David Cushman — HFS Research[00:21]
OK, so perhaps first, if you could introduce yourself to the audience.
Natasha Davydova — CEO, AXA UK and Ireland[00:21]
My name is Natasha Davydova. I’m CEO of AXA for the UK and Ireland.
David Cushman — HFS Research[00:32]
And we’ve spoken before, so I know a little bit about your story, but today I’m very interested in understanding how GenAI has impacted your business, but also to what extent you feel you could be bolder. And if that’s the case, how do you think you could be bold, and what do you need to take that next step?
Natasha Davydova — CEO, AXA UK and Ireland[00:55]
I think AI is a game changer in the insurance business, and insurance is about text, video, image, and AI is good with that. And we have done already quite a lot, from enterprise rollouts of various AI use cases. For example, we rolled out Secure GPT to all of our 160,000 employees. We also rolled out AI and data academies so that our employees know what they are dealing with. They are aware of any ethical, operational, or other considerations or guardrails we need to put in for AI. Obviously we are a highly regulated entity and we have a lot of sensitive information. We also use AI to build our own chatbots. We use AI in the context of customer communications and customer service, and we have seen fantastic return on investment, for example, in the call center space. When a customer calls with the details about their policy, it used to take the call center agent on average 5 minutes to put the customer on hold, look at the policy data, and then provide the answer. Now I can actually pick up the question that had been asked and provide the answer with the reference within 5 seconds — so from 5 minutes to 5 seconds. I have seen a lot of AI applications within the back office and productivity enhancement space, but I strongly believe that the most potential of AI is within the customer space, and I think that’s where we could be perhaps a bit bolder. But we need also to be quite careful and mindful of privacy, data regulations, and customer wishes, because, as you know, I’m often asked, why don’t we do more with AI? Well, because sometimes the customer doesn’t want it. In the UK actually, customers prefer personal touch. Secondly, in insurance, we often deal with very sensitive cases where AI, perhaps rule-based AI, cannot handle. For example, if you need to make an exception and a customer is in distress but it’s not in their policy, we would often consider that given the customer circumstances, and that’s why I think the contextual meaning of the query is also very, very important to us. So that’s why we can be bolder, but we also need to be mindful.
David Cushman — HFS Research[03:24]
That point there about the human nuance of dealing with the emotions of an individual, I think that’s an area that generative AI has been quite focused on. There’s quite a number of technologies to think of. Anthropic’s approach or Inflection AI’s approach are working towards giving you that kind of emotional connect. Does it need to take a step change before you’re going to trust it to replace a human for that kind of need?
Natasha Davydova — CEO, AXA UK and Ireland[03:46]
I think it’s still early days, I would say, and I think moving from rule-based to contextual AI is still very much in progress, from my personal perspective. And I think many customers would actually sometimes prefer to speak to a human rather than to deal with AI. So that’s why I think we need to take care of that as well.
David Cushman — HFS Research[04:13]
And I’ve heard that before. It’s often said, and I think one of the great challenges — and this is not your world, but an example I would go to is, do you want to get your cancer diagnosis from a machine? Probably not, right? That’s probably something you want to hear from a human being. And I think there’s a whole range of kind of 1 to 0 on that, of how much you’re prepared to have a machine intervene in the process. Where in insurance do you think it sits? I mean, it’s not at that number 10 level, is it? Do you think people are kind of learning to accept interaction with machines?
Natasha Davydova — CEO, AXA UK and Ireland[04:58]
I think it really depends on the use case, because we have seen already examples of wearable technologies, also within the medical setting as well. And sometimes I think machines might be helpful and maybe more precise, but often it’s in conjunction with a human being. So I’m actually looking at a human being enabled with AI, rather than human against AI or being replaced. So I think it really depends on the use cases. Sometimes if it’s a simple case, a mundane task, where you have to process a huge amount of information like an underwriting, but within a short time — and we use AI in the underwriting process as well, where the underwriter has to compare a vast amount of information — and we moved from weeks and days, or even maybe multiple weeks and days, to hours and minutes, and that is really helpful. But you still need a professional expert at the end to make the proper judgment calls. So I think it’s still kind of early days, and in certain cases, absolutely, but in some others we need to give the customer choice and be mindful.
David Cushman — HFS Research[06:19]
So maybe just one last question, and you can dismiss this if you choose, but to what extent are you personally using generative AI in your day-to-day work?
Natasha Davydova — CEO, AXA UK and Ireland[06:19]
I don’t want to dismiss that question. I think it’s a good question. I use AI fairly often. And I also love to use AI in my personal life as well. I use Anthropic, I use GPT-4, I use others, and I compare. Some are better than others in particular use cases, and I find it extremely helpful actually, and it saves a lot of time. But I think still, the experience of interaction with people, AI probably will not be able to replace that, right? But I find it extremely, extremely helpful in some use cases in my personal life. I move from things which take me probably 30 minutes to 1 hour to 5 minutes.
David Cushman — HFS Research[07:25]
And do you think your personal use of it — what’s the chicken and egg here? Were you using it personally before you thought, aha, I could use it in the business? Or was the business demanding it of you, so you started using it personally to learn a bit about it?
Natasha Davydova — CEO, AXA UK and Ireland[07:40]
I think it kind of came all together. I have been using it — I’m quite inquisitive, I love to learn, and I have been using it for quite some time. But actually, if you think of insurance itself, though a lot of people have a view that it’s kind of perhaps a traditional business, we have been using AI for a very long time, for example, in fraud detection and other areas like that, so it’s not really new. I think it became more easy to use, and we moved from maybe mid-2023, when the boards were interested to learn about AI. We made huge strides now implementing AI enterprise-wise, educating our people how to use it ethically, how to use it consciously, how to use it properly. Regulation is still to catch up with it, but I’m seeing already regulators trying to create communities, create knowledge, and get feedback from the industry as well. So I think it’s moving along very well, but I think you still should not forget about the guardrails. So maybe it’s not the gates you should put in, but the guardrails, and that’s my view.
David Cushman — HFS Research[08:49]
Thank you very much.
Natasha Davydova — CEO, AXA UK and Ireland[08:49]
Thank you. It was a pleasure.
