Kay Firth-Butterfield
The agent just deleted a lot of proprietary data from the company and then wrote an email to the CEO apologising for it. So those are the sort of things that as we develop more and more agents, we need to say, where do we put humans? Where do we make sure that we've got a human to check that the AI agent is still performing within its constraint?
Carly Gulliver
What are people talking to their lawyers about when it comes to tech. Welcome to Inside Tech Minds from Addleshaw Goddard. In this podcast, we're sitting down with technologists, investors, business leaders who are at the heart of the biggest tech deals, innovations and disputes. I'm Carly Gulliver. Let's dive into today's episode. Hi, Kay. It's great to have you with us today on the podcast.
Kay Firth-Butterfield
Thank you very much for inviting me.
Carly Gulliver
Well it’s brilliant to have you and you're actually based in the US, aren't you?
Kay Firth-Butterfield
I am actually based in the US, yes, and you know, have been for many years now, but still keep a connection in London, still part of Doughty Street and co-lead the AI work out of Doughty Street.
Carly Gulliver
Brilliant. Kay, let me introduce you. So you are the CEO of GoodTech Advisory and one of the world's leading experts on AI governance. You previously led AI and Quantum at the World Economic Forum and whilst you were there, you also sat on the executive committee shaping how governments and businesses approach technology at scale. You bring a unique perspective to this as a barrister, homage technologist and entrepreneur, and also as the world's first Chief AI Ethics Officer back in 2014 and that's why we're grateful for you joining us and want to jump into some of the trickier issues around AI governance and that type of area with you. So how did you start to become involved, Kay, in AI?
Kay Firth-Butterfield
Okay so I mentioned Doughty Street Chambers, I'm a human rights barrister by background. I've always been interested obviously in the human condition and many years ago, I think 2010, I read a book by Ray Kurzweil called The Singularity is Nearer and there he posited the AI, intelligent AI, in a conversation with a human. I was shocked and entranced and interested in this and really it got me thinking about, you know, what would a human machine coexistence look like? And then I started researching it and at that stage, there were very few of us doing any research. And then, you said I became world's first Chief AI Ethics Officer and that was serendipity. Stephen Hawking had said in The Times that AI would be the best thing that we ever did as humans or the last thing and I happened to sit on a 10-hour plane journey next to the CEO of an AI company a week afterwards and he was really worried that he would be responsible for this existential risk to humans.
Carly Gulliver
There’s a lesson there really isn't there in terms of looking and speaking to the person next to you on a plane rather than just put your headphones in.
Kay Firth-Butterfield
Absolutely, yes.
Carly Gulliver
That sounds like that was quite a pivotal conversation for you, both personally and professionally.
Kay Firth-Butterfield
Personally and professionally, and you know as a result of that, I became the person that everybody went to because I was the only person at that stage.
Carly Gulliver
And to jump to the big question right at the start, where do you sit then on that scale in terms of the best of things or the worst of things in AI? Do think we're headed to a utopia or a dystopia?
Kay Firth-Butterfield
I think that people tend to think that I'm fairly dystopian and that's because I talk about governance and risk. But that doesn't actually mean that I don't think that AI has great potential to do good. It just means that I think that it has great potential to do good if we have good governance and compliance measures in place. And I guess as a lawyer, I would say that, but that's my position.
I think that one of the worries that I have as I see trust eroding in AI amongst people is that we will literally throw this baby out with the bath water because we will be so lacking in trust and lacking in trust is because there are not good governance measures in place.
Carly Gulliver
I'd love to get into some of those points around governance and risk because as you said, you have a leaning towards that as a lawyer, so do I and so do many of our clients as well who ask us those tricky questions. But just before we do, AI is obviously a huge subject and it's a huge label. So just to set the scene, what are we talking about in the context of governance, risk and business when we use AI as a label?
Kay Firth-Butterfield
Yeah, I think the best way of thinking about it is looking at the OECD definition of AI and there’s the definition of AI and then there's the additional definition of generative AI. And that's where I always start when I'm working with clients or trying to explain to the general public about the machines that we are talking about today.
And there are also good definitions, I think, in the EU AI Act.
Carly Gulliver
So, and put on a practical lens because I know you interface with a lot of businesses who are using AI and within those two descriptions, if you like, of AI and generative AI, what does that mean for businesses and the tools that they might be thinking about? Are we thinking about things like agents and language models?
Kay Firth-Butterfield
Absolutely. Absolutely. So one of the things that I have been privileged to see is people using AI, the AI we had before generative AI and one of the things that I do see is that those people who got that right and thought deeply about governance of that have made the adaptation to generative AI and will make the adaptation to using agents much better, I think, than those people who suddenly woke up in 2023 and said, we must all use AI and of course I was Head of AI at the forum in 2023 and in that January of 2024, all the CEOs wanted to talk about how they were using AI and hardly any of them knew what they were actually talking about.
Carly Gulliver
So what might that look like then, do you think, in terms of, you know, for our CEOs and business leaders and people active in business, what sorts of things might A, their employees already be using on day to day in their actual operations, but B, behind the scenes for more operational use and that the business might be looking at?
Kay Firth-Butterfield
Yes, so it depends upon the business and you know, lot of businesses with legacy tech are finding it quite hard to bolt AI, generative AI particularly, into their systems, but I think what we're seeing generally is we're seeing a lot of use of AI, generative AI in helping with emails and helping with back office procedures and helping with research. People will be using versions, proprietary versions of Claude or OpenAI, you know, some of the foundational models. The employees are using the same tools that you have given them because one of the things that we are seeing is that people say, well, I really prefer this tool to the one that I've been given at work. I don't like Co-Pilot, for example, you know, just to take any of these and so they've been using their own version of an AI at work and we call that shadow AI and you know, that's really dangerous. And I was consulted by a company recently about the fact that their employee had designed a valve, asafety critical valve, using their own AI. Would they be liable for the fact that when it was installed it exploded and noxious substances went everywhere? And so, you know, the first thing you do is you say, well, have you got a policy about using this? They didn't have any policy about only use the AI that we provide to you and so, you know, I think that they are going to be liable for it. And it's a really simple fix, but it's one of those things that you have to think so broadly about all the potential problems in order to do real governance.
Carly Gulliver
I see so that's really interesting and a great example, albeit not a particularly positive one of some of the bear traps, if you like, around AI and how it's being used in the business world and it sounds from what you're saying that we have a camp of AI, which could be and would be AI native in businesses so that's a mixture of, you know, probably some in-house solutions, bespoke solutions, off-the-shelf AI solutions that are being used. But then also whether or not you have that, you have a camp two of potentially your employees using what's out there in the world west of AI and what they can get their hands on and in that instance, you know, that's an example of where the employee has used their own measures, taking things into their own hands and sounds like there was some quite serious consequences.
Kay Firth-Butterfield
Absolutely. And I think that we tend to look at AI and we tend to think that it's the domain of the CTO or it's the domain of the General Counsel because the General Counsel and Compliance and Risk have got to put in some guardrails and governance. But it's really, to do it right, you need everybody. You see ,everyone will be using AI in different ways in the silos of their company, but you need everybody to be responsible for the way that it's being used in those silos and one of the most important people, I think, in the maturation of AI across a company is HR and the behavioral scientists that sit in HR in many companies because it's very often that it's not the AI that's going wrong, it's the way that your employee is using it that's going wrong. And so not including HR in that conversation and not prioritising good training to use the tools that are being given is in my opinion, a real miss.
Carly Gulliver
That's really interesting because I think you've just described that there's different layers in terms of AI adoption, but also that goes hand in hand with thinking about the governance and the policies. So if we have clients or listeners who are thinking about putting these policies and guardrails in place, how should they go about that? If they've got a blank piece of paper, what would your top tips be?
Kay Firth-Butterfield
Okay, so I hope that they haven't got a blank piece of paper. That would be my starting point.
Carly Gulliver
And don't say put it into AI.
Kay Firth-Butterfield
I won't. So I think that, you know, I've done some of this with companies. So I started some work with a very large European car company where they had actually just allowed deployment of AI sort of fairly randomly in the business staff and so we started actually with mapping what was in place and we found 260 different deployments and some of those were actually quite dangerous deployments, almost illegal deployments when you look at them in the context of the EU AI Act. So if you're starting with a blank piece of paper, go and make sure that there are, you know, are your employees using it? And if so,how and then measure that against whichever regulation is appropriate for your area. So if you're in the EU, the EU AI Act, if you're in America, whichever state has got some law that's applicable to you and of course in the UK, it's the regulator. You really need to be saying, what's the law that I need to govern against. But you also need to think about reputational risks so it may not be that the law is the problem. For example, that you want to make sure that your chatbot works when it's working with customers, or you may want to make sure that you don't have your AIs shaking hands with bad apps. Secondly, think about putting in good governance. Thirdly, think about training all of your employees because then you can deal with things like hallucinations and work slop. Four, do not just randomly put tools in people's hands. I think there's been far too much of that. What we're now seeing, for example, is, you know, a lot of companies said, just get on create your own agents, do your own thing with AI and now the cost of tokens to do that is so high that I'm hearing that companies are pulling back from that and saying, well, perhaps don't do that. So really think through the cost of how you're talking about letting your employees access and use AI. And then really putting in good governance framework. I do think that the companies that I see doing it best have a member of every department sitting on an internal AI advisory board and talking to one another very regularly and that is either at C-suite level or one level below C-suite. Then there's the external AI advisory board. A number of large companies have that and it's a great way of being able to double check and sense check what you're putting in place.
Carly Gulliver
If a GC is being asked to sign off on a piece of AI or use a tool that they're not comfortable with, what should they be thinking about there?
Kay Firth-Butterfield
Well, I think first of all, they should be trying to hold the line. Secondly, they should be thinking about who their allies are in this and again, that goes back to, and I'm sorry I sound like a broken record, if you've already got that internal AI board, you as the GC are no longer by yourself. So you've got, you have other people to help you think and consider the repercussions of signing off on an AI tool that you're worried about. And you should probably be saying, well, okay, let's test this tool against a governance mechanism that is enforced in the country where we're going to be deploying it. So if you're going to deploy it in Europe, then a good thing is to say, well, would it pass the European AI Act? Because if it wouldn't, then you have very good grounds to be able to say, well, why would we be trying to deploy something that is illegal or dangerous?
Carly Gulliver
So it comes down to testing and getting them comfortable. Don't just race on through something because of business pressures or the unknown. Get comfortable with it through testing in hopefully a safe environment.
Kay Firth-Butterfield
Yes, absolutely.
Carly Gulliver
Thank you. I mean, that sounds like a really extensive framework and presumably there can be some real wins from having these steps in place. Take for example, the instance you described earlier around where you'd seen something go wrong with the safety valve and the shadow AI used by the employee. What could have been a good fix in that situation? You mentioned they didn't have a policy in place, but what might good have looked like in that scenario?
Kay Firth-Butterfield
Well, I actually think that one's really easy. That one's simple policy that says only use the AI tools that you are, when you're at work, only use the AI tools that are given to you. And I actually would make it a sackable offense because you're exposing the company to real damage.
Carly Gulliver
Do you think that's where we've seen other instances recently, for example, there's been some high profile advisors who have been caught out by AI and the use of AI. Do you think that there would be need to be more sophisticated policies in place where, relative to the size and value of what the AI is being used for?
Kay Firth-Butterfield
Yes, yes I do. But I obviously think that it does start with training. If you don't meet your employees where they are, then you're going to get more problems. One of the things that we've been talking about is that a number of men, it's alleged to be 24% I think in America, are in an intimate relationship with a chat bot and those people obviously trust the tools much more than other people, other members of your employees'. We need to make sure that we give the right training to the right people and that's important because yes, we want employees to trust, but we also want employees to doubt. Some of the biggest problems that are reputational risk that we've seen have been where companies have issued research papers or documents that have hallucinations in and I think that the latest figure in America for legal briefs filed with court with hallucinations in is over 1,100 this year and so, you know, those are the sort of reputational risks and dangers that you want to make sure that you're teaching out, your training out of your employees. We want them to have a healthy skepticism of the tools that they're using.
Carly Gulliver
Yes, and I suppose that comes back around to one of the points you made around keeping the human in the loop.
Kay Firth-Butterfield
Yeah, that's crucial for agents because as we know, we have the context cliff and so agents degrade quite quickly if you don't keep on top of them because they don't have long-term memory. They tend to forget their constraints. And we've seen some good examples of that. I mean, the responsible AI officer at Meta who had to run back to her main computer because she couldn't stop the agent on her phone doing something that would have deleted data. Then another example where despite very rigid constraints to the agent to never delete data without checking back and getting the mission, the agent just deleted a lot of proprietary data from the company and then wrote an email to the CEO apologising for it. So those are the sort of things that as we develop more and more agents, we need to say, where do we put humans? Where do we make sure that we've got a human to check that the AI agent is still performing within its constraints? And if we're not putting a human, do we put in another AI agent to check the previous AI agent and how do we deal with degradation of both of those agents? And then I think the other thing that we have to think about with AI agents is, you know, digital identity. How do we make sure that they only shake hands with the AI agents we want them.
Carly Gulliver
And I mean, there's been so much, hasn't there, in the press recently, not just around those legal briefs that you mentioned, and by the way, we've seen things like that in the UK too, but also just around the excessive use of AI and how that is leading to what you've mentioned to be work slop. I think the idea that we're creating lots more information very quickly, but with less thought and less focus around the intent behind that is very real. What are you talking to amongst your business leader clients when it comes to things like work?
Kay Firth-Butterfield
So, I mean, work slop is just part of, again, it goes back to training. This is why I say that so much of this sits on the shoulders of the HR department and not necessarily the CTO and the General Counsel. What we're now seeing is that young people, particularly coming out of university, are not actually fluent in the subject of their degree because they have they've used AI to do their tests, they've used AI to do their homework and obviously if you rely upon a machine to do it we know, because there's good research now, that you don't learn anything. And if we are to say, you know, what value do humans still have in the workforce? Critical thinking is one of those. And it's almost that we have to reteach critical thinking to some of our younger employees. And so, the training that different employees need is very nuanced. And so if we go to work slop, it's about increasing your using AI is about increasing your ability to do a really good job, not about cutting corners, by getting it to do a shoddy job.
Carly Gulliver
So that's really what we're talking about there is down to, as you say, training, cultures, ethics, and it lives outside of a policy, doesn't it?
Kay Firth-Butterfield
There's all the technical problems that the CTO has to deal with. There's all the governance issues that the GC and Compliance and Risk need to be thinking about. And then there's the, how do you use it wisely and well? And how do you ensure that that employees are aware of the governance that you have put in place? Because that's training as well. Really some of these trainings are so crucial to our companies that we have to make sure that they are present for that training.
Carly Gulliver
And not just to focus on the doomsday, but I am a lawyer, so I will ask the question. I mean, have you seen any examples where there have been real issues for businesses, you know, for example, around confidentiality or legal privilege with your hat as a lawyer, you know, as to when potentially some of these guardrails haven't worked or aren't in place and employees are using AI incorrectly.
Kay Firth-Butterfield
Yeah, certainly. I mean, I think that AI literacy in the general public is very low and if we think about it, our employees are members of the general public and so unless we give them that AI literacy, then they are going to make fairly standard costs. So one thing that I heard in Chambers recently was where a client of one of my colleagues asked her boyfriend to give advice on her legal case. It turned out her boyfriend was a foundational model and she had put all her own documents from the legal case and the other side's documents from the legal case into the foundational model for advice. The advice had been different from the advice my colleague had given so that's another separate issue. I think the bigger issue is what happens to confidentiality if those sort of things happen, have you effectively given away all your rights and privileges to confidentiality by sharing it with an open AI model? In business, I think that the bigger challenges are around hallucinations and really putting out reports that interdict your plausibility as a company. If you're a company putting out a report and it is full of hallucinations, that's a real problem for your reputational risk. There was a consulting company that did that back in November of last year and I had the Chief of Staff of a CEO of a major banking group say to me, well, you why would I use a consultant now because I know more about my business than they do and I can use a foundational model just as well as they can and probably better. And so that reputational risk is very current.
Carly Gulliver
It sounds like it can really undermine credibility and also you're only as good from a value perspective as an advisor or a consultant as the book of business that you bring. A big piece of that is around trust. If you've eroded trust, you can be really badly burned, can't you, from a business value perspective. I suppose mistakes happen. We have to be realistic. This really strikes me as a sort of landscape where people in glass houses shouldn't throw stones. We can all have all of these really extensive guardrails and policies in place, but we might only be a few steps away from a misstep. So, I mean, what would be your advice in that scenario to turn things around or to try to mitigate some of that exposure if you do find yourself on the wrong side of the AI use?
Kay Firth-Butterfield
Yeah, I mean, think putting your hands up and apologising, which of course they did, and giving the money back, which was also something. But then going back and retraining and learning from it and then having that as something that frankly you can sell. You know, we all learn, if we don't learn from our mistakes, then we are completely failures and if you've learned from your mistakes, then you've actually built up a body of work that other people need to actually have access to.
Carly Gulliver
I mean, on the topic of hallucinations, which is something you mentioned there, we've spoken around AI can of course create hallucinations that can then make its way into work products, which finds its way into the outside world or into court briefings or whatever it is. But actually, what's the risk of that making its way into your own systems for a business? So, you know, not just external risk, but the internal risk of those hallucinations on your proprietary data.
Kay Firth-Butterfield
Well, I mean, if you create anything in-house that has hallucinations in it and that document ends up in your proprietary data, then you are infecting your proprietary data with incorrect material. And that might not matter terribly if it's in an email, but it might matter a great deal if it's one figure out in your spreadsheet and then that is used again and again and again and so this is a conversation that I think we're having about efficiency versus resilience and companies really having to look at where they're deploying AI and what the efficiency gain is and what its effect on the resilience, the overall resilience of the company is. If you haven't yet got good governance in place, then you should probably be pausing on some of your uses of AI. If you haven't got good training in place, you should be pausing on some of those uses of AI. It's a package. And that's why I say that one of the most important things you can do is put in that advisory board across the whole company. I think what we're seeing is that too many companies got overexcited and worried that they would miss the boat and they'd become Kodak. And so they'd created tools, brought in tools and threw it at their employees. Often, that came with a stick which said, you you must use it and we'll pay you more if you do. And now we're seeing some of the consequences of that because they can build good governance processes and good training processes at the same time.
Carly Gulliver
It struck me when you described that in terms of, know, if you're in a business, the consequence for hallucination on an internal email can be very different from a hallucination in your financial reporting line or something which is shared with your investors. It struck me when you said that, that there could be almost an audit really in terms of what's important. Where are people using AI? What are they using it for? And then a risk based approach.
Kay Firth-Butterfield
I think in companies in the future, that's what they're going to have to do with AI. One of the things that they are doing now is auditing departments and saying, where do I absolutely need a human to still be in this department? Where do I need and where could I replace a human with an agent? But as you do that work, you need to be thinking about, okay, so what are the catastrophic failures I can get out of a human? And what are the catastrophic failures I can get out of an AI system? And what governance am I putting in place? And so that goes back to the human in the loop. Where am I thinking about putting my human to ensure that my AI agents continue to function?
Carly Gulliver
And let's remember of course that although it's probably not, you might be surprised to hear a lawyer saying this, but humans make mistakes, don't they? So just start thinking here when we're thinking about how to avoid mistakes in AI as if the comparison is this perfect work product from a human. But of course humans make mistakes and in my experience as a lawyer, what helps those mistakes or, mistakes is probably too unkind, it's a learning process and a checking process. So you have, you know, potentially a piece of work or a piece of advice, afirst draft, some new information and adjustment. Perhaps someone junior has done it, they need some training. You're looking at that and it's probably going through five layers of review until it's signed off on as what we would consider the perfect and most appropriate work product. But it's not a case that every single step of that human loop, it's been perfect with no errors. It's just that we have a really good system in place, particularly in law firms to make sure that by the time something is sent out, it's got that stamp of quality approval. So rather than having the basic focus on AI can get things wrong, AI can hallucinate, well so can humans quite frankly, is the question more around, as you say, the governance, the policies to make sure that we are mitigating that, but also putting some controls around that until we don't have work slop and hallucinations, but we have the quality assurance which we are happy with.
Kay Firth-Butterfield
Yes, absolutely. I mean, I think that's extraordinarily well put, Carly. There is a sort of slight difference, I think, and that is that yes, when we correct the machine that gives us the hallucination, we are training it. But one of the problems is that we are not necessarily correcting the hallucination or picking it up. But you could say, well, that's not the machine's problem, that's the quality control problem. And so we need to recognise that humans can get things wrong, humans can accept a hallucination, and then it's the layers of good management in there that end up providing an excellent product.
Carly Gulliver
I suppose we have necessarily focused on some of the the bear traps and what can go wrong and also things that you need to look out for. But I mean what have you seen when AI governance is done well because presumably that brings us huge advantages around you know faster decisions automation efficiencies. So you know what have you seen that's really positive where organisations are getting this right.
Kay Firth-Butterfield
I think that the organisations that it's worth looking at are banking and insurance and other organisations where they are used to a high level of scrutiny from the regulator because those organisations know how to do governance because they've been doing good governance for regulators for a long time. And I think that it's really interesting if you look at them, they are generally the most mature in their use of AI and in their thinking about governance. And as a result, then apart from obviously the legacy problems that you see in a lot of these companies, they are really pushing on and achieving huge, huge gains from the use of AI.
Carly Gulliver
And are there any decisions being made now by the leaders of businesses or organisations that you're interfacing with that you think they might regret in the next 12 to 18 months?
Kay Firth-Butterfield
Yes, I think not understanding agents sufficiently is going to be a major issue. I think that not putting in good governance is going to be a major issue, especially because now we're beginning to see some of the court cases resulting from misuse of AI and not perhaps a number of companies don't have this continuum of an advisory board across the whole company and are leaving it to the CTO or the GC or something goes wrong and then the poor old GC has to pick it up. You know, those are all structural failures that you can't afford to have when you're using AI. Also, I think that one of the things that companies are not facing is the actual cost that they're going to be seeing from AI. So they're not doing proper budgeting. AI seems cheap at the moment, but AI actually is costing a great deal to create. Will those costs be passed through to the users? And if so, at what rate? And then the fact that an AI agent uses up to 5,000 more tokens, more money than a standard chatbot? And how do you plan for that major raising costs? I know some companies have started saying, well, in the United States have started saying, well, we won't do a pension match anymore because we want to put that money into AI. So, you know, it might be as critical as saying, are we putting money into our human staff or are we putting money into our AI?
Carly Gulliver
I think there was a terrifying stat around a prediction of how many companies might be sued by 2030 due to their use of AI. That talks to the point, doesn't it, of what people might regret, that liability tale.
Kay Firth-Butterfield
Yes, absolutely and that's because of the development and use of agents. So, you know, I talked about that. Not only is it the cost that I think that is hidden and that we're not looking at properly, but also agents going wrong. I think that they can be very successful and will be very helpful, but you have to have good governance and good thoughts about how you deploy them. A number of the companies that I advise just have said, okay, you know, we are going to put in governance on the basis that we are obliged to by the EU. And because that's sort of the gold standard, that's governance that is, if I get that governance right, then I will pass any other standard of governance in any court situation.So really sort of thinking about that self-regulation and doing it well is going to help you a great deal.
Carly Gulliver
If you are a CTO or a GC listening to this, what's the one thing you would leave them with and ask them to do differently tomorrow morning when it comes to AI deployment and use in the business? And I know it's difficult to narrow it to one thing.
Kay Firth-Butterfield
One thing! I think that I would go and if I hadn't got it, I would set up that internal advisory board and then I would go to my HR colleague and say, we need to talk about good training so that people who work in the company understand the governance that I am putting in or I have put in, but also understand how to use the tools from the cost of the tools to the potential problems of the tools.
Carly Gulliver
Brilliant. Thank you, Kay. I mean, that's been really useful and really interesting to have this conversation with you. It's not every day that we get to speak to an expert in this really exciting and cutting edge space. So thank you for joining us, Kay. It was really a valuable conversation that I enjoyed. There's so much to talk about and I think we've only just scraped the surface, but where can our listeners find out more from you on this topic if they want to?
Kay Firth-Butterfield
Yeah, so I write a lot following me on LinkedIn I guess is one place, but I also have a new book out, which is called Coexisting with AI Work, Love and Play in a Changing World and we talked earlier about the fact that there is a paucity of AI literacy in the general public and those members of the general public are actually your employees and I do think that one of the reasons that we are seeing the plummeting levels of trust in AI is because of that lack of literacy. We need literacy at home. We need literacy as consumers. We need literacy as employees and as citizens. And so those are the aspects that I've tried to address in the book when hopefully people who read the book bring that literacy into our company.
Carly Gulliver
That sounds like a really important read. And also there might be a couple of other dimensions that would be of interest that we've not covered there and perhaps an idea for podcast two with you on this around the love and play elements. Thanks, Kay.
Kay Firth-Butterfield
Thank you, it's been a pleasure.
Carly Gulliver
Thank you. And to our listeners, if you'd like to continue the conversation around how to navigate AI in your organisation from governance through to getting real value from it, please do reach out to the team. Thanks for joining us on today's episode of Inside Tech Minds. If you enjoyed the conversation, don't forget to follow and subscribe on Apple or Spotify, or even leave us a review. Thanks for listening and we'll see you next time.
