What’s in this podcast?
AI may be advancing at extraordinary speed, but making it work inside an organisation is about far more than the technology.
In this episode of Hub & Spoken, Jason Foster, CEO & Founder of Cynozure, is joined by Alex Sidgreaves, Chief Data Officer at Zurich Insurance, to explore how leaders can move from AI experimentation towards meaningful business value – while bringing their people with them.
They discuss why organisations need to start with the problem, not the technology, and why the businesses that succeed won’t necessarily be those using the most AI, but those putting it in the right places.
The conversation also tackles one of the harder questions surrounding AI: trust. From employees being asked to work differently without knowing exactly what the future looks like, to customers wanting human empathy at critical moments, successful adoption depends on understanding where technology helps and where human judgement and connection still matter.
Jason and Alex explore the role of leadership, governance, skills and culture in making that happen, and why middle managers could ultimately determine whether AI transformation succeeds or fails.
Episode timings
00:00 — Introduction
05:05 — The Chief Data Officer as a people leader
12:06 — Finding the right opportunities for AI
19:48 — Governance, value and the boardroom
24:46 — Bringing employees on the AI journey
29:48 — Building trust through continuous change
33:29 — AI, inclusion and the future workforce
37:06 — Preserving choice and human connection
39:28 — What comes next for AI
The following interview transcription has been edited for length and clarity.
00:00 — Introduction
Jason Foster:
Hi everyone, welcome back to the show. Today, I’m super excited to have with me Alex Sidgreaves. Alex is the Chief Data Officer at Zurich Insurance, and we’re talking about the human side of AI: an important topic that’s very close to my heart. I love talking about the human side of data, AI and everything else that comes with it.
Alex, I’m absolutely thrilled to have you on the show. Thank you for being here with me.
Alex Sidgreaves:
Lovely to be here.
Jason Foster:
Shall we kick off with a bit about you, your background and how you came to be a Chief Data Officer? Then let’s get into AI and its human side.
Alex Sidgreaves:
Absolutely. I’ve spent about two decades working in data within the regulated insurance industry, which is not as dull as it sounds, I promise. My background spans everything from data strategy, advanced analytics, governance and transformation to AI.
Over that time, I’ve been fortunate enough to experience lots of major technology shifts. We’ve moved from traditional reporting and business intelligence through to advanced analytics, machine learning and now, very much, generative AI.
I should probably start by saying that I’m a very happy and proud technology nerd. I genuinely love the nuts and bolts of technology, and I always have. Some of my fondest career memories are from my early years as a data engineer: writing code, building data pipelines and trying to solve difficult technical problems. After hours and hours of trial, error and persistence, I would reach the answer. I still get really excited by that today.
I find technology fascinating, and I absolutely think AI is one of the most exciting technological developments I’ve seen during my career. The pace of innovation is extraordinary. If there’s something I’d like to do and AI can’t do it today, give it a couple of weeks and it probably will be able to. I don’t remember a technology working quite like that for as long as I’ve been working.
As my career has evolved, however, I’ve become more involved in leading teams, functions and transformations. I’ve found myself learning the same lesson over and over again: technology matters, absolutely, but it’s only ever part of the answer.
Every successful technological transformation I’ve seen has ultimately depended on people. That’s where the human side of AI becomes really interesting for me. The technology is fascinating, the developments are exciting and the models are impressive, but how humans respond to AI is possibly the most interesting part.
We’re starting to ask humans to consider fundamental questions about how they work, how they learn, how they build expertise, how they interact, how we create trust and how we change. Most importantly, we’re asking how we bring people on a journey. It’s the first time in my career when we haven’t known what the end state looks like yet, and I think that’s really interesting.
05:05 — The Chief Data Officer as a people leader
Jason Foster:
That point about building towards the future without being quite clear what it is creates an odd place for transformation. We often convince ourselves that we know where we’re trying to get to with transformation, but the end point often isn’t what we thought it would be. In some ways, perhaps we’re more experienced at doing what you’ve just articulated than we thought.
There’s definitely something here that links with the pace of change, the uncertainty and the potential lack of trust in things such as generative AI outputs. It creates a really interesting dynamic.
You talked about your technical background and love of technology. As a Chief Data Officer, where do you invest your time and energy when balancing the technical and human sides? What takes most of your effort and brain capacity?
Alex Sidgreaves:
It’s 80% people. A lot of chief data officers have quite technical backgrounds, so that shift is a really interesting one. But it is easily 80% people.
That might mean the people you manage and support through their careers, the suppliers you work with, your end customers, your organisation’s employees, or the board and executive team. It’s a relationship-based role in so many ways.
Jason Foster:
When you say 80% people, do you mean spending time with people, bringing people along with you, educating them or thinking about their careers and futures?
Alex Sidgreaves:
It’s all of the above and more. The technology can be amazing, but it will fail if you don’t bring people on the journey. It doesn’t matter how clever the technology is. You have to take people with you, whether they work for you, belong to the wider organisation or sit on your board or executive team. You have to gain their trust, comfort and understanding.
What’s fascinating now is that we tend to think about taking people through a technological journey as a matter of training. You take someone, put them through training and they come out the other side able to do what you want them to do.
I don’t think that works any more. There’s no sheep-dipping AI. You can’t send somebody on a two-week training course and say, “Right, all good. Let’s go forward.” That doesn’t mean we want to make every person in the organisation a data scientist. We absolutely don’t.
We do need people to understand what AI is good at, what it’s weak at, that they’re allowed to challenge it and how they can use it. There are so many different considerations. It also comes back to the basics: if your foundations aren’t in a good place, you can’t use it in the first place.
AI is probably the first technology I’ve seen where the positives and negatives are almost equal in scale. There’s a huge number of positives for organisations. We can do things faster and explore things differently. I probably couldn’t do without Copilot in my day-to-day world now.
There are negative sides, too, because bad actors also have access to AI. We’re seeing increasingly sophisticated cyberattacks and phishing attacks. I don’t think a technology has ever arrived with quite so much opportunity and threat at the same time.
Jason Foster:
That creates a people challenge in itself. There’s an education piece involved in making sure people understand that. Most mainstream press coverage focuses on the negative side: cyberattacks or AI systems supposedly going rogue. That’s what many people hear.
You occasionally get a story about AI being able to identify a new type of cancer, but those stories are comparatively few and far between. That creates confusion within businesses. Is this a good thing? Should we invest heavily, slow down or think about it more? Has that created challenges for you?
Alex Sidgreaves:
It has, but this is also a technology that people have been able to get their hands on in a way they couldn’t with many previous technologies.
When cloud technology arrived, I always hated the word “cloud” because it made it feel like an ethereal thing that didn’t really exist, rather than a load of data centres. I remember having many conversations with people who couldn’t touch, feel or understand it.
With AI, most of us are already using it in our personal lives. I use ChatGPT all the time. I’m doing house renovations at the moment, so I might ask it which product is best to use. People are generally comfortable using AI in their personal lives. Because they have access to it and feel comfortable with it, they can be quite willing to bring it into their professional lives.
At the same time, there’s a massive fear of missing out. Organisations are making public statements about partnering with Anthropic and doing various other things. No board or executive team wants to be part of the company that isn’t taking those steps. They don’t want to miss out or be left behind.
However, we’ve recently seen stories about AI supposedly going rogue. Interestingly, that comes down to people again. We didn’t secure it sufficiently within a sandbox, so something escaped and did something it wasn’t necessarily intended to do. As far as the AI was concerned, it was simply doing the job it had been asked to do.
Jason Foster:
Exactly. It’s often positioned as though the AI chose to do something mysterious, but it was simply being technology and doing what it was asked to do. “Going rogue” isn’t really accurate, but that language enters the public psyche and people start to worry.
12:06 — Finding the right opportunities for AI
Jason Foster:
I want to return to how you bring an organisation along and solve some of these challenges, but first I’d like to step back and look at the opportunity.
Organisations often start with, “We’ve got this technology. What should we do with it?” Where is the real opportunity for AI? What problem statements suit it, and how easy has it been to identify those opportunities?
Alex Sidgreaves:
After years of working through different kinds of technology transformation, we had become quite good as an industry at talking about the opportunity or problem we wanted to address. With AI, I think we’ve taken a slight step backwards because of the hype cycle.
More conversations now start with, “Where can we use AI? How can we implement AI? Can we build an AI solution?” In reality, the better question is the one we’ve always asked: what opportunity are we pursuing, and what problem are we trying to solve?
Once you’ve answered that question, AI may be part of the solution. It may not. The answer might be analytics, machine learning, process redesign, better data or none of those things. Applying AI to an unclear problem simply gives you an unclear answer.
Another increasingly nuanced part of the conversation is the economics of AI. We’ve moved beyond, “It’s amazing, so let’s put AI everywhere,” to, “Actually, it can be quite expensive.” Do we need to use that particular model? Are we using the most cost-effective approach?
I think organisations will begin to differentiate themselves not by using the most AI, but by putting it in the right places.
From an insurance perspective, the opportunity was huge. We’re brilliant at structured data. We’ve had it for years, we’re enormous data organisations and we’ve been making the most of structured information for a very long time.
When large language models emerged, they gave us an enormous ability to derive value from unstructured information at scale, and we had tonnes of it. So much valuable information historically lived inside documents and conversations that we couldn’t access practically. AI has allowed us to reach that information at scale.
That might include claims notes, engineers’ reports, customer correspondence, call transcripts or medical reports. Previously, we couldn’t extract information from these sources accurately and at scale, so there’s a huge opportunity.
Jason Foster:
What does accessing that information allow the insurance sector to do that it couldn’t do before, certainly at scale?
Alex Sidgreaves:
A lot of it is about speed. If we can respond faster, we can provide better service for our customers and grow more. It becomes positive.
We used to do an awful lot manually. Now, technology can augment the human by summarising a 100- or 300-page document into two pages that we can trust. Otherwise, people would have to read those entire documents to understand something, meaning they might only be able to handle one or two cases a day.
If you can summarise them, people can handle significantly more and respond to customers faster. You can also extract information from unstructured documents and bring it into more traditional AI. We shouldn’t forget that traditional AI remains really valuable in insurance. It helps us understand the risks we’re writing, how we price them and how we ensure our customers receive the best service.
We can supplement the information we already have with information that wasn’t previously accessible and use it in that more traditional way. Those are probably the two key benefits.
Much of this is about efficiency and speed, enabling us to do much more with the people we have. There’s an enormous amount of expertise in insurance; it’s an expert business. By putting technology around those experts, you enable them to do significantly more than they could before, which helps everyone.
Jason Foster:
I love that idea of doing more. People often talk about trying to do the same with fewer people. When technology comes along, organisations think it’s an opportunity to reduce headcount or avoid recruiting.
The idea of doing more is powerful because every organisation has busy people who are already stretched. Hopefully, they’re working on important things involving customers, policies, queries and value-adding interactions. Helping those people do more is a powerful message.
AI can also enable access to information and data that wasn’t available before. That allows you to build traditional AI, such as predictive models and machine learning models, as well as new AI solutions that enable insurance businesses to work differently. There are several innovative angles from which it can be applied.
Alex Sidgreaves:
Absolutely. One thing we need to be conscious of is looking at end-to-end processes. As an industry, we can be bad at solving one particular part of a process without solving the whole thing. That doesn’t mean you necessarily need AI throughout the process. It could involve process transformation or many other things.
For example, if a claims handler saves ten minutes when writing a note, but the process still takes weeks because of other handovers, there’s no real benefit. You have to examine the end-to-end process and use all the tools in the toolbox. AI, powerful as it is, is only one of those tools.
19:48 — Governance, value and the boardroom
Jason Foster:
You’ve been in insurance for a long time and described it as an interesting industry. I’ve been in and around it for a long time myself, and I agree.
How is all this viewed across the sector? Are you championing a distinctive position, or is AI viewed differently in the boardroom, at executive level and across the industry more generally? How are AI and data viewed as capabilities that can genuinely add value?
Alex Sidgreaves:
We’re quite risk-averse as an industry because of what we do, but we’re also excited about technology. Because we’re heavily regulated, we make sure we have the appropriate level of governance.
One thing I found fascinating when all this first emerged, particularly generative AI, was that everyone said, “We need processes, and we don’t have them.” In fact, we were quite far ahead. We already had good processes and governance.
If you have good governance, it gives you permission to innovate. It doesn’t have to block innovation. If you know something is going to be well controlled and assessed positively because the right governance surrounds it, you’re much more likely to make it successful.
From a boardroom perspective, the conversation has changed enormously. Two or three years ago, I was going into the boardroom to explain what AI was, the different types of AI and what they meant. At one point, I was trying to explain neural networks to a board, which was a fascinating conversation. I also ran summer schools and various other activities.
The conversation has moved on massively. There isn’t one dominant theme now because the focus covers many different areas.
Value is key. We’re putting a lot into this, so what tangible business outcomes are we creating? What return on AI are we creating? What risks are we introducing? We’re a risk-based organisation by nature, so we want to understand those risks.
Trust is another key consideration. Will customers, regulators and employees trust the way we’re using AI? That’s where governance and ethics come in.
Then there are skills. What does this mean for our workforce, future roles, the future of work and our capabilities?
Customer outcomes are also fundamental. Does this genuinely improve the customer experience?
Finally, because we’re a regulated organisation, what happens if something goes wrong publicly? What is the reputational impact?
The conversation has become much less about the technology. The technology is almost a given: we’ve seen it, you’ve shown it to us and we understand that it can do clever things. Now we need to consider the business, workforce, risk and reputation decisions that sit around it. Those are equally important problems to solve alongside the technology.
Jason Foster:
That provides a useful framework for bringing the organisation with you. You start with value: where can we create the most value, stop the most value leaking away or enable people to do more? Those are things people care about, understand and can get behind because they relate to their day-to-day jobs.
Then there’s the question of what risks we’re adding. We’re going to do these things, but they may create new risks that we should understand. One of the insurance sector’s strengths is its grasp of risk. You assess and price risk all the time, and risk management is a core capability that feeds into governance.
Trust and skills are closely aligned. How do people begin to trust this? We have to give them the skills, knowledge and education. That combination takes us to the human side: understanding people and behaviour while identifying where we can genuinely drive value. It comes full circle because we begin with value and end by delivering value to customers while bringing our teams with us.
Have you found that framing the discussion through those lenses helps to bring organisations with you, or is there more to helping people understand the topic?
24:46 — Bringing employees on the AI journey
Alex Sidgreaves:
It depends which level of the organisation you’re trying to bring with you at that moment. From a board and senior executive perspective, that framework absolutely makes sense.
It becomes quite different for people on the ground. All those things will matter to them to an extent, but we’re asking them to move towards a position when we don’t quite know what it’s going to look like yet.
Jason Foster:
You’re asking them to take a leap of faith.
Alex Sidgreaves:
We’re asking them to have a huge amount of faith. The average employee may be feeling many things at once. They might be excited, curious or concerned, and there may be a lot of uncertainty.
Somebody once said to me, “Using AI feels like cheating.” I don’t consider it cheating myself, but people have much more human questions: will my job change? Will I still be valuable? What skills do I need to learn? What happens if I rely on AI too much? What is my responsibility? Where does the accountability sit? How do I know when it’s acceptable to challenge the AI? What does good look like now?
There are many questions and concerns about skills fade and how we’ll develop the experts of the future.
You can’t mandate that employees trust AI. You have to earn that trust through many small steps, continuously over time, as you bring people on the journey. It comes through practice, challenge, evidence and patience.
You can bring an organisation along using a strategic framework of pillars and priorities, but that doesn’t necessarily help a human being who is wondering whether they’ll still have a job in six months, whether their role will be fundamentally different or what this will mean for their children. That’s an entirely different matter to address.
Jason Foster:
You want to give people understanding, space to explore and opportunities to grow and step into something new. You can’t pretend that nothing will change, but you don’t want to scare people either.
There’s a middle ground where you say, “Some things are going to change. We’ll support you, educate you and provide opportunities along the way. In return, you need to think about it in this way.”
Whether that works often comes back to the cultural norms within an organisation or its individual teams. If an organisation isn’t innovative, collaborative or supportive by nature, it won’t necessarily become those things suddenly, even though that’s what may be required.
Alex Sidgreaves:
It does come down to culture, but it can also come down to individual leaders. It’s great if your CEO stands up and says, “We’re going to do this, and it’s going to be amazing.” Having the executive team behind it is fantastic.
However, I recently read something that I found fascinating and strongly agreed with: middle management is where this succeeds or fails.
Those individual leaders deal with people daily. They can reinforce the culture, tell people it’s acceptable to learn and reassure them. They can explain that roles may change, but that the organisation will support people and help them make the most of the opportunities.
That middle-management layer is ultimately where this will succeed or fail, although support from the top also helps.
Jason Foster:
It has to come from all angles. You need recognition and support from the top, including funding, space and action rather than words alone.
You also need the layer where work is led operationally and things happen day to day, as well as support from the bottom up. Those things have to happen together.
There’s a strategic answer and a real, human, day-to-day answer. They have to be connected; otherwise, you’re saying one thing and doing another.
29:48 — Building trust through continuous change
Jason Foster:
Where will organisations go wrong as they embark on this?
Alex Sidgreaves:
They’ll go wrong by treating it as a one-and-done activity. This is probably the longest training and coaching initiative your organisation will ever undertake. Whether it takes 12 months, 18 months or two years, I don’t know.
You can’t take everybody through a training course, send them out the other side and assume everything will be fine. Nor is this about turning everybody in the organisation into a data scientist. We don’t need that either.
We need to understand which future skills will be valuable. They’re very much human skills: adaptability, curiosity, critical thinking, communication and the confidence to challenge AI and say, “I don’t think that’s quite right.”
People will go wrong if they think they can do this quickly, that there’s an easy sheep-dipping answer or that they can mandate trust. People will happily tell you what they think you want to hear, but that doesn’t mean it’s how they really feel.
You can put the most amazing technology in front of somebody, but it has failed if they don’t use it.
We talked earlier about summarising documents. You can take a 300-page document down to two pages and prove through testing that the summary is accurate and produces the same answers. But if the person making the decision doesn’t feel confident using those two pages and returns to the full 300 pages every time, you’ve achieved nothing.
Jason Foster:
Much of this comes down to uncertainty. We’re asking people to take a leap of faith while telling them we don’t have all the answers, but we’re also trying to build trust.
How do you remove uncertainty, help the organisation accept that there will be uncertainty or minimise it? That feels like an important area of focus.
Alex Sidgreaves:
I don’t think you can remove the uncertainty. You can provide some clarity about the direction of travel, the principles you’re following and the areas you’re examining.
In my experience, people respond well to honesty and transparency. You might not be able to remove uncertainty, but if you tell everybody that you know exactly what will happen and that everything will be perfect, they’ll see through it in a heartbeat. It’s better to be honest and transparent.
This technology is open to all levels. One thing I love about it is that the people at the top aren’t necessarily any more capable of providing the answer than people who are just beginning their careers. Everybody has an equal opportunity to raise their hand and say, “What about this?” We’re seeing that increasingly, which is really interesting.
33:29 — AI, inclusion and the future workforce
Jason Foster:
When we think about people and culture, a lot of work in recent years has focused on diversity, equity and inclusion.
AI potentially creates challenges for that, particularly because of the bias built into models, which has been discussed extensively. Does AI create better opportunities to address some DEI challenges, or are you seeing it as a problem?
Alex Sidgreaves:
I definitely see it as a problem. I’m not sure whether it helps us solve it.
It’s similar to the issue of skills fade that I touched on earlier: how do we develop the experts of the future and protect junior-level roles?
If you look statistically at many of the roles that could potentially be replaced by this kind of technology, they’re predominantly occupied by women, including many women who work part-time. There’s a potential issue there.
If I look at my own career and my early work as an engineer, I don’t think that role will exist for much longer. I have engineers working for me now who are phenomenal, but I don’t think the days of simply coding will be around for much longer because technology can do it amazingly well. The job will be very different when the human is augmented by technology.
From a neurodiversity perspective, the nature of the skills involved attracts a lot of neurodiverse people to data and AI. Many of those jobs are going to change fundamentally. People may be asked to focus more on communication, critical thinking and other skills that could be further outside their comfort zones than the traditional job was.
As with many aspects of this technology, these are conversations we need to continue having. I’m sure there’s an enormously positive side to DEI, but AI also introduces challenges that we need to face, just like skills fade. We must acknowledge them and decide how to work with them rather than simply allowing them to happen.
Jason Foster:
If DEI is important to an organisation, it needs to remain important in the context of AI, just as it would when discussing recruitment, office moves or anything else affecting the workforce.
Personally, I’ve seen AI make attention to DEI essential, particularly when organisations are building their own AI solutions. There’s a difference between using an existing AI or large language model to work faster or analyse documents and building your own solutions. We can’t do the latter without considering the end users of the platforms and tools we create.
37:06 — Preserving choice and human connection
Alex Sidgreaves:
We also need to think about our customers. Insurance is ultimately a human business. Our customers don’t want our technology; they’re experiencing a life event, and those events can be very different.
If water pours through my ceiling and my television goes bang, I don’t mind AI being involved in that conversation. I just want my new television to arrive as quickly as possible.
If somebody I care about has died, however, I want empathy, reassurance and human connection. That’s my personal preference. I don’t want AI involved in the customer-facing side of that interaction. What happens behind the scenes is entirely fine, but I don’t want it involved in my direct experience.
Another customer might feel completely differently. The crucial thing is that every customer should have the right to choose the type of interaction that works for them.
We need to ensure the customer experience isn’t always about achieving the greatest degree of automation. It should be about meeting the customer where they are, both emotionally and practically.
Jason Foster:
It isn’t one size fits all. It’s easy to return to the question of how we can be more efficient. Your earlier point was about how we can do more, but we also shouldn’t automate away everything we do.
Sometimes you want a little friction because it causes you to pause and slow down. When somebody is dealing with bereavement, that isn’t something to automate. It requires care, empathy, reassurance, support and a human touch. In some ways, you can’t codify that.
The approach needs to reflect people’s choices. We can’t enter this space assuming that all situations are equal and AI should be applied to everything, although that’s where people can end up.
39:28 — What comes next for AI
Jason Foster:
To finish, where do you think all this is going?
You talked about how quickly everything is changing and how something AI can’t do today might become possible in a few weeks. From the perspective of its impact on people, are we past the point of trying to get to grips with it and now focused on executing well? Or is there still a lot of change, iteration and movement to come before it settles into something we can execute properly?
Alex Sidgreaves:
I think it’s a bit of both. The genie is so far out of the bottle that we’re not putting it back in, and the train has absolutely left the station.
In some respects, I think we’ll look back at this period as being like the early days of the internet. I’m old enough to remember being excited that a page loaded in 15 minutes and that I could talk to people around the world. It was amazing.
Now, I can’t deal with life when my phone won’t connect to a map in London because I haven’t got a signal. The internet changed enormously over time.
There’s already so much we can do today, and it will continue to evolve. If we look forward a number of years, I think today will feel like those early days of the internet compared with now having the internet in your pocket.
There’s a huge amount that will continue to change. For businesses, AI is going to amplify what’s already there. If your data, culture, skills and customer relationships are strong, AI can accelerate that value. If they’re weak, it’s going to expose the gaps.
The winners won’t be the organisations with the most AI. They’ll be the organisations that combine technology, trust and humanity most effectively.
Jason Foster:
I’ve already seen that. Organisations that are well set up with data and have the right culture to operate in a digital, collaborative, agile and iterative world are well prepared for this.
Those that are still trying to catch up can get there, but they have a bigger transformation and a bigger people change to manage alongside the technical change.
Thank you. It’s been great talking to you and hearing your views. We haven’t talked about insurance extensively, but when we did, it was certainly interesting and fascinating, so I agree with you on that point. Thank you for being on the show.
Alex Sidgreaves:
You’re welcome.
Jason Foster:
Thanks, everyone, for listening. We hope you got a lot from today’s episode.
As I said at the start, I’m a huge believer that the human side of AI will far outweigh its technological side when it comes to our ability to execute well. Hopefully, you’ve learnt a lot from Alex during today’s conversation, and we’ll catch you again soon.
