Hello, and welcome to today's podcast. In today's episode, we'll be focusing on artificial intelligence in the food and beverage sector. From boosting efficiency and cutting costs to reducing reliance on manual labor, AI is bringing some exciting benefits. But AI isn't just another emerging technology; it's more powerful and more far-reaching than anything we've seen before, and it will change the risk profile of businesses across the food and beverage sector.
I'm really pleased to be joined today by my colleague, Sam Haslam, Head of our Risk and Resilience Advisory Practice, and Chris Matthews, Managing Director of Svella AI Solutions. In this episode, we'll explore the potential uses and benefits of AI across the sector and how you can stay in control of your risks as the pace of innovation increases.
Sam, can we start today's conversation, please, by considering the ways that AI is currently being used in the food and beverage sector and the benefits that businesses are already seeing?
SAM HASLAM: Sue, yeah, great. And firstly, thank you for having me on this podcast. I think it's an absolutely fascinating topic, and myself and Chris love speaking about this and just researching, learning a bit more about the applications to food and beverage, really interesting.
And to start answering your question, first thing I want to do, take a little bit of a step back and think about what we mean when we're talking about AI and do a bit of an imperfect split between two categories, really, first one being what we can call general purpose AI models. So, thinking about your ChatGPT, Google Gemini, Copilot, all of those, which a lot of people are using, and the rates of use are increasing massively.
But we also have what we can call bespoke AI models, and these are models which maybe have gone under the radar a little bit. They use the same machine learning techniques, but maybe applied in very different ways by individual food and beverage companies and can have some really transformative impacts as well.
So let's focus on those in terms of how they might be being used at the moment. Examples that I've seen include use in demand forecasting and inventory management. So that might be looking at historical sales figures, looking at weather patterns, looking at social media sentiments, stitching together all of those massive data sources and using that, using a machine learning model to forecast demand in a way which a human without that AI assistance never could.
It might be looking at waste reduction. I saw one example of a global food manufacturer who reduced waste by 87% using things in production environments, such as computer vision, where a computer model, an AI model, can look at what is happening, identify things like foreign objects on a production line, or bruising on fruit which a human eye can't see, and apply that to identify what might otherwise be wasted or might not be sellable, and do things distribute that to a charity.
So even outside of the massive impact of generative AI on your chatbot models, the bespoke models have a massive impact as well, and they can lead to massively increased efficiency or more creativity when used correctly. Cost savings, yes, but a wide range of benefits.
SUE NEWTON: That's really interesting, Sam. Thank you. And Chris, can you give us your view, please, on the art of the possible, and where AI might take the food and beverage industry in the future?
CHRIS MATTHEWS: Yeah, thanks for the invite. Great to be here. For me, the future of AI in the food and beverage sector, you could probably sum it up in a single phrase: intelligent ecosystems. So it's not just about the automation. We're starting to see connected kitchens in AI, driven forecasting, and autonomous production environments that run with minimal human intervention. And that's not to say that we're going to be doing away with the humans. The advantage is that humans can now do the more high-value tasks, the real thinking tasks that the AI will not necessarily be able to do.
I think in the future, we'll see a massive rise in predictive intelligence. That'll become the norm, I think as Sam mentioned, forecasting customer demand from a production line point of view, predicting equipment failure, looking at perhaps the wider global economic picture and seeing where ingredients shortages, for example, before they disrupt operations, which is going to be a huge advantage for suppliers where I'm imagining that the loss or the having to stop a production line is incredibly expensive.
And I think another couple of perhaps less obvious points, we're going to accelerate customization. Food production, we're almost going to be able to lab-grown ingredients tailored to individual nutrition goals. So dare I say, the protein-saturated broccoli for the bodybuilding community or a particular wheat that's grown without the gluten product. So that feeds a particular a growing gluten-free market. So this hyper-personalization I think, is going to have a massive impact on the industry as a whole.
And then finally AI-powered sustainability tools. Again, as Sam mentioned, cutting waste is huge. And optimizing energy use, tracking carbon, using AI tools to track carbon right the way through the supply chain. There'll be blockchain, which is another new and fairly emerging technology, where we'll be able to track the whole supply chain from farm to fork, all using blockchain. So on an uneditable inventory of everything that's happened to that product, from it coming out of the ground to being turned into, let's say, a loaf of bread.
SUE NEWTON: Great. Thanks, Chris. And you mentioned there about changing roles, the human is still being involved in changing roles. But I think it seems likely that there will be an impact on jobs in the sector and the number of people required in the sector. How should businesses be thinking about that potential impact, Sam, and what they need to do to help employees through those changes that are on the way, really?
SAM HASLAM: Yeah, I think it's important for businesses to have an open and honest conversation about this and appreciate the nuances as well. I don't want to undersell the challenges that are likely to be encountered in waves of automation that we've just been talking about. But if you take a step back and look at the trends, I think it's almost impossible that there is not going to be an impact on jobs, and we need to be having that conversation and analyzing it.
So let's look at, firstly, maybe office-based jobs. I think a reality there is that AI literacy, which means understanding how tools operate, what AI tools you can access, and what you can do with those tools, that's rapidly becoming an essential skill. Businesses need to make sure that people who are using those tools have access to reasonable tools, which can unlock the benefits. They need to ensure people know the capabilities, the risks and how that might impact them.
And to be honest, my experience is that if you use those tools that we can all have access to with the right conditions right now, it can give you superpowers. It can unlock really good benefits as long as you manage the risks.
I think if you then look at the factory-based jobs, manufacturing-based jobs, for example, I think an awareness of the increasing capability and the decreasing cost of tech, which is directly going to impact those environments, is essential for businesses.
I mean, one thing we've not mentioned yet on this podcast is robotics and the application of artificial intelligence methods, such as machine learning, computer vision, to robots. And there is a likelihood, in my opinion, that that is a field which experiences massive growth in the next couple of years. Now, that clearly has a direct impact on factory-based jobs.
So I think awareness from a business perspective, honesty and investment in AI literacy and upskilling, not just buying people ChatGPT or Copilot licenses and saying off you go, but investing in understanding people, understanding of your people and how the tools work, the risks and the opportunities as well.
SUE NEWTON: Thanks, Sam. And Chris, I think it's fair to say that larger businesses are more likely to currently be using AI due to resource and capital availability. What do you see as the current position for smaller businesses, and how you see AI adoption for those smaller businesses in the future?
CHRIS MATTHEWS: Yeah, I think historically AI has been seen as a big company technology but that is changing fast. No two ways about it. Cloud-based AI as a service tools, OpenAI, ChatGPT, for example, Microsoft Copilot, which a lot of enterprise using, is now financially very accessible for the smaller businesses and gives them access to capabilities that even 12 months ago would have been the reserve of massive enterprise, multi-million-pound enterprise, multi-billion-pound enterprise, very often in food and beverage.
So yeah, it's a real leveller. It really helps level the playing field for smaller food and beverage operators. I mean, for example, affordable automation through things like AI-driven agents, which are relatively straightforward these days to create from scratch, managing stock, better delivery, customer service, all without adding to the headcount.
So again, we take the repetitive tasks and we get the machines, the AI to do those repetitive tasks while leaving the human in the loop, capable of doing the more or less mundane, higher value, potentially tasks that previously they would be distracted with.
SUE NEWTON: That's great, thank you. And Sam, can you give us some thoughts on what businesses need to think about in terms of risk and their changing risk profile as AI is introduced and what they need to do to make sure that the governance frameworks keep pace with the speed of introduction of AI?
SAM HASLAM: Yeah, I think there are two dimensions to this. And I think personally, from my experience working in a risk advisory practice, looking at risk registers quite regularly, it's not something that I'm seeing across either of these two dimensions being addressed particularly well or regularly by organizations.
So the first one of the two is brand new risks arising due to this technology to AI and associated technologies. I would expect quite a range of risks within risk registers at the moment, looking at the different aspects, yes, the technology itself and the cyber risks, for example. But also the people impact, the commercial impact, the competitive impact, those things need to be considered carefully, as does the second dimension, which is massive changes in known and existing risks due to the impact of AI.
So let's look at a quick example of that one first. And that's in the emergence of very, very high-quality image and video models. So if you're a food and beverage organization with a recognizable, well-known brand, we're already living in a world where, for almost no cost, an individual can put your product in any scenario, whether that's in an image or in a video, they can create an image which might show your product in a negative light. And that is so, so easy to do. It's an order of magnitude easier and cheaper than it would have been even 12, 24 months ago, that has implications.
And on the first one, if we think about the societal response to AI and the trends that we're seeing and the jobs question you asked me about before, well, if an organization is intending on leveraging this technology to, for example, do a massive automation in a factory environment, what's the impact on a potential public backlash against that? Is that a brand-new risk? Is that something that's ever been considered before? All of these things need to be considered in terms of that change in risk profile.
Now, the second part of your question was around controlling that, and the governance frameworks and how to approach it. A couple of things I would recommend here. Firstly, have a look at what you've already got in place in terms of risk governance and ask whether that is fit for purpose. If the things that I'm talking about here are completely new to the organization, then that indicates to me there is potentially already a gap because existing governance processes haven't picked up these new and emerging risks.
So then secondly, we need to adapt them and to make sure that the processes we have in place are suitable for rapidly changing technology. And this means firstly, principles which are relatively static and which don't change a lot often within your policies, within your governance, but also specifics around the current environment. All the stuff we've talked about, how are you specifically addressing that within your governance framework?
Final recommendation on this one, set up a specific body responsible for tracking and responding to emerging risks. That includes our topic today, but it also includes a wider range of emerging risks, things which are changing quickly in the modern risk environment and which need a specific and direct response.
SUE NEWTON: Great, thank you. And do you see AI changing the way risk is defined or measured? And do you have a view on the future role of risk managers and how AI is potentially going to impact that risk manager role in the future?
SAM HASLAM: So the question around the definition and the measurement is a really interesting one. And I think on balance, my answer here is no, but only if you've already got your definition and your measurements right.
So a good risk framework, a good way of approaching and looking at risk will already look at the things which are crucial to achieving your strategy and at WTW and my practice, we call those things value drivers. It'll define those things and how to measure risk to those things, and then it'll do things like setting a risk appetite accordingly.
If all of that's working and it's in place already, these AI risks we've been discussing today in AI impacts will simply slot into that framework. Though, I do personally believe they're likely to often be very high-rated risks, it shouldn't necessarily change the way that we define and measure risk.
Second question that you asked as part of that was the impact on risk managers. And I see this in three phases. Right now we are at the back end of the risk of the AI toolkit for risk managers. So having access to tools which radically increase capability, ability to do amazing things with large data sets, for example.
And we're moving into the next phase, which I see is an AI assistant, where there's increasing capability of AI systems to have some degree of agency and to go about elements of the role of a risk manager, but still under the clear direction of that risk manager. And there's still parts of which, of course, I can't just simply do at the moment.
I think all of us in the risk field need to have our eyes open to the possibility of a future where large elements of the current job of a risk manager can be increasingly automated, enhanced, and controlled to an extent by AI systems in the medium to long-term and need to think about what exactly that means in terms of the role of a risk manager.
I completely think that a human in the loop or over the loop is absolutely required. But I think as risk managers, we should be thinking about how we can leverage this technology to cover a wide range of extra risks in greater depth, and essentially do our roles much, much better.
SUE NEWTON: That's a positive. That's a positive view. Thank you. And as we close this, would you each give us a view, please, a piece of advice you would give to a business in the food and beverage sector. So your best piece of advice over and above the things we've already talked about, I guess. Chris, would you like to give us your thoughts?
CHRIS MATTHEWS: Yeah, definitely. And really interesting observation with Sam there. So the human over the loop is a great terminology to use, because AI, while it might have the knowledge, it's not replacing us any anytime soon. I think everybody should take that positive from this podcast, if nothing else.
So yeah, in terms of the final piece of advice I'd give, it's in four points. I've helped businesses over the last 18 months implement and transform areas and sometimes entire businesses using AI tools and my experience tells me is start small, scale fast, and prioritize the impact.
So first, focus on quick wins that are going to deliver a measurable ROI. So show quality control, predictive maintenance, or sales forecasting they'd be for your particular sector that we're talking about today. So those would be the three real quick win potential areas.
And then build a trusted data foundation, so data is the new oil. To coin a Bill Gates phrase, AI is really good, but if you feed it useless or incorrect or rubbish data, its accuracy and its integration into your business is going to be very, very challenging. So it's labeling and creating that data set that's going to be super important for any AI project.
The second to last is align the adoption with the capability. So invest with human capability. So investing in your workforce bringing teams on the journey, don't just foist AI on them. Help them understand that it's not about replacing them, it's about giving them an amazing tool. I always joke about that it's like having a four-year-old with a PhD that you decide. It's got lots of knowledge, but it does need reining in sometimes and helping and being told where to go and the direction you'd like it to proceed.
And then finally, and I think, above all, treat AI as a strategic enabler rather than just a technology upgrade. This technology does not belong in the IT department. Well, they may help with the technicalities of rolling it out. AI teams can sometimes sadly become the progress prevention department. And I would say having this at the strategic level, it's got to be led from the top for these projects to work. And I think it is going to define the next decade of growth across food and beverage.
SUE NEWTON: Great. That's a great advice. Thanks, Chris. Sam, piece of advice from you, please.
SAM HASLAM: So I've got two. So I'll try to be quick with these. My first one is conduct an analysis of how prepared you are for the impacts that AI is going to have across various aspects of your business. And I would keep this piece of advice the same, whether I was advising a start-up business, still producing food and/or beverages in a personal kitchen versus a global organization. It is going to impact you in various ways. Yes, in terms of your production environment, but also back office, people, customers, strategy.
I would recommend organizations research some plausible, highly disruptive scenarios for the next three to ten years, for example, and just stress test based on your current plans how you are looking at things right now. Do the two marry up? Are you prepared? Do you have an awareness of what is potentially going to happen in the next few years? Are you ready to face the risks? Are you ready to leverage the opportunities? And just doing that and use AI tools to help with that exercise will position you really well in terms of making sure you've got resilience and you can adapt.
Second piece of advice, and it's a simple one, invest in AI literacy amongst your people. Educate your people rather than simply either burying your head in the sand or simply handing them tools and saying, off you go. Give your people the skills and knowledge that they need to flourish with the new tools that they are able to access now. And that will help you, yes, mitigate the risks, but also to seize the opportunities.
Again, going back to my personal experience, getting to grips with emerging AI tools in the workplace, using Microsoft Copilot. When you're able to use this well, it feels like you've got superpowers, and it has enabled me to focus on that higher value, more enjoyable parts of my role, which can only be a benefit. So that would be my advice.
SUE NEWTON: Super. Thanks, Sam. And that brings us to a close. That's the end of today's podcast. Thanks very much to Sam and to Chris for your insights and to you for joining us. If you enjoyed today's discussion, please subscribe, rate, and share this podcast. Thank you, and goodbye.
Thank you for listening to this podcast from Willis, a WTW business. The information in this podcast is believed to be accurate at the date of publication. This information may have subsequently changed or have been superseded, and should not be relied upon to be accurate or suitable after this date.
This podcast offers a general overview of its subject matter. It does not necessarily address every aspect of its subject or every product available in the market. It is not intended to be, and should not be used to replace specific advice relating to individual situations.
We do not offer, and this should not be taken as legal, accounting, or tax advice. If you intend to take action or make any decision on the basis of the content of this publication, you should first seek specific advice from an appropriate professional. Some of the information in this publication may be compiled from third-party sources we consider to be reliable. However, we do not guarantee and are not responsible for the accuracy of such.
The views expressed are not necessarily those of Willis Towers Watson. For further authorization and regulatory details about our Willis Towers Watson legal entities operating in your country, please refer to our Willis Towers Watson website. It is a regulatory requirement for us to consider our local licensing requirements.
