Thomas: 00:00
Hello, my name's Thomas, and you're listening to the Pensions in AI Podcast. This is the show where we explore the way that AI is being used across the pensions industry. In every episode, we talk to people in pensions who are working at the frontier of generative and agentic AI. You'll hear from superusers, builders, advisers, and big thinkers, the people who are shaping the future of how our industry could look and how it can deliver better outcomes for members. This is not a show for people to come and sell you things. It's far from it. We're here to share, to have open and honest conversations, to talk about what's really going on, what actually works, and the lessons that we're all learning along the way. And if we're doing a good job, then you'll leave every episode feeling more informed and confident with AI and better equipped to use the technology both for yourself and to shape how it's used in your organisation. In today's episode, we're going to be exploring three big topics: trustworthy AI, what it's like to build and ship an AI experience, and how AI can play a role in sustainable investing. And thank goodness then that we are joined by the brilliant Dr. Andrew Tucker to be the informed voice to bring all this stuff to life for us. Andrew, welcome to the pod. 

Andrew: 01:05
Gosh, what a build-up. Lovely to be here. Thank you. 

Thomas 01:10
What's the best introduction you've ever had somebody give for you?

Andrew: 01:14
Do you know what? My godson announced to his father on Father's Day that Andy was the best dad ever to his dad on Father's Day, and that was the best intro I ever had.

Thomas: 01:27
I think that's pretty hard to top, isn't it? My goodness. Um Andrew, for anyone that's not had the pleasure of coming across you or working with you before, tell us a bit about you. Who are you? What do you do? What's the stuff that you're working on? Thank you.

Andrew: 01:38
So I'm a reformed academic, which means I used to work at UCL and at Brunel. I work on the measurement of intangible assets and realised maybe 15 or so years ago that if you wanted to try to make a real impact in the world, one of the ways that you could do it is to figure out how those intangibles, like for example, sustainability, could be modelled and therefore measured and therefore managed in a much better way than I think, frankly, is happening at the moment. So, long story short, we've been trying to work out how you can use the new AI models, particularly conversational AI, but lots of machine learning as well, to understand vast, vast quantities of unstructured data that are out in the market that are being generated by all of us every day, and try to think of it as a massive cocktail party where we're trying to listen to everybody and flip between the different conversations and then ultimately pull it all together and go, right, so the way forward is this way. Actually, for a fintech, it's a lively market at the moment. And as I think, as we all know, the the the the the AI models and the sort of the evolution of AI is happening so fast at the moment that we almost need to take a breath, take a step back, understand, just go back to what it is we're meant to be doing, not being sort of wowed by the next latest woo-woo model that comes out, and just say, right, we're here, I'm here to do sustainable investing, to use AI to make that something that everybody can do, not just people in the City or hedge funds that have huge amounts of resources to do the research. And in that way, ultimately, we can actually build a more sustainable world. And that's ultimately what we're up to. All the technology is lovely, but you've got to focus on we're here to make a more sustainable world for for now and for our kids. And how are you doing that? So two ways, and I'm not I'm not claiming, you know, these are the only two ways, but I have found that two ways where you can really move the needle. Or think of it as a sort of, you know, we found a pivot spot where we can really make a difference. One is around sustainable pensions. So there's a huge amount of money uh in pensions, everyone gets that. If you are trying to build a more sustainable world, you're trying to build infrastructure, you're trying to address climate change, you're trying to and not just environmental things, you're trying to look towards social justice, you're trying to think about better governance, just a sustainability in the round. One of the pots of money you're going to have to move is the pension money in the market. And because there's so much of it, if you only move it even by a little bit, you'll be able to achieve huge things. The second one, which is not pensions, but I'll just throw it in there just very quickly, if you don't mind, is if you think about how sustainable investing actually works, sort of the mechanics of it, there are three major stakeholders. One is corporates. So if you look at the financial materiality of sustainability topics for corporates and allow them to see that clearly and quickly up and down their supply chain, they can actually move to a more sustainable business model quicker than sort of waiting for everybody to tell them what the answer is. Second one is institutional investors. And again, you'd be amazed. I was amazed, you go and talk to these institutional investors, super smart guys, lovely city buildings, uh, lots of steel and glass, and they think they're the smartest guys in the world. It turns out you say to them, how do you actually do sustainable investing? And honestly, it's a bunch of spreadsheets. So try to sort of think through again, how do you look at the financial materiality of your portfolio so you can optimise it so that your quarterly reviews you can say, right, these are more sustainable companies they're coming into the fund, these are less sustainable funds, the companies that are coming out of the fund. And then the third, the third one is uh assurance providers. So the PwCs, the KPMGs of the world, uh, those are the ones that kind of sign off on what is financially material or not for those annual reports, and being able to use AI to drive all three of those from what is at the moment, I'm not making this up, a very manual three to six month process. It's nine months out of date by the time you publish it, it's 12 months out of date by the time anyone does anything about it. It's just in terms of just financial in terms of investing, it's just a bit crap, to be honest. So we work, we have we've built a series of algorithms that then works through in a very well data-driven way, what is actually financial material for a company, for institutional investor fund for a uh assurance provider. And that allows just everybody to just kind of jump to the end and say, right, but now we know what's financially material for the sustainability topics, we can get on and do something about it. So again, generically, it's AI, it's machine learning, it's large language models, it's you know, being able to use all the tools that everybody uses these days, not complicated, they're not hugely sophisticated, just a different stack to get to an outcome that I was mentioning earlier. So, what how do you make the world more sustainable? One of the ways to do is you make sustainable investing as easy as possible to do.

Thomas: 07:06
And so it's not like you've taken the technology and tried to use it to solve a problem, like you've started with the problem first, it sounds like 100%. 100%.

Andrew: 07:14
I mean, I you know, I'm a I employ lots of very, very smart technologists. I'm not a technologist, I'm a data scientist. I I think it's bleeding obvious. You start with the problem and then you work out a solution. You don't say, here are all my clever solutions. Look, I've got this model and that machine learning, and da-da-da. What can I do with it? Which, no offence to our you know, tech pro overlords, seems to be the way that they're pushing out all their AI at the moment. Look, you here's this and here's that. It's like let's let's just start with what are we actually trying to achieve. And I mean, I think we we were saying just before we started here that one of the things I think is super interesting is that why would you bother with frontier models and getting killed on tokenisation when you can build a self-build a fairly easy tech stack and populate it with an open source LLM. It's 85-90% effective to do exactly the same thing. It's a hell of a lot cheaper. You're not tied into any of these tech pros and tokenisation, all that kind of stuff. And and you don't have to worry about next time Microsoft or or Anthropic or whomever changes their pricing model on a whim and suddenly your business model's gone out the window. So let's just you know, understand what we're doing here and use a technology to enable us and not not be led by the technology.

Thomas: 08:30
So um, could you bring to life for us? So I've seen a demo of Cameron. I think it's brilliant, I think it's super clever. How do you describe Cameron?

Andrew: 08:39
Cameron is a Uswitch for sustainable pensions, which uh it turns out probably is a copyright infringement. Anyway, what we are trying to do with Cameron is to make it as simple as possible to make your pension sustainable. So, if I may, just take one step back real quick. If you go and talk to Gen Z or millennials, and we've spent a long time doing this, and say, what do you think about your pension? And in fact, there was a thing on the BBC this morning, you probably saw it saying Gen Z don't think there's going to be a state pension in 30 years. You go and talk to them, you say, basically pensions are screwed. That's that's their take on it. And you say, Okay, so what do you think about investing for the future? Oh, well, if you talk about sustainability, yeah, yeah, I'm all for that. Because why not? It's rational self-interest, right? I would like the world to be around in 50 years' time. I'd quite like to be able to breathe the air and drink the water. Fair enough. So you say to them, would you like your pension, you know, your your I'm sorry, I'm focusing mostly on defined contribution, so workplace pensions, which obviously is is the first sort of time somebody like a Gen Z or millennial starts contributing to their pot. And you say to them, if that was invested in sustainable investments, would you like to would would you support that? Yes. Would you potentially increase your contribution? Yeah, potentially. I mean, look, cost of living, student loans, trying to get mortgage, it's not like I've got tons of spare money, but if I'm gonna spend money, if I have to spend money on a DC and if there's any kind of thing left over, if you can show me that it's gonna be invested in a more sustainable fund, fine, I'm in. If it's a generic, put it into the default, I'm not really gonna tell you what's there. I'm not gonna really tell you how to find out. But if if if you say match the match the digital experience to every other digital experience you've got, online shopping, online banking, you know, streaming. It's no good saying here's a fairly static app and it doesn't really tell you very much. What you really want to do is go, right, if I go onto ASOS, I'm gonna have a conversation with a with an avatar, and it's gonna show me what this suit or jacket or whatever is gonna look like on me, and it's gonna be interactive and the the selling experience is gonna be seamless. Now, pension's different, I get that completely. But so what sorry, back to the original question. So, what Cameron does is provide a conversation with a in effect a sustainability expert who meets you at your level. So if you if you don't know anything at all about sustainability, say, great, let's let's talk about sustainability. What do you care about? Do you care about environment stuff or do you care more about social stuff? Or maybe governance stuff. So you can start at that level. Or you can say, no, I'm I'm really into this stuff. I've done my research and I want to have a deeper conversation. I want to talk about scope one, scope two versus scope three emissions. Yeah, you can you can do that as well. What it's trying to do is understand your what you mean by sustainability. Now, I've been doing this for a long time, you you know, you've been doing this for a long time. I bet our definitions of sustainability would be different. And that's not a criticism, that's just I think entirely reasonable. Uh sustainability just means different things to different people at different times. Absolutely fine. Understanding that to then be able to match your preferences to the most appropriate funds to invest in is a really obvious option for a conversational AI. So I can build your profile. We have a chat, so Cameron, Cameron and you have a chat. It matches to where your level of interest and expertise is, and it comes out and says, right, you care mostly about environment, you care quite a lot about social, but you don't really care about governance. It then says, right, which pension are you, which pension fund are you currently with? And you can pick as many as you want. So obviously, you know, people have more than one pension fund. And I say, right, I'm with The People's Pension, and it says, right, here is the correlation between your profile and the default fund at The People's Pension. But actually, there's an ethical fund from The People's Pension, it's actually a slightly better match. So maybe you should think about switching. So this is where the Uswitch comes in. And I ran this past The People's Pension. I said, by the way, if you've got another pension, let's say you've got Scottish Widows, you can also do that, and they were horrified. I said, Well, what's the problem? It's like, you can't go around telling people that there's a better match to somebody else's pension. I said, Well, that isn't that a serious point. Really, and I say this with great respect to my my friends at The People's Pension, it's a really interesting point, which is for them, I completely understand that as a business model. But for the sort of you know, guy in the street, I don't really care whether it's The People's Pension; you know, I want to know where my money is best invested according to my preferences. And if that's somewhere else, then it's somewhere else. That's exactly what you get in in retail banking. It's what you get in you know, hotels, it's what you get in restaurants, it's get we get everywhere else in the market, so it's not a big surprise. So if that was a real interesting learning about when you're rolling out this kind of technology, you've got to be very cognizant of what the business models in the market are because you will be disruptive. And if that's what you want to do, you know, that's great. And don't get me wrong, I do want to be disruptive. I think Cameron should be disruptive. It should be a B2C app play where you have hundreds of thousands, hopefully millions of people building their own sustainability profiles and matching their DC funds to their profiles and not the other way around.

Thomas: 14:03
How do you curate the conversations that members have with Cameron? So when we talk about conversational AI, it's almost as if sometimes we we have this idea that a conversation is one thing, but we all have many different conversations within our lives. Some of them are good, some of them are bad, some of the conversations you can't wait to have again, some of them are conversations you hope you never have again. You sound like you're speaking from experiences, but how did you kind of shape and curate that conversation? I assume it's not just a large language model running loose, right? Do you steer it?

Andrew: 14:33
Do you? So it'd be quite fun if it was? It'd be absolute chaos. No, we were very clear that we cannot just take an LLM off the shelf and go, there you go, knock yourself out. So we took, we actually ended up testing, oh my god, so so many of them. And we realised that the most efficient way of doing this, instead of we were originally going to just build it ourselves from scratch, we got an 18-month Innovate UK grant. Thank you, Innovate UK. Other funding streams are available, but though not many these days, to build uh Cameron. And we ultimately realised that the cheapest and most effective way of doing this was taking one of the off-the-shelf LLMs from what two years ago now, and then training it on about 1.3 billion conversations. So, what we've been up to for the last five or six years is building a massive database of sustainability conversations in 15 languages, using the the well-known taxonomies of sustainability. So, not to get into the alphabet soup bit, but there are there are a couple that are really well known and and kind of the industry, everybody in the industry uses them. So let's not reinvent the wheel. And working up those conversations as a training data set, which hilariously, given how many we have, was still not enough. So we ended up having to gather a lot more. So that I mean, these things are so hungry for training data.

Thomas: 15:58
And how did you know that it wasn't enough?

Andrew: 16:00
Oh, sorry. So there we worked with a very good data science team at Wyser, which is another, so we it was a it was a partnership, and they are fantastic at training LLMs. They work for the the Ministry of Justice, the Ministry of Defence and the Home Office, and a lot of uh housing authorities building those LLMs that are taking the load off the incoming conversation uh phone calls and and and user requests. So imagine if you're doing Ministry of Justice, right? You've got to be absolutely on the money. Any slip is a lawsuit. So that's why we went with them. They are absolutely top-notch at training these models. And we spent a lot, frankly, just spent an awful lot of time tweaking, tweaking it, tweaking and tweaking. And it's all about optimisation. It's like a massive, I don't know if you've seen these things, but it's like a massive, you know, in those recording studios and you have those sound desks and you have all these sort of knobs go up and down, up and down. It's basically a massive version of that. And you're looking to optimise, in effect, training data against recall, as in, you know, does if I if I say I'm really interested in air quality, for example, does it recognise that you're talking about air quality and talk to you about air quality or does it do something else? And also then what's the accuracy of what it's talking about? And there is a again, these things get a lot better, a lot quicker, but it turns out that we spent best part of nine months, I'd say, tweaking a model to get it right. And then we spent the best part of nine months building a very simple interface. That you know, that's not the hard bit, and saying actually to a bunch of users, we went to about 250 users in the end, and kind of looked over their shoulders while they were interacting with Cameron, which was a really interesting experience for two reasons. One is people are an awful lot more broad than you think they are, if I can put it that way. So we were thinking you would have a relatively discrete – as in a, you know, limited conversation about sustainability. There's environment, there's social and governance. We talk about those three things, we build your profile, done. And it turns out, of course, that people, when they talk about sustainability, they talk about everything. They talk about, you know, my granny that she got flooded last year. And how does that relate to climate change? Or they talk about I mean, you know, governance is a massive and really important part of sustainability, and that includes everything from you know business ethics to management of the supply chain. I mean, there's the spectrum of sustainability is enormous. So that's why we needed so much training data, because we realised we just didn't have the breadth, we we hadn't built the LLM broad enough to meet the breadth of what people think about in terms of sustainability. So that's why it took so long and so much effort. But but it ultimately ended up with being able to say to to look over people's shoulder. And there was an aha moment, which is always amazing when you're doing customer experience and customer research, where they where they see their profile for the very first time. Oh, I I didn't know that. I mean, I didn't know that I care more about environment than social. I thought it was the other way around. But you know, you and of course there's a there's a you there's a historical thread, you can see why how you answer that, why we got to that. And then there was a really interesting piece where we said, well, are you like your friend Sam? So let's put you in a tribe, we called it. So, you know, are you more of an environment person or more of a social person or more of a governance person or a balanced person? You can start to pull the pull the sort of uh dissect the data to to to build these profiles, which obviously you know, targeted support is is is a absolute direction of travel for this. But almost more importantly, it was that moment of recognition where you could give a name to yourself. I'm in that tribe as opposed to that tribe. And that's that was a for me, that was a super interesting, almost sort of socio-political moment where people want to understand kind of what they are. It's not that let me do this better, it's not that they're trying to put a label on themselves, but I do want to know what us is and what them is, and particularly when you're dealing with quite an abstract thing like sustainability, it's important to understand kind of where you are on that spectrum. So being in a tribe, brilliant. And then the second piece is you then match to we basically we run a very simple algorithmic correlation to your profile versus the profile of all of the defaults and all of the alternative funds in the typical DC pensions market, which again was was super interesting. All this I'm slightly scratching my head saying, why hasn't somebody done this before? You know, when you have a a good idea and you sort of do it and you think, something, what something's a bit weird here. What is it? What is it? Why hasn't somebody done this? It's so obvious. Why hasn't somebody done it before? And the answer is partly I have an incredibly high threshold for boring manual data entry tasks, as it turns out. But there were just a there was just hours and hours and hours of working through lots of fund fact sheets and making sense of what that meant in terms of what was the profile of that fund, which led to two very interesting insights. One is, gosh, some of the alternatives to the defaults are really similar to the defaults in terms of sustainability. Second one is there's a much there there was a lot of a lot of the funds, a lot of the sort of very standard major market funds are quite good on environment. But wow, there's a lot of differentials between social exposure and governance exposure. And then when you look at the kind of people, I was thinking, yeah, what would happen if kind of everybody ended up in the same fund? And the answer is absolutely not. There is a much broader variety of profiles and preferences out there compared to the number of funds available. And I completely understand you can't have hundreds of hundreds of funds. That makes no economic sense. But certainly there is in some of the major pension providers, there are far, far too few options. According to the kind of preferences that people would like to invest in.

Thomas: 22:07
And so I found that when I used Cameron and I was put into the justice crew. And when I was looking at how my default compared to some other defaults in the market, I I almost felt uh some of them are slightly better, but there's nothing out there that seems like a no-brainer. Yep. Like this this this is an example of a fund that's knocking it out of the park and is truly and deeply aligned to me. It almost felt like I was choosing from my list of the like the least worst alternative in that.

Andrew: 22:35
Yeah. Yeah. Well, I'll tell you what's interesting is if I may, so I've then started working in more of the I think it's called the sort of the wealth part of the market, which I find quite irritating. But anyway. What the title wealth. Yeah. Just like well, as opposed to. I'd like to be in the pauper part of the market. But sort of SIPPs and personal investing and uh IFAs. And the variety there is sort of the other side, right? So it is there's way, way too many options. And actually the IFAs, frankly, have no idea how to work their way around that market. So again, a really good use of AI and this and Cameron is to say, okay, we could just literally put in all whatever it is now, 350, 400 sustainable. Sorry, I'm now doing the air quotes. I just saw the flaw in that. Sustainable. So 350, 400 sustainable funds. How the bloody hell are you going to figure out which one you should be in? Yeah. So the IFA doesn't know, just doesn't have the capacity. Anyway, so we now started, it's but what's been great is we've now started working with some IFA platforms and saying here is a really good way of being able to match up your potential individual, you know, paying for your advice. And I get all of that, and I'll come back to that in a second. But being able to match an individual, being able to match Thomas to that fund over there that you've never heard of, you never would have heard of. And even if you'd done all the research world, you probably still would never have found it. But using the a nice bit of AI, you can say, right, actually, that's a 99% match to your actual uh um preferences profile.

Thomas: 24:08
It's a really big deal, I think, to make an investment decision or to move your money based on something that AI is saying to you. I think it speaks a lot to whether or not people feel that they can trust what they're interacting with. How have you thought about and tried to build trust into Cameron?

Andrew: 24:24
Brilliant. So if I may, as I said I I used to be an academic, so I used to work on trust. That was my doctoral work and my postdoctoral work. And we were trying to model how trust actually works, both interpersonal and particularly with state organisations. This was back in the days, sort of date me. I had originally had the idea when I when I heard Tony Blair stand up and say, You can trust me, you can trust me. I'm a straightforward kind of guy. I think everyone went, really? Not sure that aged well. But what we so we've always had this idea that we need to build trustworthy AI, which is a terribly generic way of doing it. And we come at it from a point, a very sort of human-centric point of view. So if you go and talk to the the the IBM engineers, right? So they they did a load of work on this about 10 years ago. So we're going to build trustworthy AI, and they built tons and tons of what a software engineer with great respect will call trustworthy. So do does my token, does my half a token match your half of the token? Do I have the right credentials to be able to access this bit of data versus that bit of data? That's not trust. That's not how humans think about trust. They they say, you know, does it do what it says on the tin? You look at Andrew. Is he is he sort of a straightforward kind of guy or is he kind of a bit shady? Okay, that it's not humans make that kind of decision 10 times a day. Do we share a set of values? Do you care about what I care about? Not saying I can't trust you if we don't, but you know, that's a that's a trust signal. Is there a continuity over time of your experience working with me and vice versa? So, you know, it used to be great, but now he's really fallen off and it's gone really sketchy. This hypothetical, just stop laughing. And that's how people think about trust. And they think about it in more sort of jargonly terms, like abilities, beliefs, and consistency, or A, B, and C. And it depends on what situation you're in, right? So we are have a nice good chat in this in this environment. And actually, at this point, ability trust, does it do what it says on the tin? Does Andrew know what he's talking about? Is the is the dominant trust driver? Values, the beliefs piece, is important but not that important. Whereas consistency, does it work over time, is not important at all because this is a one-off, unless you invite me back. I'm sure you don't at this point. So that's how people actually trust. So roll that forward into what trustworthy AI looks like. You do two things. One is you can you can use a series of proxies for for how that works, but you can also do, and this is what sort of the beauty of digital technology, you can you can build a series of metrics that look at does my trust in the interaction with Cameron evolve over time? So, do my conversations show that I am very sceptical at the beginning, but then sort of grow into it and I give longer and more complete answers at the end? Do I offer very bland statements and don't really engage in the conversation? Or do I sort of lean in metaphorically speaking and and start to you know really pick on individual bits and start to really drill into those? And you can, I was about to say metricise a terrible Americanism, but you can metricise that, you turn those into metrics, and you can start to judge how much, in fact, I trust Cameron to then give me that. Don't get me wrong, even if there was zero trust, you could still talk to Cameron and get your profile and get your tribe and get your best fund. That would absolutely work. But what's sitting behind this is we're trying to also understand, particularly for the pension providers that we're working with, can we can we put some metrics to around how trustworthy this kind of interaction is? And therefore, and of course, what comes out of that is if there is a a higher level of trust, then there's more likely to be repetition, there's more likely to be greater engagement, there's more likely to be a willingness to accept further interactions and potential suggestions, and ultimately, you know, commercially, there's frankly a propensity to invest more money in this. So trust is a very, when I say tangible measuring intangible assets, it's incredibly tangible once you understand how humans trust rather than how a machine trusts. Does that make sense?

Thomas: 28:46
Yeah, it does. I think you're the first person I've heard speak so kind of deeply and carefully about trust at the point at which AI and pensions meet, because there, I mean, there are so many events across our industry exploring this stuff. Trust doesn't come up all too often. Neither does sustainability, actually. Sustainability is rarely the headline topic, it's rarely the topic of a presentation. But it's one of the first questions that audience members always ask.

Andrew: 29:12
You've got to meet people where they're at. I mean, somebody said this to me the other day, and I thought in some ways the statement is such bleedingly obvious. But then when you look at how large organisations, and I and I, you know, the the pensions industry is in some ways completely typical of large organisations. So when we were doing our trust work 10, 15 years ago, and and I I should say not I don't know whether this is useful or not, but there's there's I'm I'm sort of mentally referring to a bunch of academic papers in my head, which I'm not gonna sort of if we could put those in the show notes. Yeah, we'll put that also in the I think large institutions massively struggle with the concept of trust. Partly it's a scale thing. So if a so a very large organisation to a single individual is just a sort of scaling problem, as it were. Individuals know how to trust individuals. Individuals know how to trust smaller groups of friends. They're very, very good at it. It's part of fundamental human 101. So then when a large organisation comes along and and acts in the in a, I don't want to say distrustworthy kind of way, actually the correct jargon is sort of mistrustworthy. So it's it's sort of, they just get it wrong. It's not like they're miles off. They're not like they're being a psychopath and being completely untrustworthy. It's they're just not being not matching their trustworthiness to meet the moment. So remember I said a few minutes ago, so so in some in some situations, and I take your point, absolutely, in some situations, trust is a it's kind of a superficial thing. It's just, you know, can we can we rub along shoulders? We're only together for the next you know 20, 30 minutes. Doesn't need to be anything particular, just you know, make sure we don't end up throwing each other out the window. Fine. Slightly extreme, and probably these windows don't open, so you're okay. But in other situations, like uh investing, you know, major investing. I'm I'm a Gen Z, I haven't got a huge, you know, haven't got a big pot at all. I'm just starting out on this journey. I want to make sure I'm not cocking it up right from the very beginning. That is a huge trust moment. And being able to meet the level of trust required at that moment is absolutely vital. And I, if I may say so, and again, with great respect, I don't think the pension industry does a great job of this. I'm not saying they're the only ones, but they do seem to stick out a little bit as just not fully thought through how this works. Now, to give you some historical context, banking 10 years ago, in absolutely the same spot. Bankers, poor reputation, industry, terrible reputation. They understood partly by what's interesting, there was a supply and demand, if I may say so. So demand was they were starting to haemorrhage customers. I'm I'm thinking mostly UK retail at this point. The experience is poor, people looking to go elsewhere, and then suddenly you get things like Monzo and Revolut and all these digital app-based things. And I don't know how much you excuse me, pay attention to these things, but there's a really interesting market shift that's going on. So the demand was kind of always there. The supply then came on stream. And if I may extend the analogy a little bit, I'm starting to see some really interesting pension apps around. Digital first, you know, meets meet the customer where he or she is, understand what they care about kind of almost before they know it themselves, be very interactive, but also being really transparent. So I have no, I mean I have no problem investing my money. I don't want to lose money, but I'm very happy to invest my money in something I believe in. And, you know, I'm a reasonably smart guy. I think most people are reasonably smart and savvy about the things. Investing isn't particularly complicated. You know, there's some things you like on principle, something you don't like on principle. Try and invest in things you like. Try not to invest in things that are obviously got a terrible business model, but you know, beyond that, it's not that complicated. What I find frustrating is that the the pensions industry is very, very smart at investing. It really is. It's been doing it for a long time. It's actually got a very good track record, has some off years, but you know, overall it does a the investment teams at the the big pension funds, I think, are decent. They know their books inside and out. Why are they not making that kind of information available to the individual members whose money it is to say this is what we're doing with your money? And one of the potentials of AI and large large data sets is you can have, and I'm you're starting to see it now in some of these pension apps, you can have individualised communications that say, you, Thomas, believe in environmental things, and here are the five companies that are in your portfolio in your default DC pension fund that in the last you know quarter they've done these things, and that has been good for their environmental credentials. And in effect, well done you, right? Brilliant affirmation of your sustainable investing. Everybody likes to be told they're doing a great job. The data's there, the technology's there. It's not a big leap. And it's that back to that supply and demand thing. I think the demand is there. I think it's now been really, really clear the demand is there. Survey after survey after survey says Gen Z and millennials, really care about sustainability. You know, nine in ten people want to invest more sustainably. Two in ten are allowed to invest more sustainably because you know it's such a hard thing to do at the moment. What an easy job. What, sorry, what an easy win to be able to say to individual members, these are the things that we invested your money in that meet your preferences and is making the world more sustainable according to you. That's a that's not revolutionary at all, I don't think.

Thomas: 34:50
And I think that's what's great about Cameron is it gives people a trusted way to navigate that, as opposed to I get a letter from my pension scheme, I upload it into ChatGPT, and I go, tell me what I'm invested in. Like, what's the alternative? What should I put my money into instead? And ChatGPT goes, Well, do you want A, B, or C? And based on, you know, the predicting the thing that comes next, before you know it, you're making an investment decision led by a large language model. And the consequences of that, yeah. I mean, I'm all for things that can help people achieve better outcomes. I think lots of people will use ChatGPT and Gemini and Claude and achieve better outcomes, but they may find themselves in a bad place and not realise that until years and years down the line because those decisions just compound.

Andrew: 35:39
It's a thing that if I may I find so extraordinarily frustrating is this seeming chasm between the people that can afford to pay for advice and the nine in ten people that can't. And they are so obviously creating an unequal investment environment, which you know, as a political scientist originally, that's what I have to say I found quite scary. So I I lived in the States for a few years, a hugely unequal society. A lot of the very strange behaviours that go on is is to do with that massive differential between the the very poor and the very wealthy. And you can start to see that in this country now. And it feels like not not just this, not just IFAs, not blaming the world's ills on IFAs, but it's ingraining that kind of inequality over generations that is gonna be incredibly hard to reverse, and the and the social and democratic, if I may say so, implications of that are massive. And actually, it's one of the things that I think if there is a opportunity again. Sorry, again, I maybe this is the right podcast for this, but again, why don't you use AI to try to do something about that? It scares the pants off IFAs. We're not we're not gonna code an IFA and just put it into a large language model, okay? I get that. No, nobody's nobody's actually suggesting that, not even our wildest dreams, but there is so much knowledge that could be made more publicly available to a lot more people to allow them to make better decisions without undermining the business model of IFAs and allowing them to then do what they want to do. And they also always say every time it's a meeting, I want to do a more value add process, I don't want to spend time filling out going through questionnaires and ticking boxes, and you know, there's so much that's such an obvious way to use AI to do the heavy lift of that kind of basic knowledge transfer. And of course, with AI, you can do it at scale. So why isn't that kind of stuff available to everyone? I mean, almost like it's a public utility. I mean, sorry to sound like some crazy socialist at this point, but I yeah, why isn't why aren't we using the AI to actually just upskill, up knowledge? That may or may not be a word, an entire generation of people coming into the workforce. Because the like I say, I think the long-term social consequences of that will be incredibly beneficial.

Thomas: 38:18
This is something that I think about quite deeply, which is that you know the the disruption of what comes next in our industry may very well come from outside of it. So if we see the work that OpenAI is doing to train their models to become better at answering medical questions, to become better at helping you with your finances. So we've just got ChatGPT personal finance out in the US. Connect all of your things into one place with open banking and then let a large language model that's been tuned to do better at answering money questions, give you something that is not financial advice, it says, but has all the feelings and substance of advice and research is is really starting to point towards when people get advice from large language models in all different walks of life, people are acting on that advice even in high-stakes situations.

Andrew: 39:06
Of course, why wouldn't you? I mean, that's that's that's a very human reaction to I I would prefer certainty to uncertainty. I would prefer you know, more generically, I would prefer knowledge to ignorance, I would prefer at least having an idea about what the future holds than sort of chaos. And I it frustrates me because you speak to individual execs in the pensions industry and they get this. This is this is not new stuff, and these are smart people and they are grown-ups, and I say that in the most you know uh uh amicable way. They they they uh they have kids, they are part of their local communities, you know, these are real people. And yet there seems to be some structural issues, particularly in this industry, and I've worked in in in other industries, and you know, I've I've the closest analogy, surprisingly, is the pharmaceutical industry in my in my head. I used to do a lot of work in pharmaceuticals, but uh in this area in this area, and again, super smart people uh clued up, uh you know, cultured, yet seemingly caught in an institutional bind where everybody realises you're getting suboptimal outcomes, but nobody seems to be able to do anything about it. And if I may read that across to pensions, it does feel like, and I see a lot of this, you know, everybody reckons a suboptimal outcome is that people don't have enough money to retire on. That's a big deal, right? You've kind of got one job. You've got to get this stuff right. Why are we not doing a better job of this? And it there's it's almost a what's the right way of doing this? I'm not saying it's a fear of change, I'm saying there's there's a a a reticence of knowing that you're gonna be held accountable if you get this wrong. And that's where I think it really boils down to it. Now, you know, I'm I'm a I'm a fintech person. I've I've been running my own business for 10-15 years, but that's what I do every day. I'm comfortable with that. Not that comfortable, comfortable enough with that. I can live with it. A lot of big, a lot of corporate a lot of execs in large corporations are very, very scared of that kind of accountability. And I'm not just but you know, you see it in the public sector as well, all the time. But unless we get a little bit more comfortable with taking a little bit more risk, and sometimes it will go wrong. Yes, it will, it will. But sometimes it'll go a lot more right. And actually, unless we try something different, we are just going to rinse and repeat. And at the moment, we are heading towards people retiring, a lot of people retiring, and not being able to support themselves. And you know, if you I'm just about old enough to remember what it was like when we brought in auto-enrolment, and that was brought in for a really bloody good reason, which is pensioner poverty is horrendous. I don't want to go through that. I don't want my parents to go through it, I don't want my kids to go through that. That that was a great government intervention, and I think it was Steve Webb that that brought that in, who was a Lib Dem who I used to work with. So maybe there's sort of a slight bias there. But anyway, that's the kind of thing that we need to we need to avoid. And it feels like we're starting to slip back towards that, and we can't just wait for the government to come wave its magic wand and fix things. Okay, that's just not going to happen. Look, look, you know, look at the whole defence situation at the moment. There is no government magic wand. The industry is there to make sure this outcome is optimised and not suboptimal, and we need to take some risks.

Thomas: 42:49
If someone wants to try Cameron, get involved and kind of help you with the mission that you're on, what's the best way for them to do that?

Andrew: 42:55
Brilliant. So askcameron.co.uk. Go there, nice big button that says try Cameron. You can use it two o'clock in the morning, two o'clock in the afternoon, in the bath, whatever, whatever. Go go and have a chat with Cameron. It will build, like we said earlier, build your profile. It's it's not, I want to emphasise, I know it's sometimes sustainability conversations can be quite scary. This is not a, you know, it's not a test.

Thomas: 43:19
It wasn't scary at all. Brilliant. The brilliant thing I thought about the conversation is that it made me realise I had strong opinions on some things and no opinion on others. Yeah. But when I was presented with a question, I should go, I was able to go, I don't know. Tell me why I should care. Like in the nicest way, like sell it to me. Brilliant. And I will hear from you and I can decide whether to trust you or not, and then I can form my own opinion. It's a nice thing to gently steered that way.

Andrew: 43:43
I think I'm I'm really pleased to say that. So there's no right or wrong answer. It's your sustainability profile, it's not somebody else's. And that is incredibly hard to figure out by yourself. So this is a really good use of technology as an empowering tool. It can be quite fun. I mean, maybe I'm really, really nerdy about that. It's interesting to know which tribe you're in and who else is in that kind of tribe. And we're starting to gather actually, there are slightly more people in the environment tribe than the social tribe, as it turns out. But don't just don't don't write off the governance tribe. You think, oh, that's you know, really boring and irrelevant. No, no, not at all. Lots of very smart people that I know are much more governance than anyone else, particularly sort of financial folks, because they get that governance is a massive driver of value. And then it's interesting to say, well, look, I've I mean, I don't know, you I think I've got five pensions. Five. And I plugged all of those in, and you get completely different results. So it is, it's a it's super interesting to realise that when you sign up for a workplace pension, I don't know, I'm sure you've done the same thing. It's not exactly an interrogation about how you which defaults you go into, right? You you you know, high, medium, low risk, some kind of geographical spread. Do you care about equities versus bonds? If you're lucky, there you go, done. That's it. And then 30 years later, I mean that's nonsense, right? So it was super interesting to realise that the one of my pensions was was very good. It really hit my profile. I'm sure it's a complete coincidence. One of my other pensions was miles off. So I'm immediately thinking, well, let's bring the one that's miles off into the one that's much closer. And there's a, if I may, there's a really obvious so what that goes with it. That was the thing. It's lovely doing all this clever technology and la la la la la. But unless there's a bloody obvious so what at the end of it, you're not gonna get a lot of people doing anything. So we were super keen to say, we'll suggest to you, this is not advice, we will suggest to you, guide you towards what is a more what are the funds that are a better fit for you. But then here's a big button, I mean a bit like Uswitch, right? Here's a big button that says, right, I'm gonna switch. And it was interesting because people want to take agency of their of their money. People aren't passive. That's it's really it's a real flaw in the system of assuming that people are very passive investors. They're not actually, especially if you can make it interesting and engaging, tell them actually what they're investing and and why, people will be much more engaged. So, yes, there's a big red button that says you can switch from here to here, but also why? Right? So, what are the things why is this fund better? Well, because it really matches your profile. Why? Because the top 10 holdings are things that actually they they they are producing business models and business business outcomes that actually meet what you want to happen in the real world. That's incredibly empowering. And I and there are a couple of people that come away from Cameron and go, yeah, okay, whatever. But honestly, nine in ten and change go: not only do I feel better about myself, I feel better about what I'm doing about the world. And yeah, if you're if you're building something that does those things, you're in the right space.

Thomas: 47:00
I'm not just saying this because we're talking together, but if I were to move my pension now, I would think differently about it, having used Cameron compared to where I was before. Interesting then to reflect on pensions dashboards, yeah. Very shortly going to be showing everybody all of their pension pots that they've got. And the question many people are gonna have is what do I do next? Yes, we'll kind of say, oh, I'll consolidate consolidate your pots with with us, and oh no, consolidate your pots with us. And people need reasons to do that, and like what a fantastic opportunity to target those people and say, hey, I think dashboards are the things you can think about to do it.

Andrew: 47:36
Have a thing that was it's really it's really if I may say so. I think it's I think it's a great time to be in the market for for this kind of thing. It's it's been yes, it's really interesting. Pensions, so much happens in pensions and and yet everything remains the same. But the I think particularly targeted support, I think particularly the dashboard, and I think particularly to your point about people starting to use AI sort of to build by themselves, as it were, I I think are three almost structural things that are going to change the market, whether you like it or not, and and that's good. And I think you're gonna start seeing it's really interesting. I'm starting to see sort of what you expect to see. Some of the large pension funds are starting to try to reposition themselves as brands, they're advertising, they're you know, having you know rebranding exercises and trying to be a little bit more consumer focused and and consumer you know driven. That's completely understandable. Get that, good, more of that, please. But they're playing catch-up. I I think I think I mean again, I I I do talk about Gen Z and millennials a lot because those are the people that we are targeting, but although uh in no way am I suggesting that anyone that's not in those categories doesn't care about sustainability. It's actually, I mean, that's palpably, palpably wrong, but I just happen to be focusing on one end of the market because they're they're more sort of digitally native. But it does feel to me like these tools are becoming available, these structural changes are happening, you are seeing disruptors coming into the market. It's it's suddenly gone from a relatively, relatively calm industry to really quite lively, which is ironic. But I also think it you've seen this in lots of different industries in the past. Two things happen at that point one is you have people that seem to catch the zeitgeist and they you know streak ahead. And the second one is that the industry goes, the industry either consolidates massively and you have two or three real winners and everybody else falls away, or you have a mess. I am not in a position to tell which way it's gonna go. I thought you might be asking me next, but I think the literally the one thing you can't do is do nothing. You can't just go, it's gonna be the same as it ever was. It it clearly isn't. I would say those those people that are embracing right very beginning of a conversation, those people have a clear idea of what they're trying to do and then use technology to do it, great. You've got you've got a good chance. Those people that just grab the next technology that comes along and sort of throw it out there and try to pretend they know what they're doing, no, disaster. And then absolutely there's people that say, no, no, head in the sand, it's it's fine. We'll be there, we'll definitely be the losers. Which makes it which makes it an interesting time to be in the market. I think it also, to your point earlier, I think also means it's a it's a market that the barriers to entry are coming down quite fast. And I think that you're gonna see a new incumbents, sorry, you're gonna see incumbents under a lot of pressure. You're gonna see some really interesting new people coming in. Let me give you a interesting example, which is I'm getting phoned up by private equity companies quite a lot now saying we've we're interested in this market because it is a classic disruptor market. It's very it's the technology is poor, there's a lot of stasis, there's a lot of money that could could move if we if we came up with the right solution, and there's an ability to bring in learnings and particularly consumer consumer-facing experiences from elsewhere into this new market. Which is great. Which is sorry, great as a consumer, a little bit scary as a pension provider, possibly, but I'd say a great opportunity for people to really, you know, make a name for themselves.

Thomas: 51:37
I don't know the demographic data on how kind of different demographics worry about sustainability and in particular the sustainability of AI, but it's a constant story in the news and in the media at the moment. Have you found that's changed the way or the likelihood of kind of Gen Z and millennials engaging with Cameron?

Andrew: 51:58
Yeah, that's that's a good question. There's two bits to the answer. One is do people care about the sustainable impacts of them running massive queries? I mean, everyone gets that. Yes. They do. We have built it, we've built Cameron in a way that it is pretty light touch. There's some clever caching that goes on. There's a lot of quite smart consolidation of data points that we've done on purpose. And I'm sure we can do a lot more. Which which has sort of led to me that maybe led me to a probably minority position where I don't think it matters that much. I understand you know, data centres and water usage and all that people care about that. That's interesting because that's now coming up as part of the sustainability conversation, but there's a a disconnect, and I don't think an unreasonable disconnect from what I've just said of an individual using an individual LLM on a sporadic basis to do one thing and then stop. I mean, yeah, we're not talking about you're on you're running OpenAI for eight hours a day. It's not it's not that at all. It's 15 minutes. What, once every couple of months, maybe? You know, that's nothing, frankly.

Thomas: 53:16
And I think it's not binary, right? It's not it's not a position of you must reject AI wholeheartedly or adopt it absolutely as an individual. And actually, if you're able to use AI in a way that can help you have a greater sustainability impact because you then decide to move your money and change the way that you're invested, actually that strikes me as a super powerful thing.

Andrew: 53:37
It's a good it's a good use of the AI, yeah. I mean, you're not you're not pissing around making pictures of cats to you know put slop on LinkedIn. No offence to LinkedIn. I think we've all seen that. It's not that. It's it's there is a there is a use case for it. I tell you what, I tell you what, if I may say so, and this is this is gonna be a massive generalisation, but I I think people are getting a bit more targeted about when they use AI and what they use it for. What's really interesting, like all these things, is that the people that are so I have a I have a sixth former uh daughter and I have a first-year university daughter, and they are both highly sceptical of AI on a on a methodological point. So, yeah, I don't want to, I don't I can't use AI at university to write my essays. I can use it to generate, I can ideate, but I'm I can't use it to actually write things. Absolutely right, says one of them. And the other one says, actually, she's very much into maths, and you use AI and you'll get complete crap. Because it's just wrong. Maths, I mean it's a large language model, it's not a large numerical model, it's just really poor at maths, kind of by design, right? So people, particularly people that are exposed to this stuff, are figuring out the use case much much faster, much easier than I think maybe some of us slightly older people have thought through. But then that's the that's an exposure thing. It's not complicated. There's some things that AI is great for and use that, and there's some things it's shit for, and don't use it for that. Duh.

Thomas: 55:07
And I think that's a perfect place for us to wrap this one up and bring it to a close. Andrew, thank you so much for your time. I really appreciate it. Yeah, Andrew's laughing because he's he's dropped – he's dropped bombs, and I've chosen that as the moment to bring our – but I think you're right. I agree with you. I like making helping people understand, helping people to make their own decisions about how, when, and why they adopt AI in their lives, and making sure that that is meaningful, values aligned, human first. It's just so incredibly important. And I wonder to what extent we sometimes it feels like there is the inevitable march of AI, which is all the things will be automated, all the jobs will be gone, we'll all be living on universal basic income, or maybe we won't and there will be no alternative. And it's quite a it's quite a bleak, seductive and easy science fiction vision to buy into, and yet like it's one that's never felt more real or more plausible. I think us all having informed opinions, being users, helping to shape the system rather than being passive within the system. It's just so incredibly important.

Andrew: 56:21
Yeah. The active passive dichotomy is is huge. And it's and don't underestimate something too philosophical about it. Don't underestimate people. They they they do get this. They are not willing to just sit there and let somebody else tell them what the answer is, particularly if they're called Elon. And actually, I'll tell you what, the the the best the the best recruiter for sustainability and Cameron is the current president of the United States. He has done more for marketing sustainable investing than anyone previously, and I thank him greatly for it. I think he's done a brilliant job, and the next two years will be, I think, a great time for sustainable investing because it is you can put your money where your mouth is and stick it to him. And God, there are a lot of people that want to do that, and it's really effective.

Thomas: 57:12
Andrew, thank you so much for coming on the podcast. It's been a pleasure to have you. Thank you.