Talking Credit Unions from Swoboda Research Centre
Talking Credit Unions delves into the latest trends shaping the financial landscape, equipping credit unions with the insights they need to stay ahead in a rapidly evolving industry. This podcast also tackles key issues facing the movement, fostering knowledge sharing and collaboration. Featuring a diverse line-up of credit union leaders, experts, researchers, and industry suppliers, it offers valuable perspectives to drive best practices, innovation, and success.
Talking Credit Unions is brought to you by the Swoboda Research Centre, which works to support positive change and transformation in the credit union movement. (https://swobodacentre.org/).
The podcast’s current producer and host, Anca Voinea, is a trained journalist with over 13 years of experience in the co-operative and credit union sectors. To suggest a podcast topic or for any other inquiries, contact her at anca.voinea@swobodacentre.org.
Talking Credit Unions from Swoboda Research Centre
EDITION 35: How credit unions can navigate the AI revolution – Top tips from the experts
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Artificial intelligence (AI) is rapidly transforming financial services, with 75% of firms in Great Britain and 54% in Ireland already adopting the technology. For credit unions, this presents both opportunities and challenges as they seek to harness AI while staying true to their co-operative values and member-centric mission.
In this podcast episode, Anca explores the role of AI in the credit union sector with four expert guests. David Fagleman, Co-founder and Director of Enryo Consulting, shares insights from his recent Swoboda Research Centre paper on AI. With a background in banking policy and innovation, David emphasises the need for inclusive and fair access to financial services as the sector evolves. He argues that credit unions can take the lead on this due to their member-centric approach.
Joining from Ireland, Micheál O'Shea and Ciara Prendeville of Metamo discuss their work supporting 16 of Ireland’s largest credit unions. Micheál explains how Metamo's "sprinkling of AI" approach augments rather than replaces human processes. Ciara, Metamo’s Deputy CEO, brings nearly two decades of leadership experience and focuses on strategic innovation and member value across product and service development. She explores some of the challenges faced when implementing AI solutions, offering advice on how to address these.
The episode also features Patrick Heaphy, CEO of Youghal Credit Union in Ireland, which serves over 17,000 members. Since taking on the CEO role in 2020, Patrick has overseen major growth and digital transformation, including launching mortgages and current accounts. With a background in accounting and risk management, he also serves on the board of the Swoboda Research Centre.
For further insights, listeners can read the Swoboda Research Centre's recent paper on AI. Swoboda members can also take part in an AI webinar in May.
Talking Credit Unions is a regular podcast dedicated to informing credit union practitioners, leaders, and opinion formers on a variety of industry topics. The podcast is sponsored by the Swoboda Research Centre.
Introduction to AI in Credit Unions
Speaker 1This podcast is brought to you by the Swoboda Research Centre, the not-for-profit independent think tank serving credit union leaders in Ireland and Great Britain. At Swoboda, we're all about sparking positive change and transformation from action-focused research and inspiring annual conferences to tailored development programs for CEOs and chairs and, of course, these podcasts. We're here to help you power credit unions forward. To learn more, visit swobodacenterorg. That's swobodacenterorg. We hope you enjoy this episode of Talking Credit Unions.
Speaker 2Welcome to a new episode of the Talking Credit Unions podcast. I'm Anca Voina, your new host, and I'm incredibly excited to take over this project. Before I dive into today's topic, I'd like to thank Chris Smith for building this podcast into such a valuable platform for the credit union community and for his support as I step into this role. As a journalist covering credit unions for over 13 years, I've reported widely on their impact across the UK and Ireland, as well as the challenges they face, from growing competition to the urgent need for digital transformation. Today's episode focuses on artificial intelligence, which is reshaping the financial services sector. In Great Britain, 75% of financial services firms are already using AI, compared to 54% in Ireland.
Speaker 2So where do credit unions fit into this fast-moving landscape? Are they leveraging AI to improve their products and services, or do they see it as a threat to their cooperative values and member-focused approach? To explore these questions, I spoke with several industry experts and credit union leaders. First, I talked to David Weigelman, co-founder and director of Enrio Consulting, a firm that helps financial services providers navigate change. David is the author of the Swoboda Research Centre's latest paper on AI, which examines the landscape, challenges and opportunities for credit unions. David, thanks for joining me today. I'm keen to hear your insights on the use of AI within the credit union sector. How has the AI landscape evolved in recent years and what are the implications for credit unions?
Speaker 3Artificial intelligence as an idea goes back to well. Alan Turing, famously, was one of the first people to think about how a machine could learn and solve problems, and that's been the basis of AI over the past 60, 70 years, where it has increased in its ability, has come along with increased computing power, and right now, in 2025, we have extremely powerful computers and that means that the AI that people are talking about is extremely powerful. So everyone's talking about AI, but really it's been used as machine learning for the past three, four decades by financial services firms in their back office processes. So what we're seeing now is the more advanced version of what's really been part of financial services for a number of years.
Speaker 2What's your opinion on the findings of the recent Bank of England report on the use of AI within the financial services sector?
Speaker 3They're talking about a range of firms. Credit unions aren't included here but it gives a good indication of where the industry as a whole is going. Three quarters of firms surveyed said that they are using some form of AI in their operations, and that's increased from just over half in the past two years. That seems like a big number. It is a big number, but the increase does match the increase in AI application and the ability to apply it over the past few years. Where they're using that ranges, but really its predominant use by these large firms is on the operational side.
Speaker 3So IT functions, legal functions, human resources, automating customer support, mitigating external risks, security measures that's where it's predominantly being used. Some are applying it to assessing credit risk, but they do have plans to continue using it and applying it over the next few years. Its true impact on the financial services firms will probably be boring. It'll probably be back office function. Ai will be the new spreadsheet. That's how we need to look at it and not be too intimidated by it, because that's where it could have the most impact.
Speaker 2So where do credit unions stand when it comes to AI awareness and applications?
Speaker 3The credit union sector that I spoke to which was Saboda members so it was the Republic of Ireland and Great Britain was a really good cross-section of an industry to look at, because you have large credit unions and you have small credit unions and you've got all of them in between.
Speaker 3So, as usual with technology, there's no one size fits all solution, but I did speak to credit unions at different end of the scale.
Speaker 3Generally, there's a very high awareness of AI from the media, and large and small credit unions have had discussions at a senior level about how they might implement some use of AI.
AI Landscape and Financial Services
Speaker 3Now that does vary from some larger credit unions thinking about using it to improve member facing services to other credit unions who are thinking about training a staff member so they can utilize chat, gpt or some other form of AI to support their marketing and website development or introducing some sort of internal process. But there is a ethical concern that credit unions have which is understandable and expected, considering that credit unions are member-first organizations around the risks that exist with using technology and particular AI, which is that, because it's data-driven, it could get things wrong and that could have severe consequences for the members' financial situations and there could be security concerns as well, but one other concern that's worth mentioning is that they understand that if they were to implement an AI process, they would need to have the right expertise to do so, and they would need to have the right governance structure to ensure that it's being introduced in a safe and responsible way. That's very encouraging. There's intrigue, and some organizations are starting to dip their toe in how they may be able to benefit from it.
Speaker 2Did any of the credit unions you spoke with have AI policies or strategies in place already?
Speaker 3None that I spoke to had a specific AI strategy in place, but AI is the next step of digitalization. I've said that before and I think it's important to get that through, not to be intimidated by it. So if credit unions have a digitalization strategy which I'm sure many of them do, this is just another part of that. So I would hope that the excitement around AI could be an opportunity to revisit that strategy and see if there's a way of kickstarting it or improving it in some way.
Speaker 2What were some of the main barriers to AI adoption that they mentioned?
Speaker 3I think there is a natural hesitancy around something new, which is totally understandable and is a good way to approach this new technology. There is a lack of resource issue that they're stretched anyway, and training someone to understand this technology would be a burden on their time and could reduce employee time that could be spent on member services. An investment in technology if credit unions want to use AI to improve the more kind of high risk applications, such as automating or improving credit lending decision process, that will need investment in new technology, and there are software providers out there, some of which are specifically tailored to ethical and community finance organizations. So there's an investment issue which is a barrier, but using AI doesn't necessarily need huge investment, but it does need some and that's a bit of a barrier. There are these ethical concerns as well, which are totally understandable and would require very strong stakeholder engagement, internal and external, if a credit union was to implement an AI process.
Speaker 2Assuming that the credit union wanted to embark on an AI journey and had the resources required to do so, what would be the main areas they should be looking at in terms of operational applications of AI or member-facing applications?
Speaker 3Well on the operational side, which is where I think it could have the most impact, and in the report we go through these, and there are lots of examples in the report of some credit unions that are doing this. But one of the main ways is fraud detection and member protection, and that's a good example of a process that's already in place that could be improved by new technology, which is AI. There are lots of ways that AI can improve employee productivity. As I said before, there is this view of AI just being the new spreadsheet, and there's lots of productivity tools that can make a big, big difference in how employees conduct their day, and then more time can be spent on what credit unions do extremely well, which is that member-facing human service. I actually had a call with someone the other day and it was a Zoom call and they were using AI that took notes of the meeting and then automatically updated the CRM system. I thought that was quite a neat way of saving time.
Speaker 3That's also a good example of where that process would be in place, but someone would need to check that as well.
Speaker 3You couldn't just rely on AI to get that right.
Credit Unions' Awareness and Ethical Concerns
Speaker 3It would need to be checked, and then there's a huge benefit, I think, that it could bring to the leadership teams of credit unions who, from my interviews, they're very stretched and there could be a lot of time that could be freed up from their perspective, that could be spent more on strategy and the kind of bigger picture things that they should be really focusing on. In terms of member-facing applications, some credit unions really want to compete with other providers of financial services and if they want to do that, ai can really help improve their interaction with members, so it can improve their support services, chatbots that can be used to provide 24-7 access, improving the app and that whole customer experience. As I said, there's also how it could automate credit and lending decision-making processes and then generally, just how it can deliver more personalized and better suited products, and that's where I think that credit unions could have a really unique and important contribution to AI, which is combining what they know about their community with data and providing the most suitable decision on the product necessary for that person.
Speaker 2I know that in your paper, you also looked at some of the risks associated with the use of AI. What would be your advice in that respect? What should credit unions be aware of and how could they address some of these risks?
Speaker 3There's generally a big risk around data, so the right type of data needs to be used in order for AI if it's being used in a decision making capacity to work correctly.
Speaker 3There are some providers software providers who are specialising community finance organisations and they can use software that can support credit unions to make the right decisions.
Speaker 3But one of the problems that exists in data processes and also specifically with these machine learning AI tools, is that they can have a bias in the decision making process if they're built on rotten data or corrupt data, I think is the correct term. So it's very important to ensure that you're using the right type of data and that you have the right processes in place. It's a transparent process of something goes wrong. You know where that has gone wrong and I think that credit unions are in a good position to understand where something goes wrong, because they will have a better idea of the customer or the member they are trying to provide. There are, of course, security and privacy concerns For using more data. You've got to have improved security processes. If a credit union was to introduce some of these processes, there would need to be good engagement, both internally and externally, to ensure that both staff and members are fully aware of what's going on and don't feel that they've been unfairly pushed into a technology that they might not be that happy with.
Speaker 2You mentioned the issue of cost. How much does scale matter when it comes to implementing AI technology?
Speaker 3I think if you were looking at introducing an AI-driven credit lending decision-making platform. I don't think that comes cheap, but it could potentially produce some fantastic outcomes, one of which could be better decisions made for members, but also it could really reduce staff time needed on some things that could be spent more on the face to face aspect of credit unions. So it may seem as though you're removing the human element to it, but actually you could be increasing it because you're working with the AI to produce better outcomes for the members. But then again, if you just want to test something out and you're thinking of using AI to look at marketing or to look at employee productivity, I don't think they're that expensive and it's worth talking to them. There's lots of examples in the report of organizations that are doing this type of thing, so I think it would be good to reach out to them and see what they're doing.
Speaker 2Have there been any recent AI regulatory developments in the UK and Ireland that credit unions should be aware of?
Speaker 3If you'd asked me that question six months ago, I'd say it was quite clear what the UK and the EU's approach to AI regulation was. Since the Trump presidency, there has been a big push towards financial deregulation and we don't touch on this in the paper because it was published before this was becoming clear. But whereas it was quite clear what the two regulatory environments were doing, I'd say that that might change slightly, as it seems as though there has been this new, rejuvenated push towards becoming the country that's delivering the best AI, or the most AI. So, while there are protections in place, particularly at an EU level, which is a much more proactive regulatory approach to AI, there's already been some conversations around diluting that a little bit. So I do think it's important to stay on top of what's happening on a regulatory side.
Speaker 2Before we wrap up, what final piece of advice would you give to credit unions in Britain and Ireland that are considering adopting AI?
Speaker 3If you're curious about something, give it a go and have a look at it. If you haven't tried ChatGPT or Google DeepMind or one of those applications, try it. Give it a go and see what you think and start having that conversation around. What is your AI strategy and how can that be combined with your digitalization strategy? So that's number one. The second is collaborate with other credit unions. See what they're doing. Maybe there's an opportunity to collaborate on investment or share learnings on what's happening. Everyone is testing this out. No one really knows where it's going to go, so it's great to be in touch with other people in your sector. And the third is how could credit unions be the sector that adopts AI ethically and be a leader in it? There's all of this talk around how great AI is and how great everything is. What I know, and what a lot of Swoboda members will also know, is that using digital processes really exacerbates a lot of inequalities that exist in financial services and people's financial lives.
Speaker 3There is a risk that AI will make that a lot worse, but there's a big opportunity for credit unions, using the knowledge they have about the community and their members, to combine that with this new technology and produce a really, really great outcome. So explore how it could be done ethically is my third one. Oh, and if I could give a fourth one, read the report.
Speaker 2To get a better understanding of how credit unions in Ireland are using AI, I arranged an interview with Keira Prenville, Deputy CEO, and Michal Oshie, the Chief Technology Officer at Metamo, which is a 50-50 joint venture between 16 Irish credit unions and Fexco, one of Ireland's largest financial services companies. Before diving into AI, I asked him to tell me more about Metamo.
Key AI Applications for Credit Unions
Speaker 5I guess. At Metamo, we're driven by an ethos of collaboration with our credit union partners and they're very much committed to a brighter future for themselves, for their employees, their members and their communities in which they serve. We know that the general public have a great sense of affinity with credit unions and the levels of trust remain very high. However, we know that we must ensure credit unions continue to be relevant into the future, and what we have seen recently is that the credit union personal loans market share has declined from 61% in 2020 to 53% today, and you have to wonder is this down to competing with neobanks and fintechs, who are, I suppose, really ahead in redefining that convenience and that experience?
Speaker 5We at Metamo have a number of products and services designed to support the evolution of the credit union business model. One of the areas that's most relevant to today's podcast is we've also designed and integrated a best-in-market end-to-end AI solution to automate loan applications processing and approvals. What that does is really important because it enables credit unions to play in the 24-7 on-demand lending market. We have seen a significant challenge for credit unions around creating a differentiator in a digitally disrupted financial services market, and what we are doing in Metimo is supporting them to solve for this. That's where we're at in Metimo, and we're deeply conscious that technology is not meant to replace that human touch, but it is there to enhance it so that credit unions can continue to be relevant, especially to that younger demographic.
Speaker 2What impact has the use of AI had so far?
Speaker 4I like to describe our use of AI in Metamo as a sprinkling of AI.
Speaker 4We don't use it exclusively for anything, but we use it to augment and complement several business processes.
Speaker 4But we use it to augment and complement several business processes. We have built a lending automation platform called TLS, and part of the process of approving and quickly approving loan applications is we use AI in that flow and how we use it is we take an extract of credit unions' data the historical loan book data and we apply some AI logic there to basically tell loan officers what are the chances of the loan being approved based on what you know, based on historical data. The other very successful use of AI that we have is we've partnered with a very young, innovative company that have developed a chatbot AI powered chatbot specifically for the credit unions, and this chatbot has now been deployed in quite a number of our own credit unions and has resulted in quite a number of reduction and efficiencies and people's queries talking to the chatbot. On top of that, then, what the ai engine can do is it can summarize all the chat interactions and present credit union staff with a detailed summarization of the daily or weekly or even real-time basis of what are members' pain points why are?
Speaker 4they talking to the chatbot, are they having trouble logging in, are they having trouble with their pins, or are they having trouble applying for a loan, or are they trying to inquire about products and services? And it enables accreditudians, I guess, to quickly respond to users' interaction via the chatbot. We're also looking at expanding the use of AI A number of pilots underway. One of the things that some of our credit unions would like to solve for is call summarization. So quite often, as Ciara mentioned, there are credit unions in Ireland that are community-focused.
Speaker 4A lot of people still like to pick up the phone. However, in the past, those phone interactions were almost in isolation, whereas now, with the use of text-to-speech or speech-to-text technologies and AI, those calls can be summarized, similarly aggregated in using AI, and credit unions could get a summarization of the weekly calls and see well what are members calling in about, and it enables credit unions to quickly respond to users' requests.
Speaker 2And what has been the feedback you've received from the credit unions using?
Speaker 4it. The feedback has been very good. You mentioned there earlier about efficiencies and I suppose one of the things we're seeing is like a cost reduction, time reduction in terms of if we look at our TLS platform, for example, the time to decision is now a matter of minutes. Augmented again back to the sprinkling of AI. Augmented with AI. It's important that you can't just drop a loan application into ChatGPT, for example. This is part of a workflow. This is complementing the loan officers. In terms of using chatbots, some of our credit unions have reported 30 to 50% more member interaction. The chatbot is available 24-7. It can speak multiple languages. Increasingly, credit unions have a diverse member base and some of them English may not be their first language Personalization. It can recommend loan products based on what people can feed into it, and it typically tends to resolve maybe up to 70 or 80 percent of customer queries without human intervention.
Speaker 2Were there any specific challenges you faced when implementing AI solutions? I'm thinking of, for example, potential biases or issues of that sort and how did you overcome those?
Speaker 4Irish credit unions face several unique challenges when implementing AI. As you know, they're shaped by their community focus. We have in Europe, and indeed in the UK, a tight regulatory environment and resource constraints. Ai engines typically like data. They like clean, structured data and lots of it. Many credit union core systems and credit unions both in the UK and Ireland would rely on data that has been sitting there for a long time in systems and maybe it's siloed, it's inconsistent, the address field is in the wrong column, and so on and so on, and it takes a bit of time for the credit union to cleanse their data, to make their data fit for purpose, regulatory and compliance risks. As I mentioned, I guess we have a number of overseers in the Irish sector. We have the Central Bank of Ireland. They hold Irish credit unions to strict transparency standards. Indeed AI black box decisions would, I'm sure, raise red flags with the Central Bank. We have our GDPR regulation in Europe and something similar in the UK, and we have the upcoming AI Act that's coming out in August.
Speaker 4You mentioned about bias. One of the challenges we faced, for example, when we rolled out TLS, was that we noticed and indeed it is a fact that gender is a factor. When you look at the delinquency rates and the paybacks and then the historical loan book, typically females tend to pay back loans better than males, and that is a fact. But gender is not something that we can use in an AI decision engine. So that was a field, for example, that we had to take out. You can't introduce bias, for example, based on gender, so we have to be very careful in how you deploy AI in those situations. The other thing we have to be very conscious of is what's called explainable AI, or XAI as it's known.
Speaker 4In the TLS decision engine, we don't decline a loan. We either approve it or we refer to lenders. So we make sure that there's a human, that there's no bias, that there's no automatic decline of loan. But if there was and if somebody did get refused for a loan, they may be able to come into the credit union or indeed go to the one of the governing bodies, like the central bank, and say I would like to know why my loan was refused.
Metamo's AI Implementation Experiences
Speaker 4And credit unions have to understand that it's simply not a defense to say, for example, chat gpt told me not to give anchor a loan. You have to have what's known as explainable ai and you have to have documented and proven methods to say, well, the loan was not given for these reasons. That's why we have the the human in the loop type construct. If on our system that a load gets approved, that's well and good. But if it doesn't, we introduce what we call the human in the loop. So I guess that's some of good, but if it doesn't, we introduce what we call the human end of it. So I guess that's some of the challenges. There will be some other challenges, ira. I guess you might be able to touch on cultural resistance and staff challenges perhaps.
Speaker 5Thanks Michal, thanks Anca. I suppose in terms of some of the challenges, look, any transformational journey is difficult in terms of that cultural journey for credit unions and that is why I suppose we work very much holistically. In Metamo we have the people support element on the HR business partner model and that is supporting credit union CEOs, management teams to ensure that they have the right organisational structure and embedding the why with organisations and credit unions as well is really important. So, explaining why it is that we need to go down this route in embracing technology and looking at AI and it's all about, I guess, embracing it in that prudent and controlled manner. So for us it's explaining and getting the buy-in from staff that consumers not just members so we're talking about the wider landscape of consumers and not just credit union members that they now expect that seamless, fast, personalised digital experience and modernising the digital offering it's no longer optional.
Speaker 5Experience and modernising the digital offering it's no longer optional, it's actually a strategic necessity for growth and competition is everywhere and we want credit unions to be at the forefront to really disrupt the financial services sector.
Speaker 5And, in particular, if we think about the millennials and Gen Z, they expect instant gratification and that is definitely more so than any previous generation. It's largely driven by technology and that digital culture. So, if we think about our smartphone streaming on-demand services, what we can do through our TLS or our total lending solution is make sure that credit unions can play in that 24-7 on-demand lending market. And that is at the heart of our technology solutions in Metamo, but also at the heart of our people solutions and our HR solutions in Metamo as well, because we are passionate about the fact that that outdated digital presence can make the credit union appear out of touch to those who want to engage online. So really, we look at it in a holistic view on how we support our credit unions in rolling this out. You know how they organise themselves and what that means. You know in terms of structure, in terms of competencies, in terms of skill set, and we support them on that journey.
Speaker 2And if each of you had to offer advice to other credit unions looking to implement AI solutions, what would that advice be from? Obviously, the HR perspective as well as the technical perspective.
Speaker 4So I suppose, if a credit union is looking at deploying, AI be it in the Irish or the UK market.
Speaker 4My advice would be to start with some clear goals. I would identify one high-impact, low-complexity case. Let's start small. For example, the AI chatbot that Metamo could indeed help any credit union with is pretty easy to roll out and it will give a credit union a lot of experience for the first three to six months in seeing what AI can bring to their credit union and from there then they can leverage and they could look at implementing, say, tls or other upcoming AI enhanced products. Once you get into rolling something out, I guess on a deeper level, cleansing your data, audit your data, be cognizant that there is a law called GDPR and there is another one coming called the.
Speaker 2AI.
Speaker 4Act. I'm pretty sure there'll probably be something similar in the UK. Avoid using AI with fragmented, unstructured data. It doesn't do well. Use AI, as I said, in small use cases and sometimes credit unions can fall victim to. Sure we can do it ourselves, but leverage partnerships for me would be key. Metamo are in the Irish market. I'm pretty sure there's. Do it ourselves, but leverage partnerships for me would be key. Metamo are in the Irish market. I'm pretty sure there's something similar in the UK market. Make sure you touch base. Indeed, when we were developing our TLS, we ran it by the central bank and we got their blessing. And bring people on the journey. Bring your members on the journey. Tell your members via newsletters or marketing campaigns that we will have an AI power chatbot or we will have an AI component in our new lending solution. Do a pilot. We found pilots useful for credit unions. We ran some internal pilots for us just to get people comfortable with it. Ciara, you probably wanted to speak, maybe, about upskilling teams and internal stuff.
Speaker 5Yeah, I think it's really important that the capability to move with speed and agility is harnessed, and that is critical. So that means that having the right HOR people strategy in place is key. The need to invest in a holistic people strategy which effectively accomplishes successful change should really be at the heart of every credit union strategic plan. Michal mentioned around sharing best practices in leveraging that collective strength and scalability, because the alternative is being a pioneer-like orientation of figuring things out for ourselves. So really, I think, what are the skills of the future?
Speaker 5Looking at your existing organizational structure, shifting the mindset from today to tomorrow, and what does the credit union of the future look like? And how do we build on the competencies and develop staff to enable us to deliver on the credit union of the future? What does that look like? And really doing your analysis of the gap and then putting a plan in place to work and to bridge that gap and to drive forward with the strategic ambitions of the credit union sector. Michal and I are very passionate about the transformational journey that credit unions are on. I think that we want to shape the future together. We are very open to collaboration with credit unions and welcome more partnerships. So, for any of your listeners. Please do feel free to reach out to us and we'd be delighted to have a conversation and explore any synergies or partnerships.
Speaker 2Metamo is an example of how credit unions can collaborate to adopt AI. That theme of collaboration also came up in my conversation with Patrick Heafey, the Chief Executive of Yorle Credit Union in Ireland. Patrick is a steering group member of the Credit Union Strategic Alliance and a director of the Swoboda Research Centre. This is what he had to say about his credit union's approach to AI.
Speaker 6Primarily we were using it for back office items like pulling together a terms of reference or getting best practice or understanding. But in the last six months we've moved on to our lending platform and we've brought in a piece of software called Visualize here in Ireland and what that does? It does all the open banking for us. It does the financial analysis, all AI generated. At the moment we're running them side by side with our existing lending that we do manually and what we're doing is trying to compare and contrast. Are they getting the same result for approving members automatically that we would do manually? And of course, there's differences. So we're working on those differences and it might be that the settings were set up for a certain limit or they misunderstood things. So there's a lot of kinks to it.
Challenges in AI Adoption for Credit Unions
Speaker 6It's not as quick as we thought, but we're working through it and the whole idea is we want to get to a position where the results we're getting in the automated solution is mimicking what we're doing manually. I don't think we're in a position to go straight to automated once that happens, because what we do at the moment is it lands in our approval page and it says do you? It either says approve or it either says refer. It either says approve or it either says refer. So all our staff review it all at the moment. But in the future, once we can trust it more, we will just go straight to approve. Or refer means we'd have to review areas. But there's other credit unions in Ireland who are on the same software, on a different journey.
Speaker 2They're at it for three years and now they're gone, fully automated for certain loan type. Is AI then?
Speaker 6part of your strategic planning. In our strategic plan we do have an AI piece in us, because we're looking at AI from a whole suite of products, but the problem we have is we are stuck with our vendors, so quite a lot of the time we can't do as much AI as we like because maybe the vendors we have don't have it developed for certain areas. I'm a member of the Credit Union Strategic Alliance here in Ireland, which is a group of, at the moment, 20, soon to be 30 credit unions that collaborate and one of the key areas is AI. So we're looking at standardizing and automating things like BoardPack, and we're looking at things like KRIs, for the key risk indicators to be automated. We've already engaged transaction monitoring, a bit of automation and AI in that regard, so we're starting to bring it in more. It's becoming the norm in terms of conversation, in terms of development, but I still think we are a little bit behind where we need to be.
Speaker 2Are there any other key areas you would like to use AI in?
Speaker 6A key area and it was in the paper from Sobota is personalisation.
Speaker 6The credit unions are amazing at doing personalisation in person, but what we need to do is to be able to translate that to our online as well, and I think our online have been quite generic.
Speaker 6They've been quite out of the box. I think now, with the actual reduction in cost of AI, you know, we can really bring personalization to that level. And the one thing that's driving, let's say, the credit union strategic alliance is that we're looking at AI and automation and we're saying we're relatively medium to small size credit unions. But you're looking at competition versus big bank, for example. Ai allows us kind of almost even the playing field to a certain level, because up to now they have. I think we heard recently they had something like 300 or 400 in one of the banks in Ireland working just in AML, and we're sitting here saying, you know they might have 100 or 200 or whatever the number is, but we're kind of looking at that, saying, well, if we work smarter, you know we can utilise technology to mimic what they're doing, and I think that's going to really help credit unions in the background.
Speaker 2Are some aspects of the technology available to smaller credit unions that may not have the funding required to invest in product innovation, but they might use it for process innovations.
Speaker 6Yeah, I'd agree. I actually agree. I don't think it's there for every credit union, but that's where the likes of the collaboration really sets it apart. I suppose that's where the CUSA that I'm involved in the Credit Union Strategic Alliance, that's where it was formed. It's because y'all might be able to afford some level of technology, but other credit unions were smaller wouldn't. But together we will get access to far better technology and we've even seen it in an initial working group. We've done. We've engaged with a vendor who's not in the credit union market because they're too big, but as a collective we're big enough for them to be interested. So we've started talking with them and just bringing them in alone. When we split the costs out, it's the same that we're currently paying, but we're going to get far better value and far better compliance, governance, ai, development going forward, because there's all great about AI, but then you have to think of the likes of the ISOs and the DORA and things like that. So it's to cover both ends and I think credit unions traditionally would focus on DORA and compliance and regulatory and not enough on development.
Speaker 6To me, the key product development that credit unions need to do is the personalized product to really use AI to offer members what they want, how they want it, and the technology is there. We just need the scale to adapt it. Collaboration allows everybody benefit and we look to the US and we look to Canada and where they're doing this on a bigger scale and you can see that it works for credit unions of all size. If you go to a small credit union, they have everything. You go to a large credit union, they have the same products, whereas I know in Ireland, and it's probably in the UK, there's difference in product levels depending on the size of the credit union. And I know in Ireland the central bank has intentionally done that, where they've said you needed to be over 100 million in assets to get access to certain levels of lending or a current account. So there is an issue there. But collaboration is the only way we'd be able to do that and even from simple things marketing is a great example.
Speaker 6We've used AI to kind of help us generate ideas, so we put into various AI solutions and we'd say you know we're looking for an idea on X, y and Z, or how do you attract young members from 18 to 24 in rural Ireland, and a lot of the ideas, you know, but there's always a few nuggets. And even something I use quite a lot is an app called Napkin and the napkinai, and what that does is it takes text and turns it into visuals. So when we're doing training for the staff, instead of having a load of text it turns it into a visual and actually it's much easier to consume. The other area we use it a lot is for our SOPs, our standard operating procedures.
Speaker 6We have a piece of software called Scribe and what Scribe does? It mirrors what you're doing. So, for example, if it was logging into the central bank portal, you log in, scribe is mimicking it and then it creates a SOC document with the screenshots, with everything, brings it out into a Word document. You can edit it if you want and it's done and that's the SOC created. So it's areas like that we're using at LODES, but it's more in the member space is where it needs to go to.
Speaker 2How have members responded to the introduction of AI technology, and how have staff members responded as well?
Speaker 6The staff were probably more sceptical than the members. I think the members are more used to it because they do it in the banks, they do it with insurance companies. We're probably one of the few traditional businesses still on the high street. You know. A lot of them have gone online. The staff members, members fairness to them they've taken to it because they know this is the future and what we've tried to do it is a change management journey internally and what we've done is taking them on a journey from.
Youghal Credit Union's AI Journey
Speaker 6This isn't to replace your jobs. This is to take the existing kind of manual labor that you hate doing. Automate that and so you can focus on the relationship management. You can focus on dealing with the members and dealing with the more complicated loans, for instance. So, where our members really like it, we brought in two different kind of technologies recently. One was, as I mentioned, visualize, so when a member applies for a loan, they don't need to upload bank statements. It's a simple open banking API it does fully automated. We have it on our screen almost instantly whether it's to be approved or not. We haven't utilized it properly. As I said, we're working them in tandem with our manual process, but the members are already loving the simple, open banking platform Another one we did.
Speaker 6We had an issue with reactivating accounts, so our policy was always you have to come in here and reactivate. But we have people who live in the UK, who live in Australia, who wanted to reactivate their accounts. So we set up a solution with an Irish company called IDPal and what it is. It does the facial recognition of someone's ID and you put their ID up as well and does the face and all that. And for us, that's welcoming members who are maybe looking to move back to Ireland in a couple of years or they live in Cork and they can't come down, or they just don't want to come in and it allows them to reactivate the accounts.
Speaker 2You mentioned that you had to reassure staff. Were there any similar challenges?
Speaker 6Yeah, there was plenty of challenges because you have to bring the board on that journey and you have to bring the board to show that we're not alienating the existing members, for digitalization, shall we say. There's also the policies. All our policies would have all been written for in person and manual stuff here and there, so, and the sop. So all of that had to be changed. And quite often then you have internal audit will come in with their review of the process and they might have their own input. So it is a kind of an overarching project.
Speaker 6Every time you bring one in now with the likes of visualize, because they've done about 70 credit unions at this stage. It's a pretty straightforward process. They rolled it out. They are able to work with the IT providers. So the whole process is quite simple. If it was a brand new technology, you have to bring your IT provider, your core banking platform. All those have to be brought on the journey. Even our cybersecurity advisors need to know who's logging in from where. Our data protection people need to know where's our data flowing to and from. So there is always a journey.
Speaker 2Were any other issues you had to take into account, for example, potential biases that AI might have, or you mentioned data protection.
Speaker 6You know, when you're building, let's say, your criteria for automating loan decisioning, you have to pick certain metrics and you have to pick certain limitations and you're trying not to be biased, but ultimately you know there can be something slipping for somewhere. Now it's quite difficult because a lot of it is based on financials, but you will obviously have things that may or may not impact on the decision making. So we haven't seen massive issues. The issues we're going to see is the technology clashes. So we have information and there's an upgrade of this system. This system doesn't agree anymore and then all of a sudden, the system falls. But I think, in terms of biases, there's not enough information.
Speaker 6I think we're gathering the kind of bias.
Speaker 6Now what I would say is you could set up a bias quite easily in terms of you could say if you're not from the area, or if you're not from Ireland, or you don't live here a certain length of time, you could build in criteria to say we're going to treat you different to how we would treat someone who's lived in Ireland for so long.
Speaker 6Now we haven't done that because we have quite a lot of people from abroad who live here and people who live here who are now abroad and we haven't built it in, but that's not to say we'll make a decision for one reason and it will cause a bias on the other side, and I think when you're looking then at more advanced automation, certainly I find when we're gathering information, you have to be very careful on the papers that they're referencing, and I think that's something that a lot of credit unions will fall down on.
Speaker 6Who you pick when you do your research is really important and, like we've looked at co-pilot, we've looked at them all and we use one called Perplexity at the moment and you can get it to reference only papers from central banks and departments of finance and things like that. So we find that quite good. And I think when you're using AI to gather information, there's definitely a bias based on what information is on Google and what is everywhere. So I do think you have to be very careful with that sort of information. But when it comes to the lending and stuff, I think it's based on your finances usually.
Speaker 2And finally, based on your experience so far, what advice would you give to other credit unions in Britain or in Ireland considering adopting AI?
Speaker 6You can get AI for free to do various things and they're very good. The game changers the real game changers are in collaboration because the technology is there but it's probably too big for most credit unions on their own. The collaboration is the absolute essential for the future of credit unions. Because you look to places like the US who are slightly more mature profile. They have CUSOs, corporate credit unions. They really do collaborate through those vehicles and there's different vehicles. It doesn't have to be a collaboration under an ISO, it could just be through a CUSO but it has to be done collectively. All the good stuff that's been done in Ireland has been done collectively.
Speaker 6There's some credit unions out there who have huge ambitions for AI but they're not getting anywhere because the cost and the scale of the development is there. But there's many credit unions who are not looking at it and we need to find the ground where we can bring people on the journey because ultimately, the only way this works is if we have standardization across the board. If you can go to your local credit union nearby here and get a mortgage, a current account, a debit card, you can do the same in Yall. You can do the same in one in Dublin. That's what we need.
Speaker 6I couldn't, in Yall, sit here and say we are going to change the world with AI on our own. You need the vendors, you need IT companies. Fintechs, I think, are the ones the credit unions are not really taking advantage of and, as part of CUSA, we are actually really focusing on getting FinTechs involved with credit unions because they have the technology, we have the traditional banking background with the community and what we need is their skills and what they need is our scale and there's a perfect mix there. You know, Revolut are an IT company that happen to do a bit of finance and we need we don't think as an IT company, we shouldn't. We're not, we're for the community and that's the way we should think but we need to engage people who think like an IT company and bring them in and have the conversation and put our two expertises together and really, I think then, when we determine as a personal service for our members, I think we clean up.
Collaboration: The Path Forward
Speaker 2We've heard how Irish credit unions are embracing AI to improve operations and engage members, with collaboration and human oversight emerging as key themes. For more on this, check out the latest paper on AI from the Soboda Research Centre. Soboda members will also be able to take part in an AI webinar in May. Thanks for listening and I hope you'll join me for the next episode.