they would almost always book it.
• 31:22
  SPEAKER_05: And so they want to go through this sort of process of understanding and learning and feeling like they're in control of the product, not us just like booking a hotel for them.
• 31:30
  SPEAKER_05: So when I see these AI products now, like Muse, book me a hotel in New York for Climate Week.
• 31:35
  SPEAKER_05: It's that's not gonna be a good rewarding experience.
• 31:37
  SPEAKER_05: And there's no taste in that type of journey for a user.
• 31:40
  SPEAKER_05: And so, how do you think about the human, the human side of it and bringing your product to where they are?
• 31:45
  SPEAKER_01: Yeah, it was very much agreeing in lots of different concepts.
• 31:48
  SPEAKER_01: But I think the same is like I work a lot on on throughout my years, on carbon markets.
• 31:53
  SPEAKER_01: And and now we're looking at, okay, like what's AA governance for carbon markets and how do we make sure that like we we we take care of this markets and you know, support it to scale and strengthen integrity and quality and all that stuff.
• 32:04
  SPEAKER_01: But there's also a like a a taste piece in there.
• 32:08
  SPEAKER_01: Like what what carbon credit a company would want to pursue in the end is very yes, there is a mathematical, you can do a lot of math around it.
• 32:16
  SPEAKER_01: But then there's also like that kind of sense.
• 32:18
  SPEAKER_01: Is there a story around that?
• 32:19
  SPEAKER_01: Is it like does it connect to something?
• 32:21
  SPEAKER_01: It could be connecting to some uh you know, CEO and her like childhood memories of something.
• 32:27
  SPEAKER_01: Like you can't mathematically control for those things.
• 32:30
  SPEAKER_01: So I think that's one thing that like you you're still gonna need like that fingertip kind of humanness of things.
• 32:37
  SPEAKER_03: Self-plug coming here, so stand by.
• 32:39
  SPEAKER_03: Uh we actually hold the world record for the highest price anyone's ever paid for nature-based carbon credits.
• 32:44
  SPEAKER_03: So both Woodland and Peatland, a big proportion of that relates to the storytelling and the connectivity that that come from it.
• 32:51
  SPEAKER_03: I I did a uh a panel at one of the world's leading law firms last night, and someone in the front row went, Well, no one would ever pay more for a carbon product than they would an an all an alternative carbon product.
• 33:01
  SPEAKER_03: Why wouldn't you just buy it?
• 33:03
  SPEAKER_03: And I couldn't help with the irony that she had a Chanel bag under her chair.
• 33:07
  SPEAKER_03: And I was just like, some people choose to buy things because of other reasons rather than the utility value.
• 33:12
  SPEAKER_03: So it's really interesting.
• 33:13
  SPEAKER_03: Um, anything else on the the people and the talent side of things?
• 33:16
  SPEAKER_00: Yeah, I think just to I was thinking while they were speaking that maybe product managers will become cool again because you know, I was starting to feel uncool about this whole AI thing.
• 33:28
  SPEAKER_00: Um, but I think in a just in addition to that, it's like problem solving.
• 33:32
  SPEAKER_00: I think that's something you cannot AI your way into.
• 33:36
  SPEAKER_00: So I think uh problem solving will and intuition, you know, intuition and problem solving, I think will be cool again, will be great again because of you know that you cannot AI your way through some problems.
• 33:52
  SPEAKER_03: We we tend to uh we tend to use sporting analogies quite a lot in terms of our recruitment and and what we do.
• 33:58
  SPEAKER_03: And and we're looking for hybrid athletes now.
• 34:01
  SPEAKER_03: So before we used to hire for specialists, and we'd want someone who could come in and code, or we'd want someone who's distinctly an economist, and uh and that's been really good for us.
• 34:08
  SPEAKER_03: We we have got more PhDs in the business than DeepMind had when it's all to Google.
• 34:13
  SPEAKER_03: That's great, that's helped us get to this point.
• 34:15
  SPEAKER_03: However, now with young people and we look to bring them in, we want the generalists because you can go find an expert opinion in minutes using a frontier model, but to be curious and open-minded and hardworking and to be able to embrace and seek change rather than be nervous about it, those skills are more important than they've ever been before.
• 34:34
  SPEAKER_03: Um But if you were if you were in the room and you were a young person thinking about their career right now, what advice would you give them about the early career steps, first jobs, or even whether to go to university, things like that?
• 34:46
  SPEAKER_03: How would how would you reflect on that now?
• 34:52
  SPEAKER_00: I would still go advice for university.
• 34:55
  SPEAKER_00: I think just the the learning, you know, un university I think teaches you how to enjoy learning, how to learn.
• 35:02
  SPEAKER_00: Um, but I'm biased of it.
• 35:04
  SPEAKER_00: I still want to go to university.
• 35:06
  SPEAKER_00: But I think you mentioned something, Rich, that would that I would advise.
• 35:11
  SPEAKER_00: I think curiosity, um, and not be scared of just curious learn about other of different things.
• 35:20
  SPEAKER_00: You know, I think that's something that's priceless, that cannot be replaced, cannot be cannot be um sold, cannot you cannot have an algorithm for it.
• 35:31
  SPEAKER_00: It is just a a personal thing, just be curious and learn about other different things.
• 35:41
  SPEAKER_01: Maybe I think that's really, really great um suggestions, actually.
• 35:45
  SPEAKER_01: Uh kids at home, so I I think about this a lot, but maybe kind of bringing it forward a little bit.
• 35:49
  SPEAKER_01: This is some recommendations that I sometimes give on LinkedIn when people reach out and like, oh, how do like how can I I want to be more active in LinkedIn?
• 35:55
  SPEAKER_01: What do I do?
• 35:56
  SPEAKER_01: And I'm like, I I think it's quite interesting time to collaborate and like you can make yourself really visible and get work-life experience, even though you might not secure a job, because I do think that it's going to be quite a tough market out there right now, I would say.
• 36:11
  SPEAKER_01: But kind of volunteering your time or like, I'm writing this blog post.
• 36:14
  SPEAKER_01: Anna, you seem to be working on something similar.
• 36:16
  SPEAKER_01: Should we do you want to co-write this with me?
• 36:18
  SPEAKER_01: And I don't have a lot of time.
• 36:19
  SPEAKER_01: But if someone comes with to me with a draft that like is quite interesting, yeah, if you add value to someone else, people are quite keen to collaborate and and engage.
• 36:28
  SPEAKER_01: And and I think this is, you know, like kind of setting up if you have a platform, bringing in people to talk on that platform is a perfect, like it adds value to everyone.
• 36:37
  SPEAKER_01: Like that's a really kind of you give to the audience, to the participants, to yourself.
• 36:41
  SPEAKER_01: Like I think that's really valuable.
• 36:42
  SPEAKER_01: So thinking a bit more out of like how can you be involved and partner up with people and doing things that way so that you get visibility and experience that might add to your CV or improve your position, but like not need that like, you know, vacation plan and the salary and the because that's hard to get these days.
• 36:59
  SPEAKER_03: I um I feel at risk of being a hypocrite about this at the minute.
• 37:01
  SPEAKER_03: So I stayed at university far too long.
• 37:03
  SPEAKER_03: I was there eight years.
• 37:04
  SPEAKER_03: I went degree straight into PhD, largely because I just wanted to avoid having an actual real job.
• 37:10
  SPEAKER_03: And people ask me now about should I do a PhD?
• 37:12
  SPEAKER_03: Surely that's a good thing.
• 37:13
  SPEAKER_03: I just think locking away four or five years of your career right now with how fast they my business didn't exist five years ago.
• 37:21
  SPEAKER_03: Um, the world is changing so, so quickly.
• 37:23
  SPEAKER_03: That's the bit that I feel nervous saying to people, yeah, go for that thing.
• 37:28
  SPEAKER_03: Um, am I wrong?
• 37:29
  SPEAKER_03: Should I be giving different advice?
• 37:31
  SPEAKER_03: I mean, I always caveat that with you don't want to take my advice, do the thing that feels right for you.
• 37:35
  SPEAKER_03: But if you're insistent on it, here's some of the questions.
• 37:38
  SPEAKER_05: I I mean, I went to law school and I had uh for three years and I was a lawyer for two years.
• 37:42
  SPEAKER_05: So it's kind of a bad ROI.
• 37:44
  SPEAKER_05: Um but I do think, you know, a lot this kind of goes to the point on what I would focus on.
• 37:50
  SPEAKER_05: Um, law school taught me to communicate really concisely and write really well.
• 37:54
  SPEAKER_05: And I think that skill is enormously valuable now, maybe more so than ever.
• 37:58
  SPEAKER_05: I see so much AI slop out there, and I don't know who people are anymore because I'm just reading stuff that their agent has written for them.
• 38:06
  SPEAKER_05: And so when we're hiring, when I think about who I want to build relationships with and businesses with, it's people who understand who they are and can communicate who they are.
• 38:15
  SPEAKER_05: I think school can't help with that because you're surrounded by other intelligent people who are kind of constantly pushing at it.
• 38:20
  SPEAKER_05: And so that skill I think you absolutely want to invest in.
• 38:23
  SPEAKER_05: Uh, because if you just use AI tools to write your emails, then you're gonna sound like every other person on the planet and it's gonna be pretty boring.
• 38:29
  SPEAKER_05: And so uh I think that's a key part.
• 38:31
  SPEAKER_05: And then the other key thing that I think about a lot is you know, as old like Mac versus PC ads.
• 38:37
  SPEAKER_05: There was this early days of Silicon Valley where software was quite romantic and it was all about kind of supporting humanity and making us better at what we already are.
• 38:47
  SPEAKER_05: And uh I think that has an opportunity to come back now.
• 38:50
  SPEAKER_05: Um, and really, you know, we're gonna have unlimited processing.
• 38:53
  SPEAKER_05: So we're not gonna need the PC side of that advertisement anymore.
• 38:57
  SPEAKER_05: Um, but what do we do with all of that processing?
• 38:59
  SPEAKER_05: And and are we building a future that we believe in and that we care about?
• 39:03
  SPEAKER_05: Um, are we just kind of trying to make money on markets or whatever?
• 39:06
  SPEAKER_05: Uh the people who are building things that can connect to other humans that are going to leave the planet better than they found it, people others will be drawn to that person.
• 39:15
  SPEAKER_05: And that person will always have opportunities because it's actually very hard and scary to put yourself out there and do something, yeah, that that really aligns with your vision for the world.
• 39:25
  SPEAKER_05: And so understanding what that vision is, and maybe it takes eight years in school to figure it out.
• 39:30
  SPEAKER_05: Um, but whatever it takes to get there and then orient your life around that, I think that will serve you well.
• 39:36
  SPEAKER_01: I think building on that, I I I would I I think less of like the actual format, but I do think that that formative time in life when you do things and you are with people that are maybe not your family or like kind of the those you grew up with are so important.
• 39:50
  SPEAKER_01: And you can probably create some kind of experience for for some years that gives similar shape that's even outside university context.
• 39:58
  SPEAKER_01: But I think kind of like being an interesting human being is also part of like having a story and having done some weird stuff and you know, being able to communicate that and share that in in a kind of a constructive way.
• 40:09
  SPEAKER_01: So I think that's that's really would be what I would try to lean into if I was a young person right now.
• 40:14
  SPEAKER_01: Like even if you're taking an untraditional, like maybe not law school, or you're doing some kind of field work somewhere, like that's still gonna help shape you and make you a really interesting potential employee versus someone that just looks and speaks and does the same.
• 40:28
  SPEAKER_03: We um we read CVs bottom up.
• 40:31
  SPEAKER_03: So we look at the thing that's on the end.
• 40:32
  SPEAKER_03: What are the interests, the experiences?
• 40:34
  SPEAKER_03: What have you done?
• 40:34
  SPEAKER_03: Have you achieved something spectacular that was difficult, unpleasant, show discipline?
• 40:39
  SPEAKER_03: If you do, let's have a conversation.
• 40:41
  SPEAKER_03: Because then if the person wants to be somewhere, they know what it takes to be successful that way.
• 40:45
  SPEAKER_03: Now we're really fortunate we get thousands of applicants for each role because we're in a really exciting sector to be in.
• 40:50
  SPEAKER_03: So it means they do come with the academics as well.
• 40:52
  SPEAKER_03: But I've always looked CV's bottom-up first because I think it's it's more about the character than anything else.
• 40:58
  SPEAKER_03: You mentioned about the people building really positive things for the world and people and humanity.
• 41:05
  SPEAKER_03: Um, opportunity to plug or shout out any businesses, companies, platforms, tools that you like, what they're doing and how they do it at the moment.
• 41:14
  SPEAKER_01: What comes to mind quickly is I'm really excited about some of the smaller uh LLM companies.
• 41:20
  SPEAKER_01: Like Green PT is one group of people in Holland, I think, that are like building a smaller, greener um alternative to kind of a general purpose chatbot.
• 41:29
  SPEAKER_01: That I think is great because I think a lot of people in my sector are quite concerned about the energy footprint of AI.
• 41:34
  SPEAKER_01: So having a model that is more transparent and you know, there are options kind of similar, like ethical and data safety, and you know, people working with indigenous people are really concerned about data sovereignty and et cetera.
• 41:45
  SPEAKER_01: And I'm we're starting to see products that are catering to that.
• 41:48
  SPEAKER_01: So I mean, you spoke really wisely of like find your customer and build something that's really good for them.
• 41:52
  SPEAKER_01: There's a big, big world out there, and there are lots of different needs.
• 41:56
  SPEAKER_01: So I think that's I'm really excited to see that that we're not just kind of doing these massive mainstream models, but actually people are um trying alternative paths.
• 42:06
  SPEAKER_00: I'm gonna come up with a boring example, insurance.
• 42:09
  SPEAKER_03: So I don't think insurance is I used to think insurance was boring, and then I was wrong.
• 42:13
  SPEAKER_03: Insurance is fascinating.
• 42:14
  SPEAKER_03: And also it's like um it's the canary in the coal mine.
• 42:17
  SPEAKER_03: I don't know if that analogy translates across the air.
• 42:21
  SPEAKER_03: It lets you know what's coming.
• 42:22
  SPEAKER_03: It's a bit like taxation.
• 42:24
  SPEAKER_03: If insurance and tax are interested in something, it's going to explode because it's worth doing.
• 42:29
  SPEAKER_03: So I love it.
• 42:30
  SPEAKER_00: So I think one of the things, at least in the US, I and you know, what we're seeing now is we've been inundated in the news about insurers, big insurers pulling out of certain um states because of it's become too expensive to in to cover, you know, Florida for for flooding or for or wildfire or California for wildfires, et cetera.
• 42:54
  SPEAKER_00: But what I've been looking at um lately is some of these, there are now some insurance uh they're startups, they're tech native startups that are coming up with models to insure houses, for example, that have retrofits.
• 43:13
  SPEAKER_00: Um so they only insure houses that have uh retrofits that are they're meeting fortification standards, for example, that are the US government.
• 43:21
  SPEAKER_00: And I think this this is uh life-changing for for um, especially for residential, um, for for a lot of us.
• 43:32
  SPEAKER_00: It might be, you know, because it it really is material to a household.
• 43:38
  SPEAKER_00: You know, if it's like it it can be, if you have um, and I don't know, I'm not a civil engineer, but you know, if you have certain types of roof, then you get lower premiums for that.
• 43:49
  SPEAKER_00: I think that's the model, and that's I don't know a lot about it yet, but I've been looking at it, and I think it is life changing for a lot of households that are in this, you know, in a hurricane belt.
• 44:00
  SPEAKER_00: Or in wildfire areas because now there is an alternative, right?
• 44:03
  SPEAKER_00: It's not just you wake up one day and oh my God, my my insurance company is not gonna cover this postcode anymore.
• 44:11
  SPEAKER_00: Um what am I gonna do?
• 44:13
  SPEAKER_00: There's wildfire, it's wildfire season.
• 44:15
  SPEAKER_00: What what happens now?
• 44:16
  SPEAKER_00: But now there's an alternative.
• 44:18
  SPEAKER_00: And I think that's very exciting.
• 44:19
  SPEAKER_03: I think wildfire gives us a really interesting case study to dive into this.
• 44:23
  SPEAKER_03: So um we were both unfortunate and very fortunate that a significant um the UK got hit by record levels of wildfire this year.
• 44:31
  SPEAKER_03: Um nothing in comparison to US scale and standards, but for for us it was a huge amount.
• 44:36
  SPEAKER_03: Uh a significant wildfire wiped out 5,000 acres of land a kilometre from our boundary.
• 44:41
  SPEAKER_03: And literally we were sat having operational meetings saying, this is coming, what are the fire breaks we can cut, how can we respond to it?
• 44:47
  SPEAKER_03: But it does illustrate how AI is affecting that in every different way.
• 44:51
  SPEAKER_03: And my favorite use case is the whole array of companies that are coming up with early warning systems and ways of geographically representing that data and information.
• 44:59
  SPEAKER_03: And my particular two I want to shout out are Lovable and Replet.
• 45:03
  SPEAKER_03: So companies that allow you to build individual bespoke software applications incredibly quickly.
• 45:09
  SPEAKER_03: In the face of a massive wildfire in Scotland, the team were able to build a visualization tool that scraped every form of social media for photographs, for data points, for anecdotal references of where this fire was happening on the ground, and literally give that to the responders that were working with it to rapidly improve the data they had available.
• 45:28
  SPEAKER_03: It wasn't perfect, but it was way better than having nothing to respond in that way before.
• 45:32
  SPEAKER_03: And that just wouldn't have been possible without it.
• 45:34
  SPEAKER_03: So I love that example.
• 45:36
  SPEAKER_03: Um you you took us on to um smaller LLMs and open weight models.
• 45:42
  SPEAKER_03: Um maybe you could give uh uh an explanation to the room of the difference between a frontier model and something that's open weight, just at a high level to frame the conversation.
• 45:50
  SPEAKER_03: Oh boy.
• 45:51
  SPEAKER_05: Um put me on the spot here.
• 45:53
  SPEAKER_05: Well, the the frontier labs are obviously uh I don't know how much AI knowledge there is, but essentially what AI is doing is it's predicting what will come next constantly and at rapid speeds, and it's using lots of language to do that.
• 46:06
  SPEAKER_05: Uh so it's a very, very simple explanation.
• 46:08
  SPEAKER_05: And so what they're doing is applying uh different weights and um uh different methods to kind of uh shape the outputs that that they think are appealing to their users and that have broad uh use across all the use cases that the businesses that that use them uh care about.
• 46:25
  SPEAKER_05: And so, you know, that's one version of it.
• 46:27
  SPEAKER_05: When you when people think about anthropic or open AI, they're probably thinking about Claude and ChatGPT, but those are actually products built on top of the models.
• 46:37
  SPEAKER_05: Um, and so there's these underlying models that the organizations themselves are building and it's their proprietary IP.
• 46:43
  SPEAKER_05: Um, you know, now it's in the news a lot is something that we need to secure for national defense purposes.
• 46:48
  SPEAKER_05: But all of that is fundamentally owned by a private company.
• 46:50
  SPEAKER_05: Uh there are open source versions of similar models that are taking a lot of the same methodologies and processing power.
• 46:58
  SPEAKER_05: They also have access to data centers and all these things, although not as much capital.
• 47:02
  SPEAKER_05: And they're open sourcing the weighting and the methodologies to provide the outputs.
• 47:06
  SPEAKER_05: And so if you think about that, um, you're not now bound to a single private organization's decisions on sort of how to evaluate and and uh return results.
• 47:17
  SPEAKER_05: Um, you're looking uh at a large community, and those communities can can change.
• 47:20
  SPEAKER_05: You know, there could be a a nature uh group that is building a model for the purpose of evaluating biodiversity data or something, but you could have the same thing in finance and all these other areas.
• 47:29
  SPEAKER_05: So it's really exciting.
• 47:31
  SPEAKER_05: Um, those models are a lot are much lower cost, largely because they use less compute.
• 47:35
  SPEAKER_05: They're also not as powerful and not as good, but there's a lot of debate right now around how big that delta really is.
• 47:40
  SPEAKER_05: Is it a you know 80% delta or is it a five to 10% delta?
• 47:44
  SPEAKER_05: And you know, if you talk about five to 10% less processing power, but um, you know, uh 80% of the of the results and you know, five percent of the cost or something like that, uh it starts to become really interesting.
• 47:57
  SPEAKER_05: And it seems as if they're coming closer and closer together.
• 48:00
  SPEAKER_05: And so uh there might be a world where open AI, anthropic become a lot less important and these open weight, open source models become much more a part of our core workflows.
• 48:10
  SPEAKER_05: And I'd say that's very much in the current debate right now.
• 48:13
  SPEAKER_05: And honestly, a lot of a lot of people think that this regulatory capture play by anthropic and open AI is related to that because they're scared of these open weight models.
• 48:20
  SPEAKER_05: So there's a lot we could go into here, but um, that's the somewhat short version.
• 48:24
  SPEAKER_03: We get a round of applause for anyone.
• 48:26
  SPEAKER_01: It's really good.
• 48:27
  SPEAKER_01: That's so great.
• 48:28
  SPEAKER_01: Do you want to be a coach for some of our trainings?
• 48:31
  SPEAKER_03: Oh my gosh.
• 48:32
  SPEAKER_03: Um, but the rest of the panel, have you got any experience with the open weight models, the lighter weight LLMs yet and working with those, or are you still very much working with the frontier models?
• 48:41
  SPEAKER_01: We work a lot with small uh local solutions.
• 48:45
  SPEAKER_01: Uh also can can you can kind of contain very specific type of information in there.
• 48:49
  SPEAKER_01: So if you're, you know, if you're working on like eDNA um and you're a science collaborative and and you like you don't need to know you don't need to use massive models to do that because they don't also not train on very relevant information for that.
• 49:03
  SPEAKER_01: So I think, yeah, that's definitely something we're seeing a lot, maybe more in nature and biodiversity than kind of climate and decarbonization, I would say.
• 49:10
  SPEAKER_03: Um you don't need your e-DNA model to be aware of the entire works of Shakespeare.
• 49:15
  SPEAKER_01: That's true.
• 49:16
  SPEAKER_01: Why is that?
• 49:17
  SPEAKER_03: Sheila, do you do you get to access those yet?
• 49:19
  SPEAKER_03: Or because the guardrails you talked about is very much so.
• 49:22
  SPEAKER_00: I think we have there are specific models that we can use.
• 49:25
  SPEAKER_00: Um and so it's pretty limited.
• 49:28
  SPEAKER_00: We don't have as much um flexibility as a smaller hour, but a bit more audience participation.
• 49:34
  SPEAKER_02: Can we have a show of hands if you're playing with or experimenting with open weight models yet?
• 49:40
  SPEAKER_03: Three hands in the room.
• 49:41
  SPEAKER_03: That's four.
• 49:42
  SPEAKER_03: Thank you.
• 49:43
  SPEAKER_03: Um that's interesting.
• 49:44
  SPEAKER_03: I think if we do this in a year, we'll have virtually the whole room, I think, in terms of the pace at which things are changing.
• 49:51
  SPEAKER_03: Okay, we we've stayed very much on the positive side of things, self self-exclosed um proponent of everything AI.
• 49:57
  SPEAKER_03: Um let's go to the other side of things.
• 49:59
  SPEAKER_03: So uh Anna, you mentioned it.
• 50:00
  SPEAKER_03: How conscious are you about the let's go with the environment first, the environmental impact of AI and what it means for the planet?
• 50:08
  SPEAKER_01: Uh how well, very conscious.
• 50:11
  SPEAKER_01: That's probably the most direct answer.
• 50:13
  SPEAKER_01: Um, I think it is after having worked in climate or nature or adaptation or resilience for 20 years, I'm really frustrated that we haven't done more.
• 50:22
  SPEAKER_01: We have not bent the curve when lack of a better, more precise word.
• 50:26
  SPEAKER_01: So I am super excited about this technology and its ability to both make us do more and and like see things in a different way, rewire the economy so that we can bake in these externalities and actually fix what's driving nature loss.
• 50:39
  SPEAKER_01: And but not at the cost of a 25% increase in global GDP that is drive driven largely by like unresponsible commercial, fast fashion, fossil fuel extraction.
• 50:55
  SPEAKER_01: I mean, like we talked touched a little bit upon this kind of scariness of the models going crazy, but like we've lived with the reality that like our climate is going crazy.
• 51:06
  SPEAKER_01: Like we are like we're losing nature and biodiversity at a rate that's not gonna sustain human life.
• 51:11
  SPEAKER_01: You know, like yes, quite a lot longer.
• 51:13
  SPEAKER_01: But you know, my kids are their kids are their grandkids, whatever.
• 51:17
  SPEAKER_01: Um, so I think it would feel really, really, really irresponsible to go down a technology path or a model path that undermines all the progress we've done on decarbonization, on nature restoration, on biodiversity protection.
• 51:33
  SPEAKER_01: Like that doesn't add up to me.
• 51:35
  SPEAKER_01: Like you can't say, like, oh, because we can now, you know, make a digital twin of this ocean habitat, so we can therefore emit like, you know, triple the amount of natural gas power plants beyond the grid that are operating in the US.
• 51:48
  SPEAKER_01: Like that doesn't work.
• 51:49
  SPEAKER_01: Like, so I I I'm very conscious conscious of that.
• 51:52
  SPEAKER_01: That I do also think, though, that there are two paths to fixing that to try to be constructive and positive.
• 51:58
  SPEAKER_01: Like you can push for regulation, and I think there are some people doing that.
• 52:02
  SPEAKER_01: Um, and then you can also, like, my kind of theory of change is I need to empower the people who understand this well.
• 52:08
  SPEAKER_01: Both like the the nature and climate people who don't understand AI, I'm helping them understand AI so they can like understand what's under the grid and under the hood and therefore feel an agency and take action, but also vice versa, like help technologists to actually understand decarbonization and the importance of nature restoration.
• 52:25
  SPEAKER_01: And again, that's actually where the technology can help.
• 52:27
  SPEAKER_01: So like you don't have to be an expert in both anymore.
• 52:30
  SPEAKER_01: You can actually use technology to kind of um sorry, that was a long winding answer, but very honest.
• 52:39
  SPEAKER_00: Yeah, so I think just supporting Anna, it it is uh very cognizant about that.
• 52:45
  SPEAKER_00: Um, I think we're this is the irony of it all because you know, we're excited about all the simulations we could do for climate, but the reality is they that they're stored in data centers.
• 52:58
  SPEAKER_00: And you know, I I think everyone knows the the controversies over data centers.
• 53:04
  SPEAKER_00: You know, there's it it is impacting uh power supply, impacting water, um, at a time when uh heat uh is also you know, there's also correlation between all of these physical hazards and um so it's the impact and I I think the the risk and the impact are like coming together in this day data centers.
• 53:27
  SPEAKER_00: And so that is why we're seeing so much um uh pushback, right?
• 53:32
  SPEAKER_00: I guess the data centers because of all of these um environmental impacts.
• 53:38
  SPEAKER_00: Um and so I don't I I think uh on the positive side that uh data center operators do need to be I think it has to be more than philanthropy.
• 53:52
  SPEAKER_00: It has to be more than, you know, just I think they have to be really invested in uh in making in in in in nature, I guess.
• 54:01
  SPEAKER_00: They have to be really invested in making sure that there's water supply, that it doesn't affect the grid as much.
• 54:07
  SPEAKER_00: They're using renewable energy.
• 54:09
  SPEAKER_00: Actually, I think a lot of hyperscalers are using renewable energy, but there's uh there are other things.
• 54:15
  SPEAKER_00: You know, there's the the water is a big issue.
• 54:17
  SPEAKER_00: I think that that has been the theme of data center discussions this year has been water.
• 54:23
  SPEAKER_00: So it is that irony of it all that you know every we it's accelerating things, but it's also accelerating environmental impacts.
• 54:31
  SPEAKER_03: So we've um we've seen some good examples of the hyperscales in terms of that.
• 54:35
  SPEAKER_03: So you know, Microsoft have largely single-handedly built the global voluntary carbon market.
• 54:40
  SPEAKER_03: And we've had uh Meta especially come out recently and said we are gonna significantly increase entire the quality of entire watersheds around data centers.
• 54:49
  SPEAKER_03: Do you think we'll have to see more from the hyperscales and the frontier models if they do want to get their environmental and social license to build these data centers?
• 54:58
  SPEAKER_00: Yeah, I think I think so because it is it is material for them, right?
• 55:02
  SPEAKER_00: Like uh it's not just I I think it's uh it all of these things, it could lead to operational constraints for them.
• 55:12
  SPEAKER_00: So it could lead to uh uh you know, uh protests against pushback, regulatory issues, um, even operational issues.
• 55:19
  SPEAKER_03: If you're not gonna be able to do that, yeah, we see an 80% anti-data center sort of um like sentiment.
• 55:25
  SPEAKER_00: Sentiment in the UK in the US, yeah.
• 55:27
  SPEAKER_00: And that's regulatory, that translates to regulatory issues, operational issues.
• 55:32
  SPEAKER_00: If you don't have power, if you actually if you actually have power cuts in the area where you are, you could be the world's biggest cloud provider and you still that will affect you operationally.
• 55:44
  SPEAKER_00: So I think they need to do more.
• 55:46
  SPEAKER_00: Um that's very high-level personal opinion.
• 55:50
  SPEAKER_00: They need to do more.
• 55:51
  SPEAKER_00: They need to be invested in it because these things are material to data centers.
• 55:57
  SPEAKER_03: Do you want to add to that?
• 55:58
  SPEAKER_05: Uh there will be a backlash from the backlash, uh, probably sooner than we realize because um, you know, not only are they investing in these communities, they're developing closed loose system closed loop water systems.
• 56:11
  SPEAKER_05: So the water usage is pretty low now.
• 56:13
  SPEAKER_05: And these new data centers, um, they're obviously bringing lots of tax dollars into the communities that they they come into.
• 56:19
  SPEAKER_05: They're helping build out uh great infrastructure, which is useful beyond data centers.
• 56:24
  SPEAKER_05: So there's a lot of positive aspects of data center development that isn't getting talked about much.
• 56:28
  SPEAKER_05: But you know, Loudoun County, Virginia, where historically we've built a lot of data centers in the US, has some of the best schools in the country, has some of the highest property values, people there are fat and happy.
• 56:37
  SPEAKER_05: And uh I think there's gonna be other communities that that want that and are craving that.
• 56:41
  SPEAKER_05: And so um, you know, the stories are all on the negative side, and I don't want to downplay the negatives.
• 56:45
  SPEAKER_05: There are very uh you know scary things about the development, but I think a lot of the sphere is also about what AI is going to do to our to our livelihoods and and our families and our environment uh broadly.
• 56:55
  SPEAKER_05: But but on the overall like use of these capabilities for uh environment and why I said yes to the question at the beginning, um, we, you know, climate change is something near and dear to my heart.
• 57:06
  SPEAKER_05: I've spent most of my uh recent career on it, uh, Carbon Direct before this, we did a lot of the Microsoft work.
• 57:12
  SPEAKER_05: Um, and there's kind of three parts to it.
• 57:15
  SPEAKER_05: One is, you know, is the climate changing?
• 57:16
  SPEAKER_05: And I think everybody, you know, with the brain at least, uh, is going to agree the climate is changing.
• 57:21
  SPEAKER_05: And then there's, is it overall negative?
• 57:24
  SPEAKER_05: And are those negative consequences sufficient that it's worth the cost to combat them?
• 57:29
  SPEAKER_05: And I think those, those second two, there's there's room for debate depending on your context, right?
• 57:33
  SPEAKER_05: If you're looking globally, probably overall it is quite negative.
• 57:36
  SPEAKER_05: Uh, but if you're looking locally, some places aren't seeing pure negative impacts.
• 57:40
  SPEAKER_05: There's there's farmland areas that are more productive now because of climate change.
• 57:43
  SPEAKER_05: And so um, so AI uh and the ability to take all this information, process it, and tell stories at the macro level, but also go down to the regional and micro level at the same time, I think is going to really elevate our perspective on what this issue is and where we should be investing on mitigating, where we should actually be appreciating the change and not painting just this pure uh negative picture.
• 58:05
  SPEAKER_05: And as people learn to trust the outputs of the models, they're going to trust more of these claims that that people in this room and make are making and I've made in my career around the overall negative consequences.
• 58:14
  SPEAKER_05: And it won't be such a leap of faith to say, in 30 years, this may happen and we should invest trillions of dollars to prevent it.
• 58:21
  SPEAKER_05: It might be actually this probably is going to happen and we trust that because of all these data inputs, but also we're trusting sort of the guidance on what needs to be done at as low cost as possible to uh change that trajectory.
• 58:33
  SPEAKER_05: And so I think, you know, we'll just be able to make better decisions for long-term outcomes versus now where we're kind of relying on on, frankly, a fairly small number of scientists and a small number of actors to make really, really major decisions around the global economy.
• 58:46
  SPEAKER_05: And so I think it's been a challenging period, but it could become much less challenging.
• 58:52
  SPEAKER_03: Um Alan, did you want to add to that?
• 58:54
  SPEAKER_01: Well, I just also want to so this I run uh one of our programs is exclusively for women, like senior women in climate and nature.
• 59:00
  SPEAKER_01: And we had this very, not upset, but like quite emotional debate last week when we talked about uh the emissions of AI.
• 59:07
  SPEAKER_01: Because like data center emissions, like if you divide them in three, you've got the embodied emissions, the operational emissions, and then the enabled emissions.
• 59:14
  SPEAKER_01: So like data center operation running data centers 24-7, that's like the middle one, the operational emissions, what it takes to run these models and the data flowing 24-7.
• 59:23
  SPEAKER_01: And that has been a really big topic, right, in the US for a long time now, also spreading to Europe and the rest of the world.
• 59:28
  SPEAKER_01: It's super important.
• 59:29
  SPEAKER_01: And I am concerned about it.
• 59:31
  SPEAKER_01: And I think these companies, to your point, before they've like created carbon markets, they've also been incredible pioneers in energy efficiency around data centers and chip manufacturing.
• 59:42
  SPEAKER_01: Like, so that like they've like a lot of these technology companies have a history of like pioneering clean energy industries.
• 59:48
  SPEAKER_01: I think we'll get back to that.
• 59:49
  SPEAKER_01: I'm actually less worried about data centers today, and this is what got us, got me in trouble in my community last week, because there's new data now from MIT as loan talking about the enabled emissions.
• 59:59
  SPEAKER_01: So, like if the economy is increased by like, I think in modest scenarios, like AI will boost the economy for 25%.
• 1:00:06
  SPEAKER_01: So GDP globally will increase by 25%.
• 1:00:09
  SPEAKER_01: That's not necessarily green 25%, right?
• 1:00:12
  SPEAKER_01: That's brown 25%, most likely, because our economy is re-jiggered to premium consumption and fossil fuels and other unsustainable like nature loss is baked into the systems, et cetera.
• 1:00:23
  SPEAKER_01: That worries me actually more than the data centers.
• 1:00:25
  SPEAKER_01: I think we will fix the data centers.
• 1:00:27
  SPEAKER_01: We can build data centers that are like super positive for communities and watersheds and ecosystems around them with a little bit of public and regulatory pressure.
• 1:00:36
  SPEAKER_01: But how do we make sure that this superpower that we're embedding in the economy that pushes for a more green future?
• 1:00:45
  SPEAKER_01: That's really interesting.
• 1:00:46
  SPEAKER_01: But then so many people got really angry on the call and they're like, start like data centers a big issue.
• 1:00:50
  SPEAKER_01: So I think I don't know, this is a bit controversial.
• 1:00:52
  SPEAKER_01: But I think we should look at now and here and what we can see and feel, which would be data centers, and then think about like how do we collectively make sure that that's a positive 25%.
• 1:01:02
  SPEAKER_03: I think we're highlighting the fact that the world's too complicated for absolute rights and absolute wrongs anymore in almost every situation.
• 1:01:10
  SPEAKER_03: With with everything that's changing at pace, you know, there might have been um when we did this event in London, whilst we were on stage, the Prime Minister resigned.
• 1:01:18
  SPEAKER_03: So if anything massive has happened in the news, please let me know before before I go into this part of it.
• 1:01:23
  SPEAKER_03: But that there might have been three new frontier models announced while we've been sat here.
• 1:01:26
  SPEAKER_03: How do you keep up to speed with everything that's changed?
• 1:01:29
  SPEAKER_03: Because we are in this bubble, and I hear people go, I just can't, like, how do I even keep up to date with things?
• 1:01:34
  SPEAKER_03: So, what are your hints and tips for keeping up to date with everything AI and technology?
• 1:01:40
  SPEAKER_00: That's a hard question because I can't.
• 1:01:42
  SPEAKER_00: It's hard to give up.
• 1:01:44
  SPEAKER_03: Do you do you find that difficult?
• 1:01:46
  SPEAKER_00: Yeah.
• 1:01:47
  SPEAKER_00: Yeah.
• 1:01:47
  SPEAKER_00: Um I think it's hard because it becomes there's so much going on as well in the world, right?
• 1:01:54
  SPEAKER_00: So it's you you open the news, you open you know, the app, New York Times, whatever, Guardian, uh, BBC, and there's just so much.
• 1:02:02
  SPEAKER_00: So it's very hard to to keep track.
• 1:02:05
  SPEAKER_00: Um but what I do what I do do personally is I try things.
• 1:02:15
  SPEAKER_00: So if there's something that you know comes up and it looks interesting to me, I play around with it, no matter how smart.
• 1:02:22
  SPEAKER_03: There's no substitute for that, right?
• 1:02:23
  SPEAKER_03: You've got you've got to get hands-on on experience.
• 1:02:25
  SPEAKER_00: And so I I don't know uh everything.
• 1:02:29
  SPEAKER_00: I don't know every model or every tool out there, but there are tools that if I find it's very useful and it's interesting, I can build something really quickly, then I try it and do something with it.
• 1:02:40
  SPEAKER_01: I mean, the short answer is I don't.
• 1:02:42
  SPEAKER_01: I don't keep up, and but I'm quite comfortable with that.
• 1:02:45
  SPEAKER_01: I think there is magic in like white space and quiet and long walks and nature.
• 1:02:50
  SPEAKER_01: So, like taking in some stuff and then giving myself time to like reflect on that and let it mature in my brain, and then like something interesting will come out.
• 1:02:58
  SPEAKER_01: That's I'm more interested in that than like overloading and just feeling frantic.
• 1:03:03
  SPEAKER_03: I'm literally the exact opposite.
• 1:03:05
  SPEAKER_03: So um, we had uh a gentleman called Martin Berg.
• 1:03:08
  SPEAKER_03: I don't know if anyone knows him.
• 1:03:10
  SPEAKER_03: He's the chief executive of climate asset management in the UK, incredible organization.
• 1:03:14
  SPEAKER_03: He had the exact same answer as you.
• 1:03:15
  SPEAKER_03: He's like, it uh in fact, he took it even further.
• 1:03:17
  SPEAKER_03: He's like, it's just too much for me.
• 1:03:18
  SPEAKER_03: It's like I do not want the continuous noise that all the flow.
• 1:03:22
  SPEAKER_03: And I was like, oh, I I find that really reassuring.
• 1:03:25
  SPEAKER_03: So so I love the Moonshots podcast and the 20 BC podcast and um all in, which is more of a political angle on those things.
• 1:03:32
  SPEAKER_03: They are my go-to every week.
• 1:03:34
  SPEAKER_03: So um I watch those and a and a load of others ever on the gym in the morning when I when I work out, and and I love it.
• 1:03:40
  SPEAKER_03: I don't watch soap operas, I don't watch TV.
• 1:03:42
  SPEAKER_03: My interest and excitement is staying in the flow of all this type of stuff.
• 1:03:46
  SPEAKER_03: And I think loop background to young people coming to business, help them find the way that they want to learn.
• 1:03:51
  SPEAKER_03: Do they want to take the Anna approach where they cherry pick the best of the best and have space to let them think about it and ruminate on it?
• 1:03:57
  SPEAKER_03: Or like me, do they just want the mass flow of things and to pull out those ideas that are there?
• 1:04:02
  SPEAKER_03: Are you one of those spectrums or in the middle?
• 1:04:04
  SPEAKER_03: How do you think about keeping up to speed with things?
• 1:04:06
  SPEAKER_05: I mean, I definitely am more on your side of the spectrum, but I also think, you know, Anna has a very clear perspective on her role.
• 1:04:13
  SPEAKER_05: I think hers might be the better one.
• 1:04:15
  SPEAKER_05: Yeah, totally.
• 1:04:15
  SPEAKER_05: And you know, she's communicating really complex topics to large audiences and doing a really good job of it.
• 1:04:20
  SPEAKER_05: And and those large audiences also aren't on these tools every day.
• 1:04:24
  SPEAKER_05: So, like being able to connect to, you know, who you are at your core.
• 1:04:27
  SPEAKER_05: And, you know, I think we're both, I'm guessing, I don't know you that well, Rich, but you know, technologists at our core.
• 1:04:33
  SPEAKER_05: And I think it's exciting for me to keep up with it and I get really energized by that.
• 1:04:37
  SPEAKER_05: But um, but I also am a human being and I think, you know, taking a step back and creating space, and you know, I do, well, I don't do it as often as I like, but I have a goal to do, you know, Saturdays without a phone and without, you know, uh playing with any AI tools or anything.
• 1:04:51
  SPEAKER_05: Um, so I think that's part of it.
• 1:04:53
  SPEAKER_05: I also think surrounding yourself with a blend of people, uh, you know, there's somebody on our team, shout out to Jasper, he's iconic, but uh he's he's like in the mix.
• 1:05:02
  SPEAKER_05: If there's if those models have come out during this podcast, he's already been playing with them.
• 1:05:06
  SPEAKER_05: Uh and so he can translate some of that to me.
• 1:05:09
  SPEAKER_05: But I also have uh people in my life who have no idea what I do.
• 1:05:13
  SPEAKER_05: And, you know, we just go to shows together, play music or whatever.
• 1:05:17
  SPEAKER_05: And it's really important to have that balance.
• 1:05:19
  SPEAKER_05: And so I I don't know, I think I'm maybe somewhere in the middle in that sense.
• 1:05:22
  SPEAKER_03: Love it.
• 1:05:23
  SPEAKER_03: Um so I'm gonna finish off with a few rapid fire questions on this one.
• 1:05:26
  SPEAKER_03: So please keep as sharp and as short as you can on them.
• 1:05:29
  SPEAKER_03: Um if we're sat here in a year's time, what do you hope has changed in the world of AI and technology?
• 1:05:39
  SPEAKER_05: I hope that we can just trust it more and use it for the incredible intelligence offering that it is.
• 1:05:47
  SPEAKER_05: Right now it's processing, but I don't think it's anywhere close to intelligent yet.
• 1:05:52
  SPEAKER_05: Uh and there's a huge amount of human um translation that needs to happen and understanding, and it's frankly quite burdensome.
• 1:05:59
  unknown: Uh
• 1:06:00
  SPEAKER_05: And so we end up getting a lot of outputs that aren't helpful and just add more chaos and more noise and more information to a world that's already inundated with it.
• 1:06:07
  SPEAKER_05: And so I think it's actually kind of a exciting time, but I wouldn't say it's a it's a great time to be an AI.
• 1:06:13
  SPEAKER_05: Like I think there's a it's there's a lot of challenges with it.
• 1:06:15
  SPEAKER_05: And so I hope that it it becomes much closer to a teammate than uh sort of like a newsfeed, if that makes sense.
• 1:06:23
  SPEAKER_01: I love that.
• 1:06:24
  SPEAKER_01: That was really well, but more of a teammate than a newsfeed.
• 1:06:27
  SPEAKER_01: Uh I guess yes, yes, on trust, but I would like transparency to build trust.
• 1:06:33
  SPEAKER_00: I'm gonna borrow less noise.
• 1:06:36
  SPEAKER_00: I think there's just so much noise.
• 1:06:37
  SPEAKER_00: I I hope next year it's less noise on AI.
• 1:06:43
  SPEAKER_03: What would you want there to be more noise on?
• 1:06:45
  SPEAKER_00: I think yeah, it's more like again, borrowing his it's more trust, more less noise and more credibility, I guess.
• 1:06:54
  SPEAKER_03: Um so I did a keynote for um uh a group of CEOs in Europe last week.
• 1:07:01
  SPEAKER_03: And at the end of the the keynote, the presenter and I don't have the prop, sadly, picked up a big red old-fashioned dial-up phone for those of you in the room long enough to know.
• 1:07:10
  SPEAKER_03: And they went, right, you can phone anybody, dead or alive, to have one conversation to ask their advice.
• 1:07:16
  SPEAKER_03: I didn't know this question was coming.
• 1:07:17
  SPEAKER_03: Who would you call?
• 1:07:18
  SPEAKER_03: So this is one for my audience.
• 1:07:19
  SPEAKER_03: It's coming to each of you.
• 1:07:20
  SPEAKER_03: You can phone one person, dead or alive, for a conversation, advice.
• 1:07:24
  SPEAKER_03: Um, who would you call?
• 1:07:26
  SPEAKER_03: And I'd like a why as well, if it's an interesting answer.
• 1:07:30
  SPEAKER_05: I I should have prepped for this, but I think Abraham Lincoln comes to mind dealing with this period of incredible change and chaos, but ultimately getting back to what connects us all and what I'm not good on my American history.
• 1:07:44
  SPEAKER_03: Would he be confused about the concept of a phone call?
• 1:07:47
  SPEAKER_05: Oh, well, you said anybody, right?
• 1:07:49
  SPEAKER_03: Yeah, yeah, no, you guys take some explanation first.
• 1:07:51
  SPEAKER_03: They're like, don't freak out.
• 1:07:53
  SPEAKER_03: This is all fine.
• 1:07:56
  SPEAKER_05: That's there's an inverse of this.
• 1:07:57
  SPEAKER_05: There's a great YouTube video you should look up of uh some kids getting one of those rotary phones and just saying, like, call somebody and that they spend 10 minutes trying to figure it out and they keep thinking they get it and then it doesn't work.
• 1:08:08
  SPEAKER_05: It's so funny.
• 1:08:09
  SPEAKER_03: I've got these people in the team we've hired now that um uh younger than YouTube, which is just depressing as a concept.
• 1:08:17
  SPEAKER_03: Like forgive me.
• 1:08:20
  SPEAKER_03: Anna?
• 1:08:22
  SPEAKER_01: Um I I don't know.
• 1:08:25
  SPEAKER_01: I I would probably my kids go to a school called Ral Wallenberg.
• 1:08:28
  SPEAKER_01: It's it was a Swedish diplomat that uh unfortunately kind of vanished in the second part of the Second World War and was left to disappear and rot in a Russian prison, which is sad.
• 1:08:37
  SPEAKER_01: Sweden wasn't very brave at the time and got him out.
• 1:08:39
  SPEAKER_01: Um but he saved a bunch of people and he was really brave and he really thought about like kind of peace and community and collaboration.
• 1:08:47
  SPEAKER_01: And I would love to call him and be like, what are we gonna do here?
• 1:08:50
  SPEAKER_01: How are we gonna forge like closer collaboration between people who don't trust each other?
• 1:08:54
  SPEAKER_01: We've got such a polarized world.
• 1:08:56
  SPEAKER_01: I think that would be a cool, cool call to make.
• 1:08:59
  SPEAKER_00: I think it's David Attenborough.
• 1:09:01
  SPEAKER_00: He's still alive, right?
• 1:09:03
  SPEAKER_00: Yes.
• 1:09:03
  SPEAKER_00: He's 100.
• 1:09:04
  SPEAKER_00: 100 years old.
• 1:09:06
  SPEAKER_03: He would know.
• 1:09:06
  SPEAKER_03: Yeah, he'd be familiar with the city.
• 1:09:08
  SPEAKER_00: Okay, I can call him.
• 1:09:09
  SPEAKER_00: Um because I think we need more storytelling on climate and nature.
• 1:09:15
  SPEAKER_00: I think the science is there, the technology is there, but people still need I think people just need to know more about it.
• 1:09:25
  SPEAKER_00: Yeah.
• 1:09:26
  SPEAKER_03: Um so um my answer's changed as a result of this wonderful panel.
• 1:09:31
  SPEAKER_01: You got to practice too.
• 1:09:32
  SPEAKER_03: That's true.
• 1:09:33
  SPEAKER_03: I didn't at the time.
• 1:09:34
  SPEAKER_03: Um although I'll do one story on that.
• 1:09:35
  SPEAKER_03: On on our regular podcast, our final question is what's the one question I haven't asked you you wish I had?
• 1:09:40
  SPEAKER_03: It doesn't really work with a group setting.
• 1:09:43
  SPEAKER_03: Um the team, whenever we hit a milestone, so we're just about to our 200th episode.
• 1:09:47
  SPEAKER_03: For the hundredth, thank you.
• 1:09:49
  SPEAKER_03: For the hundredth episode, the team made me be the guest, and I'm being the guest again for the two hundredth.
• 1:09:54
  SPEAKER_03: And they got to the end and went, What's the one question?
• 1:09:56
  SPEAKER_03: And I forgot we did that when they made me the guest.
• 1:09:58
  SPEAKER_03: I was like, I don't I don't have an answer for this, so I'd forgotten to to prep for it on the spot.
• 1:10:01
  SPEAKER_03: When I was asked in front of this group of CEOs, my answer was actually Mark Zuckerberg, because that morning he'd just bought a big estate in Ireland, and he was the the person at the at the front of the frontier world who actually seemed to be talking about the holistic piece of nature and climate and water and environment.
• 1:10:20
  SPEAKER_03: And and I think for what it's worth, kind of what was alluded to here, whilst we are going to get the backlash of the backlash, I think to have the environment on the social license to operate, those hyperscalers and frontier models will become the biggest investors in nature and climate we've ever seen.
• 1:10:33
  SPEAKER_03: And they will bring an institutional level of capital.
• 1:10:36
  SPEAKER_03: They they can put$100 billion into nature and environment without blinking.
• 1:10:41
  SPEAKER_03: That's not even this year's intake into the OpenAI Foundation.
• 1:10:45
  SPEAKER_03: And it would change this sector a hundredfold overnight.
• 1:10:48
  SPEAKER_03: But listening to you all, my my actual answer to this question moving forward will be I'd call Alan Turin.
• 1:10:54
  SPEAKER_03: So for you geeks out there, Alan Turin is the man who broke the Enigma code and is one of the people who, because of his work, we're all here today in terms of technology and things.
• 1:11:03
  SPEAKER_03: And boy, did the British government treat him terribly.
• 1:11:05
  SPEAKER_03: So I'd ring and apologise and say thank you.
• 1:11:08
  SPEAKER_03: Um, speaking of saying thank you, could you all say a massive thank you to my incredible panel, please?