AI has dominated the headlines this year and is the cause of seemingly bottomless investment in markets like the US. This is technology theatre at its finest, accompanied by flashy presentations and even flashier share prices at the world’s biggest tech companies.

But what impact is it actually having on the corporates on the ground who are the eventual users of the service?

To unpack where AI works (and doesn’t) in the real world, Shane Cooper (Head of Digital Advisory at Forvis Mazars in South Africa) joined me for a frank discussion. We covered a number of topics including:

• AI as the “smartphone moment” of this tech era
• The key considerations for executives in AI implementations
• Global and local examples of AI implementations in companies like Zeiss, Sasol, Siemens, Walmart and Shoprite
• How SMEs are viewing and applying the AI opportunities
• Disruption to the internet economy as we know it
• Why AI is still “hallucinating” answers to questions
• The near-existential questions that AI poses for humanity
• How to think about your job in the context of AI’s strengths and limitations

Buckle up. This is a powerful discussion about probably the most important business topic in the world today. You can connect with Shane on LinkedIn here.

Listen to the podcast here:

Transcript:

The Finance Ghost Welcome to this episode of the Ghost Stories podcast. We’re recording this after there’s been a lot of news flow in the market around AI, although it feels like I could probably say that every single week.

We’ve seen some really big deals between the likes of Nvidia and OpenAI and AMD, and lots that investors have been looking at. And we keep reading about this stuff – we keep reading about AI. It’s everywhere at the moment. I think at this point, if you haven’t at least tried out some kind of LLM, you probably are at serious risk of making yourself redundant in a world where it’s going to be hard even for those who do actually put in the effort to equip themselves with AI knowledge. Please don’t be the VCR in a world of CDs and then DVDs and then streaming – gotta stay on top of these things.

And to help us understand what is actually going on out there in this world, we have Shane Cooper. He’s the head of Digital Advisory at Forvis Mazars in South Africa. Shane and I have not done a podcast before together, so this is Shane’s first time on the show, and I’m very excited to have him here with me today and we’re going to talk about how companies are thinking about this new world and what’s actually going on out there.

Shane, one of the terms that came up is “technological theatre” – what a thing! I love that. I guess today we’re going to understand a little bit more about this big theatre production and especially what it could actually lead to and what the useful things will be. Thank you for your time on the show.

Shane Cooper: Thank you, Ghost, for inviting me and it’s a great pleasure. I’m sure we’re going to be covering a number of critical issues today with your audience.

The Finance Ghost: Yeah, absolutely. I mean, this stuff is literally everywhere. As I said, it really is incredibly important.

I’m keen to understand a couple of things – I guess one is more your background and how you got into doing the digital stuff. That’s always interesting. But then on top of that, why digital advisory services are actually such a natural fit at Forvis Mazars, why this is part of your service portfolio?

Maybe let’s do some of the setting the scene here, a little bit about your involvement, but then also why this makes sense at Forvis Mazars in South Africa.

Shane Cooper: Perfect. It probably doesn’t feel natural for an accountant to be talking about tech. I’m an accountant by trade, but when I finished my articles in the early part of the century, I moved into tech pretty much immediately. So I’ve been in tech for 20, 25 years – the last 10 years been part of the Vodacom Group, led strategy there for half a decade, and then was part of one of their startups in the Internet of Things space for the last five years.

And during my time there, what I began to understand is just how organisations are battling to deal with generating value out of data. But coupled with that, we have this emergence of what I guess you could call the consumer version of AI with the launch of ChatGPT – what is it now, well over two years ago? And I think for the very first time we’ve had a significant form of technology find its way to consumers first. We’re now in a position where organisations are looking to embrace it and find ways to get value out of it. And you’re right, the “tech theatre” that you referenced at the beginning of the podcast is exactly where we are.

I mean, I’d love to talk about things like whether we’re going through a period of significant hype or whether there is a bubble, I’ve got some firm views on that.

But tech for 25 years, there is no debate that at the minute the most interesting part of technology around the world today is Artificial Intelligence.

The Finance Ghost: It feels like the biggest thing since the smartphone. I always think about the iPhone and the iPhone came out in 2007 – I was at university, I remember, and it was only the very rich kids at varsity who managed to get one or two of these things from overseas. It was quite a big deal. The rest of us were on Blackberries with broken keypads trying to BBM each other. Those were the days!

And things change, things change really quickly. I mean, just look at what’s happened in not even two decades since the iPhone came out. It feels like AI is that next big push.

And you can’t tease me like that on your views on the bubble without me asking you then, is this technological theatre going to be comedy or tragedy? How’s it going to end for investors? Or is it going to be a little bit of both? Because it feels very much like comedy at the moment when I look at what’s going on out there.

Shane Cooper: Yeah, it’s fascinating. I mean, you referenced the OpenAI investment in AMD and the Nvidia investment in OpenAI. So you have a circular reference where cash is simply moving from one organisation to the other.

The Finance Ghost: It’s that Spiderman meme where they’re all pointing at each other. All of them are Spidermans.

Shane Cooper: Exactly! And you have OpenAI having a conversation with AMD saying, well, how do we fund this? Well, all that OpenAI needs to do is mention AMD in the media and AMD’s share price is going to rise. There we go – you’ve got your funding for the acquisitions.

The Finance Ghost: Ta da! It’s amazing, right? It’s ridiculous.

Shane Cooper: Exactly! I don’t think there’s a debate that there is a bubble and absolutely there are going to be fingers burnt for sure. But I think the broader question is if you – and you referenced the smartphone moment with Apple, I think it’s probably a bit bigger than that. I think probably the more likely equivalence is either the internet era or even more seriously the invention of electricity and the railroads. We all know that the story about railroads, there was a huge investment, there was a bubble there and there were some fingers burnt during that process. But at least the rails were put down. You could argue now the data centres that are being deployed is useful infrastructure for future use anyway. I think the challenge here though is that the depreciation cycles for data centres are far shorter than they were for railroads. So that’s the challenge we sit in.

For sure there is hype. For sure there is a bubble. But is there value in the use of AI? Absolutely! And this is where I think the conversation today is useful in how organisations should be thinking about AI to bring benefit to the organization.

The Finance Ghost: I love the railroad analogy because it is the exact right one I think – it is about building out this infrastructure. But some important differences which you’ve also highlighted there around depreciation cycles, etc. Also, railroads didn’t constantly need energy just to exist. I think that’s a big question mark around data centres. Every time AI gives me a stupid answer, I feel extra irritated by the fact that energy was used to actually go and waste my time with this nonsense.

Shane Cooper: Energy. Water.

The Finance Ghost: All these things.

Shane Cooper: Yeah.

The Finance Ghost: But anyway, this is part of technological progress for us. The funny thing is, if you go back and watch like a sci-fi movie from 30 or 40 years ago and then their idea of “The Future” – the future is inevitably somewhere in the 2030s. I don’t know why that seemed to just grab their attention. Maybe it was, well, what will we be doing in 50 years? People jumping through black holes into deep space, I’m just trying to get AI to actually read a document and give me a correct number here in 2025!

It feels like we are behind where sci-fi would like us to be. And unfortunately there’s going to be a lot of growing pains with that. It means we need to invest in this stuff and we need to see where it all lands. It’s growing pains, right? There are going to be fingers burnt. I completely agree with you.

I think the value of this discussion is understanding what is actually going on at ground level because that’s the test of whether or not this is real. I pretty much ignore – obviously all the listed companies at this point in time, it’s the “turkey voting for Christmas” issue. They’re going to tell you that this is just going to keep going, that demand is enormous.

For listeners, if you really want to go and see the hype cycle play out, just go and look at Oracle and go and look at some of the targets recently put out by them. They’ve got a big investor day in Vegas – of course, where else? – where they will be giving all these extra targets to show how insane the growth is.

Be that as it may, I think you can bring a fantastic view here of what is really going on out there from a client perspective. Let’s move into that because real-life case studies, good examples of where this stuff is actually being done out there as a digital transformation – that’s always valuable. I’ve seen it a lot in the retail space, obviously – eCommerce, omnichannel, big data that comes through there. And it’s easy for us to understand as people because we are customers of retailers who have invested in this stuff. So that’s an easy example, I suppose, but there will be lots of other ones. So let me open the floor to you, to just take us through some really good examples of where AI is actually being used in practice.

Shane Cooper: So I think, Ghost, what is important for me is let me just set the scene quickly around why I’m at Forvis Mazars and what we’re doing, because I think that’s part of what will add more context when we deal with the use cases being deployed in practice today.

So for us as Forvis Mazars, we know that digital transformation – and we’ll talk a little bit about what digital transformation means – is essential in today’s world for organisations to understand and to embrace in order to remain relevant and competitive. Now for Forvis Mazars, we (together with some of the other larger audit firms) built our brands over decades on one fundamental principle, and that’s trust. It’s a highly regulated environment and for us, we obviously have to provide an opinion on financial statements, for which investors and shareholders place confidence in those numbers in order to make the right decisions.

In order to arrive at providing that opinion, we need to have an intimate understanding of the customer. It’s not just a superficial assessment of numbers on the income statement and balance sheet, but rather a deeper understanding of how an organisation functions and in particular the decision-making process that drives performance.

Now when you overlay that in today’s world of technology with Artificial Intelligence, this understanding for us on the levers that one can pull in an organisation is pure gold when you approach AI implementations, because we’re seeing this “tech theatre” play out where shiny tools are deployed and then you have executives, shareholders and board members disappointed by ultimately the outcome of those AI implementations.

Where we focus is on ensuring that the right strategy is in place, that the right value pools are targeted. And then I think very importantly, as you implement the AI solutions, you keep track of how the metrics that you’ve targeted are changing. Because if you’re not changing metrics, then you’re simply wasting your time and you’re simply enjoying a bit of theatre.

If I use that as a preface, I’ll jump into giving a few examples of where AI in particular over the last two years – because let’s remember AI has been around for decades and even if you look at AI in terms of the various elements thereof on Gartner’s maturity curve, you’ll see something like computer vision sitting on the far right-hand side of its maturity because that’s been deployed in particular over the last five years comfortably across organisations around the world. And then more recently you have the LLMs, which is where a lot of the focus is today and moving into agentic AI.

Let’s use a few examples. None of these are Forvis Mazars customers, I thought I’d keep it a little bit generic just to give a sense of how AI is being deployed around the world.

The first one, maybe a little bit close to my heart as I’m an amateur photographer, if you look at Zeiss, the lens manufacturer, they’ve deployed AI-powered inspection systems in 2023 already. And these deployments, all that they’re designed to do is to detect issues with products that they’ve manufactured. The reason why AI is deployed there is because you’re producing tens of thousands of products on an hourly basis, there’s no chance a human QA process can solve that. Historically, QA processes used to be on a sample basis, whereas now you do 100% checks with AI. They’ve trained their engines over years of providing it with images of products that are defective and they simply apply that and an alarm will go off and a product will be removed from the proverbial conveyor belt when there’s an issue that’s been detected. That allows for a significant improvement in the quality of the products that eventually roll off their production facilities.

In South Africa, a little bit closer to home, if we look at one of our investment topic favourites – Sasol. They’ve had an interesting deployment there and I’m mentioned them specifically in that they actually haven’t spent a whole lot of money on AI for automating the process of extracting information from emails. Now, I mention that because it’s a simple use case and it’s a problem that, if you think about emails been around for many decades, it still hasn’t moved on much. Information flows via email can sometimes be hugely problematic because you’re not picking up all of the required information that you need and there’s zero automation in the email process. So they’ve deployed an interesting project. This was actually driven by some champions inside of the business, where when they deal with the extraction of metadata from suppliers, they’re able to immediately identify where there’s missing information and an automated email gets sent back to say, we’re missing information. Now, ordinarily you would have probably hours of human intervention there to assess this, very often missing things, and a simple application of AI technology solves that.

Let’s move on a little bit, maybe back to Europe. Siemens, as an engineering firm, have developed what they call their MindConnect AI platform. They’ve actually deployed this across 50 of their facilities in 2024. And what they’ve done is they’ve moved on from prediction of equipment failure to optimising production lines in real time. Can only do that with AI!

Essentially what the engine has done over the years of its deployment is that it’s understood what are the attributes that eventually lead to equipment failure and how do we pre-empt this by employing various practices during the course of the maintenance process. And what this has done for them is they’ve seen some remarkable improvements in their efficiencies of their operation because they don’t have the typical stoppages of production facilities that one finds in highly mechanical production environments. If we move on, you mentioned retail a little bit earlier. There are two names in the retail space that one should always talk about. And I’m not going to talk about Amazon in this, I’m talking about Walmart and then Shoprite, seeing as Walmart may make its presence known in South Africa in the next year or so.

Walmart are an extremely forward-thinking organisation from a tech point of view. They’ve deployed technology with their partner Wiliot across about a thousand stores in 2024 and they’re using ambient IoT sensors which track every product from warehouse to checkout. And that’s interesting because now it’s a combination of using AI and edge technology to significantly improve inventory management. Many retailers in South Africa battle with inventory management. This is the holy grail – to understand what product needs to be placed on which shelf, where, when in order to optimise your working capital. And also it’s helping Walmart in optimising their store layouts in real time. So a really cool use case.

Closer to home, Shoprite – they’re obviously a bit of the darling of the retail space in South Africa and with their loyalty programme, I think the stats are that they process somewhere around 3,000 card swipes a minute on their loyalty programme. This is a huge source of amazing data for use in their technology hub. I understand that they’ve developed some pretty sophisticated AI and analytics tools to optimise stock levels, also to provide personalised offers which is really important in our – as a sample of one, I know when I log on to my Sixty60, I always have personalised offers based on my buying patterns.

The Finance Ghost: Shane, you’ve touched on a lot of really important points there, one of which obviously being the importance of just tracking metrics in these projects, just to take you right back to the preface to that discussion about some of those real-world case studies. I think that’s very important because we’re kind of in a theme at the moment of people are not quite sure what this tech will do and they’re throwing money at it. You’re given some really good real-world use cases there that make a lot of sense.

Stuff like quality assurance, taking out errors, taking out that human element, I mean that makes sense – people are not great at having a 100% rate of no errors. That’s what we’re bad at. So that’s precisely where we should be using stuff like AI to actually get involved.

I also enjoyed your reference to amateur photography. Adobe is one of the companies that I’ve been following closely. I mean they were the poster child for transitioning into Software-as-a-Service from way back in the day where you’d go to a store and buy your software off the shelf, then you’d start paying a monthly subscription. Adobe was one of the first that took us there and they did incredibly well as a result. But now their share price is really struggling because there’s so much uncertainty over: what does it mean for the creative industry with these AI tools? What does it mean for Adobe? To what extent do people really need Adobe?

Disruption is the thing that just keeps happening, right? That’s how it works in the world of tech and what you’ve referenced there is a lot of large organisations investing in what is essentially disruptive technology, but it’s also in many cases to protect their market share, to protect their leading positions.

Sometimes with smaller companies and challenger brands, the AI actually becomes a really good way to carve out a new market position. Suddenly you’ll see a new business emerge and maybe you’ll also see SMEs that can suddenly compete with the big guns because they actually have access to a system rather than needing an army of people.

I guess that leads into a conversation around the extent to which you are seeing AI implementations in smaller companies. I’m loathe to say small companies because I think most small companies are happy to just plug something into Copilot. Perhaps I’m wrong? Happy to be corrected if I’m wrong there. But certainly as you get into medium-sized enterprises, what are the sort of AI deployments that you’re actually seeing out there?

Shane Cooper: Yeah, I think that’s a good point. I think what we should all remember is that just over two years ago when ChatGPT arrived, it was almost a “where were you when” moment when you first discovered ChatGPT. For some of us in the tech space, it was that profound a moment. Now if you think about small- to medium-business owners, they in their personal capacity would have understood the power of the technology at their fingertips. And I’m for sure clear on the fact that those business owners would have thought about how best to use this technology for their businesses.

That’s what I mentioned earlier on, that you have this consumerisation of technology that now larger corporates are beginning to understand. And corporate friction in the large organisations is a thing. Deploying new technologies into large corporates is a challenge because there are a host of things that one needs to deal with – whether you’re talking about regulatory compliance or policy restrictions, and of course the good old fashioned change management – how is AI going to impact my job? Which we can talk about a little bit later.

In the small-to medium-space, you mentioned Copilot as a technology that organisations are using. And I wouldn’t disparage that. I think there is a powerful set of use cases that one can use Copilot for. Of course, what one has to be aware of, is that if you’re not careful with how you use it, you do run the risk of your data being exposed. Understanding when to use it and how best to use it is critical. I think for small-to-medium organisations, you do need to think about how you engage those organizations to embrace a more secure access to LLMs in order to protect your business.

But I mean, just on some of the use cases, if you think about the typical series of activities that various departments undertake on a day-to-day basis, there is a lot of repetitive activity that one can now use AI for. We’re all familiar with the use of robotic process automation, and now the application of agentic AI inside of the LLM space allows you to execute what we may be a little bit disparagingly referred to as mundane tasks that you can execute on an automatic basis. Whether it is the generation of invoices, whether it is the issuance of POs, whether it’s doing basic reconciliations of your bank account into your financial records – all of that stuff can now be done via a robot. And for us, what we see in this space is that what this does is it frees up the time to do better work around properly interrogating your information and making better data-driven decisions.

The Finance Ghost: This term “agentic AI” has come up a couple of times and I think there are so many fancy terms in this “technology theatre” that I am loathe sometimes to assume that people know what it is. So maybe just because it’s come up a couple of times, can you just give us the TL;DR on what “agentic AI” actually is?

Shane Cooper: So “agentic AI” is in the first word, it’s agentic, so it’s an agent. Essentially what you’re doing is you’re creating an agent to conduct an activity that you’ve instructed it to do. Whether it is to fetch an email, analyse it, understand it, give it context and respond to the email. That could be one agent. And then what you do is you stitch various agents together to fulfil on a task end-to-end. What I think is important just to make sure that if someone does want to embrace this, I do encourage you to insert a human in the loop element there, where there’s some review. As we know, AI does tend to hallucinate and I think that’s important to understand. We’re all hoping that hallucination will decrease over time, because from what we understand over the last few weeks, the designers and builders of these LLMs have now finally understood why hallucination happens.

The Finance Ghost: Well, that is exciting, because then we can stop getting such nonsensical answers from time to time. I don’t quite remember exactly where I was when ChatGPT came out, but I know that I was “navigating the landscape” of a whole lot of articles that suddenly had the same starting sentence. So I think that was when it became apparent to me that everyone was starting to use this thing, which as a writer is just very painful to get through. But it is what it is

I enjoy the concept of stitching agents together. This sounds a lot like The Matrix, which also had a human in the loop at some point. So this feels like the world we’re going to, just go watch a bit of sci-fi and you’ll get there.

Jokes aside, the thing that worries me, and I think that worries a lot of people, is what is going to be the real impact on employment here? Because it is lovely to think that humans are all going to be way more efficient and we’re going to do these fantastic things, but it’s also not the case that we’re just going to flick a switch and suddenly say, oh well, we can do everything we do as a species now, but we can do it with half the number of humans. The other half are still there. They still need to be gainfully employed, they still need to earn an income, they still need to do something. And I think this is the ethical discussion that is honestly only just warming up. I don’t think we’re anywhere close to understanding where that might actually end up and what it might do.

I’ll tell you my theory and then you can tell me yours. My theory is that if your job has an exact right answer, then you’re in trouble. So if you don’t have to make a lot of judgement calls, if you don’t operate in the grey, if the answer is white or black, the answer is 1 or 100, then at some point there is a very good chance that something is going to be trained to get there which doesn’t make mistakes and doesn’t have off days and doesn’t take leave and doesn’t get sick. So for you to justify all of those things that make us human, you’ve got to bring a lot of judgement to the table. You’ve got to operate in areas that are grey, because that’s very hard to train machines to do.

That’s my theory at least. I’m keen to hear yours and your thoughts on the broader impact, just on employment, I guess, and how jobs will change.

Shane Cooper: Ghost, spot on! And if you add a little bit more nuance to what you’ve said around the reference to a job being black or white, if you were to think about it in sort of AI terms, around deterministic and probabilistic, if your job is deterministic, where you have a very particular outcome, for sure your role is at risk. And that applies, for example, to environments that are highly regulated. If you think about a lawyer, although we have heard some very interesting cases around AI being used in the legal space to the rather embarrassing outcome where the judge finds out that case reference that you’ve used is completely fake and made up. That’s happened quite a few times.

The Finance Ghost: Yeah, no hallucinating in court! Don’t be doing that with your AI.

Shane Cooper: These conversations are happening today and I do think that the world is going through a crisis, if not necessarily evident to many people, and I do think we should precipitate crisis conversations around AI. If you listen to someone like Dario Amadei who is the CEO of Anthropic, remember he was part of the team that was initially at OpenAI and he broke away. His view is that more than 50% of jobs are going to be affected by AI. In fact, he said more recently that he expects that the bottom rung of employment is going to fall away within the next five years. Now, essentially what he’s saying is that entry-level jobs, whether they are deterministic or not, could be taken up by Artificial Intelligence.

The Finance Ghost: Because, sorry Shane, the problem there is the entry-level job doesn’t have enough probabilistic elements, right? Because people don’t have the experience to operate in the grey. The judgment calls they need to make are not that advanced, right? That’s the issue. That’s what scares me.

Shane Cooper: And that’s why I say it’s more nuanced. If you consider that an LLM today is probabilistic, you can ask it the same question five times and more than likely you’re going to get five different answers. They may be all correct, it’s just the way that the answer is given.

That could be a junior role in an organisation where you’ve joined a legal firm, you’ve joined a consulting practice, you’ve joined a journalism job, and your first year or two or three of activities are relatively, let’s say, simple in the context of your longer journey. Those simpler tasks can be undertaken by a probabilistic LLM. Because the view today is that an LLM is essentially a student, potentially with an honours degree coming out of university.

If one was to talk about the age of LLM, it’s not quite the doctorate student yet, the view is, but it’s a collection of very intelligent honours students.

The Finance Ghost: Hallucinations included, sometimes depending on how their year went at varsity.

Shane Cooper: Absolutely, hallucinations included. And I think that’s what we should remember is that hallucinations are simply errors made by the LLM. And part of the reason for this, is that LLM’s are trained to give a confident outcome. They’re not trained to determine whether something is right or wrong. So they get rewarded if the answer is confident during the training process. Mostly it’s around accuracy, but it’s also about confidence.

The Finance Ghost: That explains a lot about some of the bad answers I’ve been given. The AI never says, well, I’m not sure, but maybe, which is actually what a grad would say.

Shane Cooper: Yeah.

The Finance Ghost: The AI says, oh, it’s definitely this. Absolutely.

Shane Cooper: For sure! And that’s why when people talk about the risk of hallucination in the corporate world, my answer to that is, well, you would review a junior’s work, wouldn’t you?

Assume that for now you have the power of as many juniors as you would like. Just make sure you review the work before you submit it. Over time, as these LLMs learn, just as people do, the margin of error will decrease.

Coming back to the question of the risk on jobs, I do think that we’re moving into a world where there should be huge concern around the impact on jobs. If you listen to, and as a journalist, you’d probably have noticed that the tone of the last year, and potentially more the last six months from CEOs around the world, is that they’re far more comfortable in talking about the fact that jobs are going to be lost to AI than they were before. There was always a reluctance to talk about the risk of jobs being lost. But we are now for sure moving into a space where decisions around hiring for someone who has resigned, there’s a pause that says, right, let’s just see first whether the AI can fulfil the job. So I do think that over the next three or four years we are going to have a fundamental shift in how the job market looks.

Now coming to the point around the benefits of AI and what that can bring to bear in society, is that if we do see a significant uplift in global GDP and the benefits that AI has promised come to fruition, what does broader society and policy makers, what decisions do they make around the concentration of revenues into these larger corporates, these owners of the LLMs, in terms of the broader dissemination of wealth? Because you now have people who are able to do a week’s worth of work in three days, do we go down to a three-day work week? Do we have universal basic income? Because think about it, the AI tools that we deploy are not free. You have to pay someone, largely in the US given that there’s a concentration of the frontier models in the US.

Then there’s the broader debate which we probably have a separate discussion on, which is the slow death of the internet economy. I’m convinced that the internet economy is probably going to die in the next two years. The whole ad generation and sharing of revenue that Google led the path on many, many, many, many years ago – AI is changing that. I think there’s a huge question to be asked around policy making and how people today, who are in the workspace, take a very, very clear view on how their role can be impacted by AI and what they can do to ensure that they don’t become a victim.

The Finance Ghost: Yeah, Shane, there’s so many fantastic points in there. I mean, we could probably do another entire podcast or two on just all of the debates around this stuff. So universal basic income – the problem there, right, and you said it is so much of the value from AI is going to flow into the US because that’s where these companies sit, but the people who are impacted by it are sitting everywhere else in the world. And you can be very sure that the likes of Microsoft, etc. are definitely not going to pay a universal basic income back to the people who lost their jobs because of a model built in the US.

There’s a lot of stuff coming down the road which is not pretty. That’s my view, which is not good. I’m equally bearish on Google’s ad revenue business. I must be honest. I wonder how we then end up in a world where there’s actually motivation for people to have niche websites to create this kind of content that were previously supported by Google Ads. Because the AI still needs to go and read something, otherwise how will we ever know what is real or not? I mean, it’s proper sci-fi stuff at the end of the day.

Shane Cooper: I was just going to say: scarcity creates the value exchange. So if there is a point at which content creators no longer feel that they’re getting what they should be receiving, they’re going to stop producing content, they’re going to go somewhere else. That’ll create a vacuum and eventually an economy is going to be created elsewhere. But for sure the disruption that Google is doing to themselves as well as, in particular ChatGPT is doing to the internet economy, is real. I think we’re going to see some real pain for people who generate their revenues from ads on the internet over the next year or so.

The Finance Ghost: If we are lucky and maybe just talking selfishly now because my model has never been clickbait and a lot of other platforms out there are very clickbaity because they just get rewarded for the number of people who arrive and so they write accordingly, etc. If we get away from an internet adverts world, then we also get away from clickbait. We get back to maybe having fewer voices who are credible and then who attract brand partnerships that support that voice and platform and want to get to that audience as per this precise partnership with Forvis Mazars.

So it’s interesting, that’s how this stuff plays out. The other thing I just wanted to touch on before we bring this home, was you mentioned “journalist” earlier and that’s a big part of what I do. But it’s just there’s an important lesson in there actually for people thinking about, well from a career perspective, how are they going to navigate this stuff. I’m also a CA by background. I did my years in banking and now I get to do this for a living, which is fantastic. But I’ve been able to do it because I’ve had to go and learn how to at times look like a journalist to the outside world, managing the website, doing all of that myself, etc. You have to learn how to sell. You have to learn so many different skills and I think that’s going to be the trick. If you’re just sitting there in university right now and you’re saying, well, I’m going to be the greatest XYZ type of lawyer or doctor or accountant or whatever in the world and I’m going to just hyper-specialise – you are looking for serious trouble, I think. Personally, I mean none of us know, but I think you have to get very good at something and then go broad after that. You definitely can’t just go extremely good at something and then sit and say, well that’s it, now I’m safe. That’s just not the world we’re in anymore. You’re going to have to pick a skill to get really strong at and then you’re going to have to learn a whole bunch of other things just to make sure that you are not replaceable by one model or one prompt or one system.

It’s hard. It’s not easy. I’m very glad my kids are too small right now to be impacted by this. By the time they get to university, we’ll know how this panned out. I’m grateful to not have kids going out of high school right now looking for advice because I wouldn’t know what to tell them. I’ll be honest, I really wouldn’t know what to tell them.

Shane Cooper: Yeah, it’s a very difficult situation to be in. If we were to look at the future by seeing what happened historically, the fact is the internet world that gave birth in the latter part of the 80s, early 90s, did give rise to a number of new kinds of jobs and work. So I would imagine the maturity of AI will also give life to a whole host of jobs we would never have guessed today.

I don’t think it’s all doom and gloom. But I do think that given the philosophical prescience of sci-fi authors, we are beginning to reach a point where we wonder about the future of humanity. I mean, isn’t it insane that we can sit here today and talk about how some of the original founders of these large language models inside of Google are talking about the risk that this technology has for humanity? Now, if you just use that as a proxy for the power of the technology and the impact that it could have on our lives, even if we were to assume humanity continues to exist, we do have real existential questions to ask ourselves about what is the meaning of this all and how do we continue to eke out an existence into the future?

So I think if you are a parent with teenagers, the anxiety is real, for sure. For me, ultimately the hope is that people still like to engage with people. In the business of dealing with buying and selling, people still buy from people. My view is that provided you’re a well-rounded human being and you can engage well and you can distill and synthesise information and articulate well with people, you should be okay. But that doesn’t mean that you shouldn’t be asking yourself some critical questions about how this technology can impact you in your life.

The Finance Ghost: Yeah, absolutely. Just a really good example of a career or line of income that’s come through recently and where I think the model could go, is something like YouTube. Being a YouTuber is an actual thing. People use this to disintermediate traditional media to get a message out there. And people love YouTube! I’m consuming more and more YouTube all the time about niche stuff, like hobby stuff and whatever, because you can just find anything you need, anything you’re interested in is on there. And then you find creators who are thought leaders in a specific niche and they get rewarded accordingly, which I think is fantastic, it’s a good example of where the internet economy still works. I think YouTube is by a country mile Google’s most impressive long-term, wide-moat kind of business. We’ll see how that plays out.

I guess to start to bring this home, Shane, and maybe just bringing it back to business stuff and I’m glad that we could talk so much about the human side of AI because too often that just gets brushed under the carpet actually as people talk about the business application. And the reality is that as a business owner you also can’t sit back and say, well you know, I hate all this and I’m not interested and I’m not getting involved because you’re kind of going to create your own downfall. I think that’s the unfortunate reality. Can’t ignore this stuff, just can’t.

So if you wanted to leave listeners, I guess, with one essential message about AI and business strategy, something that might get them to open a conversation with you to try and understand more about what the team in Forvis Mazars could assist them with, but even just outside of that, what is that essential message around this stuff?

Shane Cooper: Maybe just before I give the formal close on this, is that you’re absolutely right – organisations cannot ignore this. I would prefer that organisations, if they decide not to embrace AI, at least make an informed decision about it. Understand how it’s going to impact your life. If your decision is not to embrace it, at least make it an informed decision. I think that when organisations are considering their future use of AI, they mustn’t think of it as a technology project. It has to be for me, a board-driven transformation aspect of the business. It has to include how it impacts the organisation in its entirety. And please do not let this be the preserve of your technology department or your CFO department. This has to be something that’s embraced by the organisation in full. And I often refer to the principle of vertical and horizontal deployments of AI. Just make sure that when you are embarking on an AI project, pick the value pools that are meaningful. Where are the biggest levers in your business that you wish to either improve or protect? And work hard to understand how AI can impact that environment because then at the very least, you’re clear that you’re building a shield against that competitor who arrives out of your blind spot that you didn’t expect, who’s embraced this technology and looking to make a difference in the industry.

The Finance Ghost: Shane, I love that and I think there’s been a ton of really great insights here, so thank you. It’s been fascinating to speak to you. I’ve really genuinely enjoyed it.

And to listeners, if they’d like to connect with you, I imagine LinkedIn will work. Should they just ask their LLM how to get hold of you and the LLM will hallucinate some kind of incredible answer about your phone number or what is the correct way to find you?

Shane Cooper: They can find me on LinkedIn, it’s Shane Cooper. Or they can find me on the Forvis Mazars website. Name.surname@forvismazars.com if you want to drop me a line.

The Finance Ghost: Cool. Shane, thank you so much. Really interesting times for all of us. And to the listeners, I’d welcome your views here. What do you think about this whole AI world? What do you think is going to happen in the future and how are you using it in business? That’s also fascinating feedback.

So, Shane, thank you so much. All the best with this stuff. You are right at the cutting edge of what’s going on out there right now. Must be very fun, slightly scary. Good luck to you with that.

Shane Cooper: Thanks very much, Ghost. It’s been a pleasure and I look forward to chatting again.