Transforming The Future Of Banking With Generative AI
Breadcrumb Home Icon  // . //  INFocus //  Transforming The Future Of Banking With Generative AI

06:53

Double Quotes
Generative AI is really going to change the way you interact with the bank. For the last 15 years, the interface to the bank has been the mobile phone. I think generative AI is going to get us away from tapping to talking
Donald MacDonald, Head of Group Data Office, OCBC Bank

Oliver Wyman Partner Na Zhou joins OCBC head of Group Data Office, Donald MacDonald, as he discusses responsible and ethical artificial intelligence (AI) in the banking sector, and how he envisions an AI-enabled future for both employees and consumers of the bank.

INFocus Series
INFocus provides exclusive insights and trends from experts and leaders across the Asia Pacific region, exploring the forces, opportunities, and challenges shaping its future.

Go to series

Donald MacDonald

Generative AI is the thing that's going to make banking conversational. It's going to make it truly personalized, and it's going to make it always on. To me, that's fundamentally transforming banking.

[series intro clip]

Na Zhou

You mentioned certain things about your central team is responsible for building guardrails, evaluating models. I also know you're a bit of an industry leader in terms of ethical AI, responsible AI, fairness. So, Donald, tell us a little bit more about how OCBC is scaling up AI in a fair, safe, and responsible manner.

Donald

We take it very seriously. We believe you can't have generative AI without responsible AI – these two things go hand in hand. I think we're fortunate to live in Singapore because Singapore has been quite open about wanting to be a global leader when it comes to the use of responsible AI. We're very fortunate to have very forward-looking collaborative regulators. We have MAS on the financial services side, and IMDA on the media side, and they actively collaborate with the industry to define what are those industry guardrails going to look like. What I especially like is they're not just writing large documents; we're actively crafting open-source software that we can deploy into our processes for things like fairness testing or red teaming on LLMs. So, I think Singapore has done a great job when it comes to AI safety.

As we develop our applications, we've been progressively building our internal guardrails, to make sure that those GenAI applications are done securely. The kind of things that we've been building [are] evaluation frameworks so that when new LLMs are launched on the market, we're able to accurately test those models to find out where are their limitations. Are they toxic? Is there gender bias, for example? So we know the limitations.

When it comes to building applications, we've been building internal guardrails like our RAG system. We can control the context of what's actually going into the models and building things like hallucination detection, and also using FLARE for hallucination rectification as well. We build these as common services that we deploy; available to anyone in OCBC when we have our LLM available as a service.

Na

I also feel quite fortunate to be operating in Singapore. I definitely second what you said about MAS being more practical and actually releasing things that the practitioners like you and I can use. Oliver Wyman also worked with MAS on a number of projects in that regard.

Maybe switching gears a little bit, if we can fast forward to 3 to 5 years in the future, what in your mind is a generative AI-enabled OCBC? What does it look like?

Donald

I think it looks very different for employees and also for customers. For employees, the way I imagine generative AI changing the future is that generative AI will really be augmenting our employees. It will be helping to automate a lot of the drudge work that every employee faces in their day-to-day tasks. If I can use generative AI to take away some of the brain fatigue things that you don't enjoy doing, so that you can spend time focusing on the higher value tasks – potentially even getting home to see your family an hour earlier every day – then to me, that's transformational.

If I can take, 30, 40 minutes out of every employee’s day, then already that's a huge productivity gain across 30,000 employees. Eventually, if I can get to the stage where I'm taking one hour out of every employee's day across the organization, you have to start asking yourself: do I need to be working five days a week? Or is this the thing that ultimately gets us to the four-day working week? And that is truly, you know, transformational, for society.

For customers, I think generative AI is really going to change the way that you interact with the bank. For the last 15 years, the interface to the bank has been the mobile phone. I think generative AI is going to get us away from tapping to talking. It's about having a more conversational interface with your bank, whether that's through voice or whether that's through text chat. The kind of future I'm imagining is you wake up in the morning, you're having your breakfast cereal, and you ask your phone, “hey, what happened in the market last night?” and the generative AI tells you, “hey, there was volatility in the market, and this stock went up and this stock went down.” Okay, what should I do? The generative AI knows your profile; it knows your risk appetite, what assets you hold today, and it says this is what I recommend that you do. And I say yes, execute that. Behind the scenes, the agent frameworks pick up on that. They take your order; they go and execute it. By the time you go to work, already these tasks have been completed and you get an email confirming it.

To me, generative AI is the thing that's going to make banking conversational. It's going to make it truly personalized, and it's going to make it always on. To me, that's fundamentally transforming banking.

Na

It's truly, fundamentally transforming banking. It’s going to look completely different. And I think what's really resonating with me is not only all of those sounds fantastic, but they are also within reach – a lot of the technology we're talking about is already there. We are actively experimenting and building those solutions.

Donald

Yeah, agreed. Two years ago, those things were science fiction. Today, generative AI can already understand the voice. You can have a good, sensible conversation with a generative AI, and agent frameworks are a real thing. Today, 70% of the calls to large language models in OCBC are made by agents. Already a lot of our complex workflows, we have agents behind the scenes executing those things. I think all the pieces are there. It's just about putting them together and having the right guardrails in place so people are comfortable.

Na

Yes. Fantastic. What are some of the advice, to wrap it up, you would like to give other industry leaders who's on a similar journey scaling AI?

Donald

My key advice would be: don't wait. You know, I think already, the time for experimentation and the time for POCs is over. I think the benefits are clear. There's lots of examples, like OCBC, where we're showing real benefits from generative AI. I think there's already clarity on what are the main use cases. It’s about using generative AI for coding, for customer service, for discovering knowledge, and curating new content.

Someone's already identified what are the key use cases. There's no need for you to experiment anymore. Go out there and start capturing those benefits.

Na

Thank you so much once again, Donald, for your insight and your time. We covered a broad range of topic under AI, scaling AI. We touched on your own journey at OCBC starting 2019. We mentioned this multifaceted approach that everyone needs to embrace in order to scale up AI.

We also touched on how fortunate we are in Singapore, to embrace responsible AI. Finally, I love the vision you've panned out and is so within reach. Let's just take action right now.

Thank you very much for the insight. And congratulations on everything you've achieved so far. Really look forward to the future that you described to us.

Donald

Thank you very much. Really enjoyed the discussion.

    Oliver Wyman Partner Na Zhou joins OCBC head of Group Data Office, Donald MacDonald, as he discusses responsible and ethical artificial intelligence (AI) in the banking sector, and how he envisions an AI-enabled future for both employees and consumers of the bank.

    INFocus Series
    INFocus provides exclusive insights and trends from experts and leaders across the Asia Pacific region, exploring the forces, opportunities, and challenges shaping its future.

    Go to series

    Donald MacDonald

    Generative AI is the thing that's going to make banking conversational. It's going to make it truly personalized, and it's going to make it always on. To me, that's fundamentally transforming banking.

    [series intro clip]

    Na Zhou

    You mentioned certain things about your central team is responsible for building guardrails, evaluating models. I also know you're a bit of an industry leader in terms of ethical AI, responsible AI, fairness. So, Donald, tell us a little bit more about how OCBC is scaling up AI in a fair, safe, and responsible manner.

    Donald

    We take it very seriously. We believe you can't have generative AI without responsible AI – these two things go hand in hand. I think we're fortunate to live in Singapore because Singapore has been quite open about wanting to be a global leader when it comes to the use of responsible AI. We're very fortunate to have very forward-looking collaborative regulators. We have MAS on the financial services side, and IMDA on the media side, and they actively collaborate with the industry to define what are those industry guardrails going to look like. What I especially like is they're not just writing large documents; we're actively crafting open-source software that we can deploy into our processes for things like fairness testing or red teaming on LLMs. So, I think Singapore has done a great job when it comes to AI safety.

    As we develop our applications, we've been progressively building our internal guardrails, to make sure that those GenAI applications are done securely. The kind of things that we've been building [are] evaluation frameworks so that when new LLMs are launched on the market, we're able to accurately test those models to find out where are their limitations. Are they toxic? Is there gender bias, for example? So we know the limitations.

    When it comes to building applications, we've been building internal guardrails like our RAG system. We can control the context of what's actually going into the models and building things like hallucination detection, and also using FLARE for hallucination rectification as well. We build these as common services that we deploy; available to anyone in OCBC when we have our LLM available as a service.

    Na

    I also feel quite fortunate to be operating in Singapore. I definitely second what you said about MAS being more practical and actually releasing things that the practitioners like you and I can use. Oliver Wyman also worked with MAS on a number of projects in that regard.

    Maybe switching gears a little bit, if we can fast forward to 3 to 5 years in the future, what in your mind is a generative AI-enabled OCBC? What does it look like?

    Donald

    I think it looks very different for employees and also for customers. For employees, the way I imagine generative AI changing the future is that generative AI will really be augmenting our employees. It will be helping to automate a lot of the drudge work that every employee faces in their day-to-day tasks. If I can use generative AI to take away some of the brain fatigue things that you don't enjoy doing, so that you can spend time focusing on the higher value tasks – potentially even getting home to see your family an hour earlier every day – then to me, that's transformational.

    If I can take, 30, 40 minutes out of every employee’s day, then already that's a huge productivity gain across 30,000 employees. Eventually, if I can get to the stage where I'm taking one hour out of every employee's day across the organization, you have to start asking yourself: do I need to be working five days a week? Or is this the thing that ultimately gets us to the four-day working week? And that is truly, you know, transformational, for society.

    For customers, I think generative AI is really going to change the way that you interact with the bank. For the last 15 years, the interface to the bank has been the mobile phone. I think generative AI is going to get us away from tapping to talking. It's about having a more conversational interface with your bank, whether that's through voice or whether that's through text chat. The kind of future I'm imagining is you wake up in the morning, you're having your breakfast cereal, and you ask your phone, “hey, what happened in the market last night?” and the generative AI tells you, “hey, there was volatility in the market, and this stock went up and this stock went down.” Okay, what should I do? The generative AI knows your profile; it knows your risk appetite, what assets you hold today, and it says this is what I recommend that you do. And I say yes, execute that. Behind the scenes, the agent frameworks pick up on that. They take your order; they go and execute it. By the time you go to work, already these tasks have been completed and you get an email confirming it.

    To me, generative AI is the thing that's going to make banking conversational. It's going to make it truly personalized, and it's going to make it always on. To me, that's fundamentally transforming banking.

    Na

    It's truly, fundamentally transforming banking. It’s going to look completely different. And I think what's really resonating with me is not only all of those sounds fantastic, but they are also within reach – a lot of the technology we're talking about is already there. We are actively experimenting and building those solutions.

    Donald

    Yeah, agreed. Two years ago, those things were science fiction. Today, generative AI can already understand the voice. You can have a good, sensible conversation with a generative AI, and agent frameworks are a real thing. Today, 70% of the calls to large language models in OCBC are made by agents. Already a lot of our complex workflows, we have agents behind the scenes executing those things. I think all the pieces are there. It's just about putting them together and having the right guardrails in place so people are comfortable.

    Na

    Yes. Fantastic. What are some of the advice, to wrap it up, you would like to give other industry leaders who's on a similar journey scaling AI?

    Donald

    My key advice would be: don't wait. You know, I think already, the time for experimentation and the time for POCs is over. I think the benefits are clear. There's lots of examples, like OCBC, where we're showing real benefits from generative AI. I think there's already clarity on what are the main use cases. It’s about using generative AI for coding, for customer service, for discovering knowledge, and curating new content.

    Someone's already identified what are the key use cases. There's no need for you to experiment anymore. Go out there and start capturing those benefits.

    Na

    Thank you so much once again, Donald, for your insight and your time. We covered a broad range of topic under AI, scaling AI. We touched on your own journey at OCBC starting 2019. We mentioned this multifaceted approach that everyone needs to embrace in order to scale up AI.

    We also touched on how fortunate we are in Singapore, to embrace responsible AI. Finally, I love the vision you've panned out and is so within reach. Let's just take action right now.

    Thank you very much for the insight. And congratulations on everything you've achieved so far. Really look forward to the future that you described to us.

    Donald

    Thank you very much. Really enjoyed the discussion.

    Oliver Wyman Partner Na Zhou joins OCBC head of Group Data Office, Donald MacDonald, as he discusses responsible and ethical artificial intelligence (AI) in the banking sector, and how he envisions an AI-enabled future for both employees and consumers of the bank.

    INFocus Series
    INFocus provides exclusive insights and trends from experts and leaders across the Asia Pacific region, exploring the forces, opportunities, and challenges shaping its future.

    Go to series

    Donald MacDonald

    Generative AI is the thing that's going to make banking conversational. It's going to make it truly personalized, and it's going to make it always on. To me, that's fundamentally transforming banking.

    [series intro clip]

    Na Zhou

    You mentioned certain things about your central team is responsible for building guardrails, evaluating models. I also know you're a bit of an industry leader in terms of ethical AI, responsible AI, fairness. So, Donald, tell us a little bit more about how OCBC is scaling up AI in a fair, safe, and responsible manner.

    Donald

    We take it very seriously. We believe you can't have generative AI without responsible AI – these two things go hand in hand. I think we're fortunate to live in Singapore because Singapore has been quite open about wanting to be a global leader when it comes to the use of responsible AI. We're very fortunate to have very forward-looking collaborative regulators. We have MAS on the financial services side, and IMDA on the media side, and they actively collaborate with the industry to define what are those industry guardrails going to look like. What I especially like is they're not just writing large documents; we're actively crafting open-source software that we can deploy into our processes for things like fairness testing or red teaming on LLMs. So, I think Singapore has done a great job when it comes to AI safety.

    As we develop our applications, we've been progressively building our internal guardrails, to make sure that those GenAI applications are done securely. The kind of things that we've been building [are] evaluation frameworks so that when new LLMs are launched on the market, we're able to accurately test those models to find out where are their limitations. Are they toxic? Is there gender bias, for example? So we know the limitations.

    When it comes to building applications, we've been building internal guardrails like our RAG system. We can control the context of what's actually going into the models and building things like hallucination detection, and also using FLARE for hallucination rectification as well. We build these as common services that we deploy; available to anyone in OCBC when we have our LLM available as a service.

    Na

    I also feel quite fortunate to be operating in Singapore. I definitely second what you said about MAS being more practical and actually releasing things that the practitioners like you and I can use. Oliver Wyman also worked with MAS on a number of projects in that regard.

    Maybe switching gears a little bit, if we can fast forward to 3 to 5 years in the future, what in your mind is a generative AI-enabled OCBC? What does it look like?

    Donald

    I think it looks very different for employees and also for customers. For employees, the way I imagine generative AI changing the future is that generative AI will really be augmenting our employees. It will be helping to automate a lot of the drudge work that every employee faces in their day-to-day tasks. If I can use generative AI to take away some of the brain fatigue things that you don't enjoy doing, so that you can spend time focusing on the higher value tasks – potentially even getting home to see your family an hour earlier every day – then to me, that's transformational.

    If I can take, 30, 40 minutes out of every employee’s day, then already that's a huge productivity gain across 30,000 employees. Eventually, if I can get to the stage where I'm taking one hour out of every employee's day across the organization, you have to start asking yourself: do I need to be working five days a week? Or is this the thing that ultimately gets us to the four-day working week? And that is truly, you know, transformational, for society.

    For customers, I think generative AI is really going to change the way that you interact with the bank. For the last 15 years, the interface to the bank has been the mobile phone. I think generative AI is going to get us away from tapping to talking. It's about having a more conversational interface with your bank, whether that's through voice or whether that's through text chat. The kind of future I'm imagining is you wake up in the morning, you're having your breakfast cereal, and you ask your phone, “hey, what happened in the market last night?” and the generative AI tells you, “hey, there was volatility in the market, and this stock went up and this stock went down.” Okay, what should I do? The generative AI knows your profile; it knows your risk appetite, what assets you hold today, and it says this is what I recommend that you do. And I say yes, execute that. Behind the scenes, the agent frameworks pick up on that. They take your order; they go and execute it. By the time you go to work, already these tasks have been completed and you get an email confirming it.

    To me, generative AI is the thing that's going to make banking conversational. It's going to make it truly personalized, and it's going to make it always on. To me, that's fundamentally transforming banking.

    Na

    It's truly, fundamentally transforming banking. It’s going to look completely different. And I think what's really resonating with me is not only all of those sounds fantastic, but they are also within reach – a lot of the technology we're talking about is already there. We are actively experimenting and building those solutions.

    Donald

    Yeah, agreed. Two years ago, those things were science fiction. Today, generative AI can already understand the voice. You can have a good, sensible conversation with a generative AI, and agent frameworks are a real thing. Today, 70% of the calls to large language models in OCBC are made by agents. Already a lot of our complex workflows, we have agents behind the scenes executing those things. I think all the pieces are there. It's just about putting them together and having the right guardrails in place so people are comfortable.

    Na

    Yes. Fantastic. What are some of the advice, to wrap it up, you would like to give other industry leaders who's on a similar journey scaling AI?

    Donald

    My key advice would be: don't wait. You know, I think already, the time for experimentation and the time for POCs is over. I think the benefits are clear. There's lots of examples, like OCBC, where we're showing real benefits from generative AI. I think there's already clarity on what are the main use cases. It’s about using generative AI for coding, for customer service, for discovering knowledge, and curating new content.

    Someone's already identified what are the key use cases. There's no need for you to experiment anymore. Go out there and start capturing those benefits.

    Na

    Thank you so much once again, Donald, for your insight and your time. We covered a broad range of topic under AI, scaling AI. We touched on your own journey at OCBC starting 2019. We mentioned this multifaceted approach that everyone needs to embrace in order to scale up AI.

    We also touched on how fortunate we are in Singapore, to embrace responsible AI. Finally, I love the vision you've panned out and is so within reach. Let's just take action right now.

    Thank you very much for the insight. And congratulations on everything you've achieved so far. Really look forward to the future that you described to us.

    Donald

    Thank you very much. Really enjoyed the discussion.