“If Your Job’s on a Computer, It Will Change.” Here’s What to Do Next | Mark Rotheram | Ep 1

There will be disruption. The jobs that people do today, if they are doing them on a computer, are likely to be the first ones that get disrupted. That I think is a given. Welcome to the AI for Business Podcast with BCN.

I'm your host, Sinead Hammond, and today I'm joined by CTO of BCM, Mark Rotherham. How are you doing, Mark? Amazing. Amazing.

I'm glad. That's great. Just before we start, I wanted to give listeners some updates for the podcast for this year, as we put AI and automation at the centre of every episode. We wanted to create a regular place for business leaders to digest the business news and the biggest stories in development in AI as they hit the headlines.

And when it feels like there's a new story and technology pretty much daily, we're cutting through the nominees and we're giving you a rundown of what's important to businesses, why it matters, and what you can do to keep up with the pace. So we're a few weeks into the new year, and I feel like AI is really centering in every conversation around the future and strategy. 2026 feels like the sort of big inflection year for AI. And I was wondering what has changed, Mark, in the last six to 12 months that's making conversations about AI so central.

Yeah, so the last 12 months have seen AI go from almost a kind of novelty thing that's on the periphery of kind of what we're doing to a mainstream tool that is effectively ready for the prime time. So we we've effectively evolved from AI being fun, sometimes getting things right and sometimes getting them wrong, to it in maturing to a point where we're seeing it become a huge differentiator and asset and also risk um very quickly. You know, if we look at the capability that's changed in the last 12 months, it's incredible.

We we've gone from AI being okay at a lot of things and being generally pretty rubbish at a lot of things as well, if we talk about generative AI, to it becoming on par, if not better, than humans for a lot of the roles that we do today. So I I think the thing that's changed the most is that evolution from it being kind of a novelty, um potentially disruptive, to it's here now, it's disrupting, it's becoming the default, it's become cheaper, it's become something that you can actually use to do great things.

Um and uh we we're still really early in the cycle of of where it's going as well. You know, we're seeing interest peaking month on month. So it's growing interest, it's growing into every boardroom conversation and everybody in their personal life conversation as well. Absolutely.

And I think when it first started, it was very much a technical IT conversation or it was kind of reserved for that, those technical minds. But I think it's, as you say, becoming a conversation for every day. If you had sort of 60 seconds with a CEO who still thinks AI is just an IT project, what would you say? Yeah, so I mean the misconception I think is that it's just tech, it's not an IT project.

It's a business operating model upgrade. It's a way that we can enable a business to change their margin, their speed, their quality, the customer experience. The IT team, yes, will use a bit of AI, but the biggest impacts are gonna be outside of IT where it uh starts to impact the rest of the business. So the 60-second summary would be it will transform the business operating model, or at least it's got the ability to transform the business operating model, and it cannot be restricted to being led by IT.

It's got to be part of an IT strategy, absolutely, but it is a business operating model strategy, so it's a business-wide impact. That makes sense. I guess going back to that being a business strategy, obviously, in a lot of cases, things start kind of from strategy. Do you think that's the same with AI?

Would you advise if you were going to start embracing AI? Obviously, 2026, everyone's talking about it, it's the big year, more so than it has been. What would you say to kind of a business who was thinking, okay, I know that we need to embrace this. Where do you kind of go in with that first?

Where would you start on that business journey? Would you suggest? So if we're looking at having a business kind of level conversation, the starting point is is all around the people. We've got to have an understanding of what AI can do.

We've got to have an adoption strategy, we've got to understand what it's going to do to us as a business or what it's capable of doing as a business. And really looking at it from a kind of transformational perspective, what we're seeing is the broad adoption of AI is extremely spotty across many businesses. And what that means is you'll you'll have a couple of people that may be champion it and have a go, and the rest of the business just carrying on business as usual, not really knowing or wanting to know necessarily about AI.

The human-centred view of the impact of this is absolutely important. So it's it's kind of starting the adoption journey and understanding the potential impact and figuring out a strategy that allows AI to become a thing and not to be stifled and shut down or uh fail. Uh, we do see quite a lot of pilot kind of engagements where we look at a a slice of a business, and that's absolutely fine as well. You know, we see a a lot of tactical where we want to make an impact here, and AI has got the ability, so let's focus on that bit of the business, which is great to kind of prove the capability and to get some value out of it.

But taking that step back, every business needs to have a strategy of how are we going to adopt this? What's it look like? What do we look like in six months' time, in 12 months' time, what jobs are going to shift and why, and what happens if we don't do it, you know, the the biggest potential threat to a business is not doing anything with AI. We all know about first mover advantage, but there's also a no-mover disadvantage, you know.

Not doing anything just means that over time you will be limiting yourselves to, well, a non-automated, non-efficient, non-AI augmented way of working, which means you'll you will be slower, you'll cost more, uh, your quality won't be as high as it could be, and your competitors uh could potentially be doing the exact opposite and getting all those kind of rewards that AI can bring. Okay, so if we sort of accept this as sort of a big moment and businesses are really going to embrace AI this year, there's so much talk.

I mean, I hear it all the time, different terminology. Last year it was agency, agents. This year I'm hearing terms like AGI is everywhere. Firstly, what is AGI?

Because that's something that's hitting the news loads at the moment, and it's something that I'm hearing good, bad things about. And then how does it relate to business? And is there a better, more practical question businesses should be asking or thinking about in terms of these big kind of new terms that are coming out in 2026? Yeah, so I mean, there is a lot of debate around general intelligence, artificial general intelligence, and artificial super general intelligence.

So we have these phrases banded around quite a lot. And being human, our first kind of thing to try and do is put it in a box and say this is what a benchmark is that we can measure AGI against or intelligence against. And that's really hard because most people disagree on what general intelligence is or what superintelligence is. The way that I like to look at it is what can it do?

What benchmarks can we demonstrate that the AI today can achieve and do? So the focus on the capability and proven capability for me is really important. There are lots and lots and lots of different tests and benchmarks out there that every one of the new models goes and has a go at. But for me, it's really making it real.

What is the current capability of AI and how can that be used in the business context in a way that's going to make a difference? We are seeing we have entered, you know, what people call the kind of singularity, which is this exponential curve of AI getting better. You know, if we if we think about the really simple concept of a year ago we had people using chips from Nvidia and they were training AI. Yeah.

Now over the last 12 months, those chips have got bigger, faster, stronger, much better. What would a chip be? Sorry, just for info. You know, just general processors that you have in your computer, graphics, cards, those are getting better.

You know, we've got this thing called Moore's Law where basically things get faster and quicker. We're applying that to AI. In the past, we've had humans applying their intelligence to make the chips faster. What we're entering now is the AI actually looking at itself and making itself better and faster.

And what that means is the models that we use, you know, Chat GPT or Claude or Groc or whatever it might be, Google, um, we're moving into an era where they are going to compound and get better faster than we've seen before. So the capability that we've seen arrive in the last six months will very quickly be overtaken by the capability in the next six months and so on. To kind of answer where we should be looking is what can it do today and can it impact my business model today in a constructive, useful way that gives us a return that's actually going to be quantifiable and useful.

Yeah, makes sense. That's quite a it's a big step forward. I think I heard DeepMind CEO suggests that we're on a sort of five to ten year timeline at the moment. I think even quicker in some cases that we're getting, you know, this brand new way of working completely, this brand new way of existing, I guess.

So if you if you think about the way it's integrating all over the way we live and the way we work. I think what can be, I guess, challenging is just understanding like what these you mentioned benchmarks, what do they look like in a business context, not like a technical one? Like how do you start with this adoption of these technologies? Yeah, I I think by making it simple.

I I like to keep things as simple as possible. So if if we look at a role that happens today, so let's use the a role of someone that reviews a document, yeah. Their goal is to open a document and review it and check if they're happy with it. It could be anything, yeah.

Now, they will have learnt what good and bad looks like. They'll have kind of cottoned on to things that are good, bad, indifferent, and they will do that job absolutely fine. Now, if they leave and you bring someone else in, they've got to learn how to do that job. Yeah.

Now, what we're not very good at is writing down what the benchmark is there. You know, we're quite good at writing a process, review the document and say whether it's good or bad. But benchmarking is the rules in this case of what makes good and what makes bad bad, and what do you not need to worry about? It's almost having measures against the work.

And once we do that, we can then apply those uh measures and those benchmarks to AI. And we can say, right, AI, can you go and do this against this document? And then what we're effectively doing is changing the role. Yeah, we're changing the role of the person that was doing it from doing the work to defining the benchmark and then to validate the benchmark.

The compound impact of that is you probably don't need as many people reading and reviewing those documents. You know, if you had 20 people doing it, you could probably get away with one person or two people defining and checking. So we're seeing that kind of really simple. What is the benchmark for this piece of work, how we're going to measure its success, how we're going to define it, and then align that with the capabilities of our AI that we can tap into.

That's kind of how I try and make it really simple. And then we can just extrapolate from that further and further into more and more complex things. But for AI to be really good, we need to have these kind of benchmarks or these measures that we can use to actually say, look, this is your operating guidance, this is what we need you to be able to do and demonstrate, is not something that many businesses have documented or written down in a way that's very easy to digest and maintain.

Some in regulated areas are, but in a lot of other areas, we we kind of rely on people and training those people to do that kind of job. Yeah. Kind of brings on that question that everyone's talking about or everyone's worried about at the moment, is around kind of what AI is going to do around jobs. I mean, you said there that you've gone from 20 people doing document analysis and research and summarization jobs to one person managing, I suppose, an AI agent to do those things for them.

Do you think there's more a change of task as opposed to change of is it a reduction of jobs or is it kind of more a change of what specific roles or specific tasks, should I say, people are doing in their day-to-day? Like what do you think about that? Yeah, that there are a lot of different repercussions for what's going on now. A lot of them.

I I think the number one given is it there will be disruption. The jobs that people do today, if they are doing them on a computer, are likely to be the first ones that get disrupted. That I think is a given. As we kind of go into the impacts of that, people in that world that I just described that will be the most successful doing the kind of role that they were doing before will be the ones that embrace and use AI because that will make them the most successful at carrying out the role that needs to be performed.

If we use an example with, you know, writing code as it's a really good example, that's going through a shift. You know, if we go back a year or two years, every line of code would typically be written one line at a time by someone. And that's where companies were investing. They were investing on more and more software engineers, people writing code.

I remember a couple of years ago thinking, well, that's where the money is, you know, that's where the job market is, because they're going to need tons, we're going to need tons of software engineers to write all of the code constantly. Yeah, and and what we're seeing now is that is reducing. It's not fully gone, but the need for somebody to write every line of code is reducing. You know, with with the the changes we've seen in the last 12 months, AI has become much more reliable and competent in doing that.

Now, the there's two sides to this coin now. That means that the cost of providing solutions in that world is dropping because we can use AI to reduce the cost of doing it, which means we can do things quicker and faster. And so that increases the potential market for that capability. So what we're seeing is again, if we if we keep on this path, things that used to cost too much for a lot of people previously now become not costing too much.

So we have growth. But the role, the tasks that the people are carrying out is changing. So there may be more work, but the work will be different. And that's the disruption that we've got to navigate around for every business.

I mean, if we take it into the personal world for a moment, two or three years ago, none of us, unless you were super rich, had a personal assistant. And now everybody has got access to something which isn't akin to a personal assistant because the cost of that has come right down. You can go to Google Gemini or Perplexity or ChatGBT, and you can have a personal AI assistant that will collaborate with you, they'll help you, they'll talk to you, they'll do all these things, and it's costing you very little to access that.

Everyone's got a new capability that they may be subscribing to that wasn't available a few years ago. But have we disrupted the personal assistant ecosystem dramatically? Maybe a little bit, but we've seen massive growth. So you've got lots of different ways to see the impact of AI.

I'm not going to say it's all going to be shiny and nice for everyone because people like doing the work they're doing today. And, you know, when we look at the human-centered approach to AI, a lot of people will start to use AI in their work construct, realize that it's potentially going to impact their livelihood and their jobs, and they're going to have to change, and quickly bury it and stop using it. And that's where we see a lot of the adoption stifle. So it's all around how we we manage that and we set expectations that you know this is disruptive, things are going to change, but there's a plan.

Yeah, there's there's got to be a a plan at each business of to how to approach that and what to do with it. Well, can we dig into that a tiny bit more then? So obviously, you've talked about people that work on computers having being the most disrupted. I know finance, marketing, operations, sales, all of those roles that spend most of their time doing that.

I feel like, you know, if they're going to be disrupted, there has to be a little bit of a path. Like, what would you prioritize if it was you to start doing a bit differently? Even like this month, this week, in the next quarter, like what would you be doing to kind of make sure that you're keeping up with this rate of change and making sure that if the tasks do change, that you're still able to give back value, I suppose? It's it's a mindset thing for me.

Humans are really good at adapting to change. We're not very good at adapting to change very, very quickly. So to prepare for this, we've got to start thinking about what it's gonna look like. For me, it's it's all around being curious and starting to use the tools around you and learning what they can and can't do and keeping up to date with that.

That this disruption is is coming. It's already here in a lot of tangible ways that we're seeing. So it's having that curiosity, it's not being scared. That the people that embrace and adapt to AI will be much more valuable, they'll have much higher output, higher quality, and they'll be able to drive in a different direction from those that don't.

So I think for me it's looking at it from that perspective, you know, that AI is not going to get put back in the box. It's only gonna get better at what it does. So if you are working on a screen or computer and you know, you're not considering the potential AI impact. Or the value of it, then you should probably reassess your potential role of the future's going to look like.

Yeah, yeah. I think it's it happens over history. If you look back, there are always big changes and there's always resistance and there's always, you know, embracers, early adopters. And I think it's just about taking that mindset.

Like you said before, like we can change, we can adapt, we don't always like to. So I think it's a case of like choosing that mindset. I love what you said there about mindset and about making that your choice to try and find out what more AI can do with your job role. I mean, I personally love it.

I use it constantly for my job role. I think being in marketing in particular, it is one of those sectors that I think is potentially AI can start to do a lot of the roles. So I think it's looking at how we're to try and use it to augment what we're already doing and scale and make it bigger. And I think that's really exciting more than being scary, which is great.

What are the risks that start to show up? I think this has been something that's been talked about absolutely loads. I know that even Bill Gates came out a couple of, I think it was this week, actually, saying that we need to really think about how this technology is developed, how we govern it, how we deploy it, how we look after it. What kind of things would you say are like the bigger risks?

And what would you as a business say to make sure that you're putting those guardrails and governance in early on? Yeah, so as I mentioned right at the beginning, it it it brings risk. And the risk is in lots of different places. We have got the potential of Shadow AI, you know, where people just go and use public domain models, basically models that haven't got the guardrails in place that stop data leakage.

You know, one of the easiest things to point at is the potential of data that you don't want to leave your business leaking out. And that's an easy one to kind of point at, but I think there are lots of other risks. One significant risk is the over reliance on AI and not having checks and balances. It's extremely easy to give AI a question and just accept whatever it comes back with.

Whether you're writing a document, summarising something, creating something, it's extremely easy. And let's face it, a lot of people think that they can just offload all of the checks and balances into the AI and it's going to give them exactly what they want. So that's where we've got to make sure that there's an understanding of hallucination or getting things wrong. It's having the right human-in-the-loop processes to to kind of counter that.

It's that kind of governance layer of making sure that people know what they should do and what they shouldn't do with AI. We've got risks of people trying to infiltrate AI with bad data to influence the way that something happens. There are an awful lot of potential risks. It's making sure that there is a policy that says this is what is approved and why.

Then having an acknowledgement of if we are going to build something out, we understand the guardrails and the risk mitigation that we're going to put in place. I I see this being really immature. If we if we look at what global leaders, and um, I'm talking tech leaders, wanted to do a year, maybe two years ago, three years ago. They wanted us to slow down the AI race because the perception was we're not ready for this technology.

And we are not ready for this technology. Our governments aren't necessarily ready, our businesses aren't ready. We are seeing companies try and catch up with regulation and guardrails to kind of put those things in place, but they're still coming, you know, and we need to spend that time on governance and controls to make sure that we don't end up with serious issues that we didn't know or didn't see were coming. Is there anything that you would insist on human sign-off every time?

No exceptions. I think it's a maturity question. So let's use a really good real world example. If you say I want to go from A to B in your sat nav, you're likely going to trust it, you're gonna follow it, and more often than not, you you will just follow it.

You'll drive, but you'll drive safely. But yeah, hopefully not into the room. And that's the point. Yeah, exactly.

You get this point, but then let's flip it slightly differently. I i if you send an email, you expect the email to be sent, and you expect it to be 100% accurate each time. If you use technology to do a thing, we become comfortable, once we are comfortable, that it's doing exactly what we want it to do. That's kind of a maturity thing, and and I'm gonna say it depends, because the application of AI against a task will have a different level of maturity depending on the AI we use and the tasking in mind.

For me, there's always got to be checks and balances to a point, but at a point, you will likely get to the the human being removed from the loop because you've got the trust there. You m you might have different checks and balances that are automatic, but you'll get to that point. And again, that the the speed at which we get to that point, I think will increase as the AI gets better and better and better throughout the next few months. Even now it's extremely good and probably better than humans at a whole raft of tasks that adding a human into loop would probably not help, it would probably cause issues.

So uh yeah, it's definitely a consideration in everything we do, but um it's also a thing that that you've got to mature into. I I'd say it depends. It depends. That's great.

I love that open-ended. Um I guess just one final question to just finish off this episode, then. If a listener takes nothing else away today, what are the first three actions you take in the next 90 days to get ahead of this and of AI and set yourself up for 2026? Um, be curious on what AI is potentially going to do to the work that you do.

Be curious and have a go. That there's so many things out there in the personal world which are fine to have a try at. I'd encourage you to get access to a safe AI in your workplace. Co-pilot chat from Microsoft is free to everyone that's got a Microsoft license, effectively, and it's a safe place to go and try out an AI and become familiar.

So I think being curious, trying it out, and I guess being prepared for some change. It's getting more familiar with what's going on around us and starting to figure out what you're going to do with that information. Something I've done. This is my role, this is what I do.

How's AI going to impact me over the next 12 months? See what it says. It's a good thought experiment just to see, you know, what's happening and then ask, right, and what do I need to counter that? How do I use AI to kind of power uh what I do and become more efficient at it?

Love it. Definitely think the biggest takeaway from this episode is that 2026 is the year to move. This is the year for really embracing AI. The gap between AI capability and adoption is widening, and businesses that will really win this year will be the ones that start small, learn quickly, keep going, keep growing, which I think is amazing.

Thank you very much, Mark. Yep, it's been fun. So if you enjoyed today's conversation, please follow the show. We've got practical real-world examples coming that will help you make AI actually work in your business.

We want this to be a place where business leaders can come, listen to all of the different headlines that are going on and how that applies to your business and what you can do to really make that work. You can listen to these episodes at bcn.co.uk.

You can also get in touch with us, and you can listen to the podcast where you normally listen to all of your podcasts. So thanks very much, Mark, and thanks for listening.