Biggest AI Risks Revealed on Live TV!

The landscape of artificial intelligence is evolving at breakneck speed, and understanding its trajectory is crucial for navigating our future. Three-time author and AI award winner Kieran Gilmurray joins us to deliver a balanced and thought-provoking examination of where AI is today and where it's heading with implications that touch every aspect of our lives.

TLDR: 

• AI adoption is inevitable with massive investment driving innovation, but we must approach it cautiously
• No job is truly safe from AI influence even roles previously considered secure like caregivers and mental health workers
• AI has reached public consciousness faster than any previous technology since ChatGPTs 3.5 launched in 2022
• Global regulation is fragmented with nearly 200 different legislative approaches worldwide
• The potential development of artificial general intelligence (AGI) raises profound questions about humanity's purpose

Gilmurray cuts through the typical AI hype cycle with refreshing candour, acknowledging both the tremendous opportunities and sobering challenges this technology presents. "We're racing ahead with this technology and maybe not putting the brakes on where they need to be put," he warns, setting the tone for a conversation that avoids both techno-utopian fantasies and doomer narratives.

The discussion tackles the uncomfortable reality of AI's impact on employment with surprising directness. 

Contrary to reassuring narratives about certain jobs being "safe," Gilmurray challenges us to recognize that virtually every profession is already being transformed from knowledge workers to caregivers, construction to mental health services. Rather than fearing this change, he advocates for embracing AI as an augmentation tool that can make us "more innovative, more creative, more productive" if we approach it thoughtfully.

Security concerns dominate the middle portion of our conversation, with Gilmurray revealing alarming real-world examples of AI-enabled fraud, including deepfake board meetings that have authorized multimillion-dollar fraudulent transfers. The technology's ability to "hallucinate" convincing but fabricated information raises additional stakes, especially when professionals rely on it without verification. These challenges are compounded by a fragmented global regulatory landscape that struggles to keep pace with innovation.

As we ponder the future development of artificial general intelligence (AGI), Gilmurray presents both the promise of technology that might solve humanity's greatest challenges and the profound philosophical questions it raises about human purpose and agency. His insights reveal that our relationship with AI isn't just about technological adaptation it's about deliberately shaping our collective future.

Whether you're a business leader, policy maker, or simply curious about how AI will transform your life and work, this conversation provides essential perspectives on navigating the AI revolution already underway. 

Chapters

• 0:00 AI: Embrace or Exercise Caution?
• 1:20 Will AI Take Our Jobs?
• 3:01 Jobs Safe from AI Disruption?
• 4:58 Unprecedented Growth of AI Technologies
• 6:57 AI in Wrong Hands: Fraud and Risks
• 9:32 Regulation Challenges Across Borders
• 11:38 Deepfakes and Fighting Misinformation
• 15:43 AGI: The Future of Human-Level Intelligence

Transcript

Miahael: 0:00

Kieran Gilmourie, three-time author, multiple AI award winner and AI consultant. He'll help us cut through the hype, explore AI's risks. Is AI our biggest opportunity or our greatest challenge? Kieran, thanks so much for your time. I'd like to start out with. Ai is transforming everything. Should we embrace it or should we be cautious?

Kieran: 0:23

I think it's a mixture of both. To be absolutely honest, there's no running away from this, because the amount of money that's going into it for work, for leisure, for military purposes and everything else there's too much. It's too compelling, and the benefits that you get are extraordinary, and lots of companies are getting those already, as are lots of us, you know, using it for fun or for home productivity, as it were. But we should be worried as well, because we're racing ahead with this technology and, dare I say, we're maybe not putting the brakes on where they need to be put, and if we're not careful, we're going to end up in a place where we really don't want to be. So how do we run fast? How do we innovate, but do it without jumping off a cliff? That's the real questions for us, for businesses and for governments today.

Miahael: 1:09

And is AI? Is it going to take your job? Is it going to take my job? How sure we can be about the future of job security for the entirety of humanity.

Kieran: 1:20

Yeah, I sincerely hope it will take my job and I say that with the best intention in the world. If I'm still doing what I'm doing, in five years time, there's something wrong. Now this is the bit. Lots of people shy away from this. I don't. Ai will replace jobs, you know, and it doesn't matter the type of job that you're actually in. I'm finding it very difficult today to find jobs that it won't impact and therefore we have to embrace it. But look, this is a history of jobs in general. If you think about it, there's a lot less software engineers or Sadler, sorry than software engineers, and in about five years time, there'll be a lot less software engineers than there are today.

Kieran: 1:56

Jobs have always been impacted. I think. When I look at it, we fear what we don't know. Ai is one of those big unknowns, but it doesn't matter where it is, what we're doing, any particular job, and this is the worrying bit at the moment for most the so-called you know, knowledge worker jobs the bank workers, the estate agents, the educators, the website builders, the translators, the copywriters or the consultants all of those jobs it's not a case of they will be under threat. They are already under threat. Ai is embedding itself into all those roles at this moment in time. So it can, it will, and it's already impacting roles today, what we got to do is, like every other technology that came before us, is not fear it, embrace it, but understand the risks, make it part of what we do to allow us to be, you know, dare I say, more augmented, more innovative, more creative, more productive, to allow ourselves to be what I would describe as employable, not just in the short term, but in the long term as well.

Miahael: 2:54

And you just stated all of the jobs that potentially aren't safe. What jobs are safe?

Kieran: 3:01

Well, back in the day they used to say and I say back in the day because when you and I used to talk about this, it was five or 10 years ago. Now we mean about two years time the classic is the caregiver role and I keep hearing this from people, but there's been robot helpers or robot caregivers in Japan for years. We're currently working with another research group out of Singapore at this moment in time to put you know hotel reception, hotel cleaning. If you turn up at Heathrow Airport, you're not noticing. You know cleaners, who are humans they are normally, by the way, but it's robot cleaners. If you turn up in Belfast City Airport, then your burger is going to get delivered by a robot.

Kieran: 3:36

So I would have said numbers. A year ago. You know there were certain roles that were not going to be impacted by AI. You know building a house, but again, automation, ai, robotics are all going into place to allow these things to happen. You know caregiver roles. I would have thought years ago, no, that's impossible. But now you know there's robotic caregivers and the dexterity of robots improving.

Kieran: 3:59

I would have thought mental health workers would be uninterrupted. But last year, when you looked at the greatest use and there's a hbr article and they're saying what were the top 100 uses for chat, gbt or similar tools? And it was work-related tasks. The top use this year, in 2025, is mental health and relationship advice. So those are not, by the way, tools to be used for mental health advice. They're certainly not certified. There are tools out there that allow you to do it. So today, michal, I'm really struggling to point to jobs that won't be impacted in some shape, form or other, which is either very troubling, if you're going to be impacted by this and you're not ready, or very exciting, if you embrace this technology to allow you to create more, do more, achieve more and free yourself up to do the fun things in life.

Miahael: 4:46

And ChatGPT only started in 2022. It's hard to imagine now a world without it.

Kieran: 4:58

Are you surprised at the rapid growth of this technology? I am extraordinarily surprised Now. I've been in business tech for about 30 years and I've never seen anything move as quickly the adoption of this by everyone. We're not just talking about government, we're not just talking about educators, we're not talking about workplaces, but it's everyone and anyone. I'm sitting delivering a training course in London to a group of educators yesterday and I said how many of you are using this and they're still learning? Which was great.

Kieran: 5:21

But someone said my brother's doing it and I your brother doing. And they said well, he's actually a landscape gardener, so he's using chat GBT to demonstrate the before and after effects of what he's doing in a garden. You know? Someone else said well, I'm going to Japan in the next couple of weeks and I've just used deep research to find my hotels, to book my trips, to tell me the cost of everything, to teach me a little bit of Japanese, you know.

Kieran: 5:43

So this has made it into public consciousness before any other tool, or more rapidly than anything I've seen, and the exciting and the scary bit is that it's only accelerating. Now there's a flywheel working. It feels like a snowball running down a hill that we're all chasing after, but we're only on and when you think about it, chat gpt was november 2022. We're in version coming up to version 5. That's only in a very short period of time, and version 10 isn't going to be in 10 years time. It's going to be far less, far quicker, far better than anything we've ever seen. So the horse has bolted. Now the choice is jump on board or or turn away. I wouldn't recommend the second I. I recommend learning and learning quickly to embrace this tech, because it is everywhere, not just the fact it is going to be everywhere.

Miahael: 6:33

And, as you said, in terms of the dangers, it potentially, in the wrong hands, it can be a very dangerous technology. We look at cryptocurrency. We look at hackers, for example. Do we look at cryptocurrency? Do we look at hackers, for example? They're going to be able to use this to potentially hack into all of our data. How important is it to build in structures to make sure that it doesn't fall into the wrong hands per se?

Kieran: 6:57

Yeah, that's the interesting bit, because it already has fallen into the wrong hands. So now people are using AI to attack cyber defenses and ai can attack at a scale that humans just can't you or I not that we're hackers might be able to. You know, type in 50, 20, 40 passwords, maybe every 10 minutes. It's putting 10 000 password attempts at websites you know now and every moment. So, again, that's the scary part. But it's not just hackers, it's social engineers. So remember the old uh, I say the old people are still doing it. You know they're sending text messages pretending to be, you know, significant others saying they've lost their phone. Now, with three seconds of your voice or a flat image of you or a little bit of video and we're all sharing these things on facebook and instagram I can use ai to recreate you, recreate your voice and, at scale, I can start to send messages to your phone or your significant other telling them that you've lost your phone. It looks, it sounds like you, you sound distressed, but I can put words in your mouth and I can now take money.

Kieran: 8:00

There has been examples of companies now naughty companies, as it were creating Zoom calls where they faked the board, brought in the CFO. They had what sounded, looked like, in every sense of the word, a real meeting. Obviously it's a virtual meeting. They invited in the CFO and said, ok, we're buying the company, transfer $25 million, and that was the last they saw of that money.

Kieran: 8:23

The tech itself, by the way, not just the fraudulent use of it, because we've seen that for decades new tech, new uses of fraud, old ways of doing it. Not that I'm condoning it, I dislike it intensely. But now we're starting to see extreme cases at volume because this is digital and very scalable. But the tech itself you do have to be careful about because it does something called hallucinate. It's only a prediction engine, a powerful and wonderful engine, but it makes stuff up, and I've seen lawyers go to court arguing legal precedent. Having used ChatGBT and other tools to seek advice and do the research. It's created what looks like plausible use cases or plausible legal precedent, but it's not. So we as people, never mind hackers, still have to check all of this technology and be extremely careful with what it's actually producing at this stage.

Miahael: 9:15

And in terms of regulation, for example, it could be regulated on in America, but they won't have the same regulations in China. How does humanity try to make sense or to try and have a cohesive regulation body on AI?

Kieran: 9:32

Yeah, I think this is part of the problem. I think every country is starting to see this tool and every company as something that can generate, you know, huge and tremendous competitive advantage, or military advantage. So there's an old phrase, you know America innovates, china duplicates and the rest of Europe regulates. And it's not a million miles away from the truth in some regards. There's about 193 bits of legislation that I've seen when it came to AI the last time. None of them are the same. So I feel tremendously sorry for companies and we're all global citizens these days and all global companies.

Kieran: 10:05

Everybody's trying to put something together. Some are putting more regulation in place, like the EU, eu than others, for example, like the US. Some are arguing that regulation is required before the genie gets out of the box. I would argue the genie's already escaped.

Kieran: 10:18

Some are arguing that we need a lightweight legislation to allow innovation to flourish, but what you're noticing is too many voices, too many mixes.

Kieran: 10:27

What I would like to see is some common standards, because that's how the Internet made itself so successful.

Kieran: 10:33

Rather than everyone doing their own thing, there was a set of common standards that allowed it to flourish, with protection in place. So if we could come up with legislation that protected the rights of companies to innovate, whilst cautiously protecting the rights of individuals like you and I, whilst combining that with governments, you know, encouraging their citizens to learn more about this technology, not only to use it, but to learn more about the shortfalls and the risks, and learn what hackers are doing and learn about hallucinations and where the technology can go wrong. I think if we did something and everything in that regard, I think we'd be in a much better place than we are today, because I'm seeing real risks and real harm happening and companies rushing forward not knowing quite what they're doing or where they're putting their data, and that's going to lead us into a very scary and maybe not so nice space. We still have time, but we really need to rush to catch up with this technology, with this technology and in terms of image generation.

Miahael: 11:28

We've seen a lot of news even today. At the moment, the Israelis have accused people in Gaza of using AI images as a form of propaganda. Obviously, we don't believe that, but it's an interesting use case kind of going forward as well as deep fakes for crimes. What are the methods going forward that governments as well as free individuals are going to be able to use to prove that they didn't attend a certain event or they didn't commit a crime? Is it something that has to be embedded into the video or into the technology itself?

Kieran: 12:05

It's a really interesting question because you know, when we look at wallpaper money, we embedded all sorts of security controls and people still manage to fake the notes. The naughty people are going to do or get around the fakeness. They'll not put any thread in place that will demonstrate that we're about to rip you off. Now this is the oddity. It's new technology, or reasonably new. Remember AI? The term itself was invented in 1956. So technically, ai is an 85-year-old overnight success story, gen AI being the latest version.

Kieran: 12:37

Governments have used propaganda for years. Go back to the World Wars, when there was parts of government put up to mislead the enemy, to convince the public to do something. Now you're starting to see this at scale. We're talking about governments using images, fakery, video. This has been happening for the last 10 years, 20 years and beyond People themselves. You take a quick canter down Instagram, x or LinkedIn and you will see fake content all the time. The scary bit and this truly is the scary bit is that you can't tell real from fake. I've watched some recent videos where folks have demonstrated the power of the technology and you wouldn't be able to tell that it's not the real person, and that has huge consequences. So there was a robocall, which is essentially automation technology, that copied Joe Biden's voice, for example example in one of the recent US elections and there was 20,000 phone calls made says don't worry, uncle, uncle, you know, joe has got this, we don't need you to vote, and I've seen that repeated across the world. You know Bangladesh, other countries. It sounds like the person. Now we're seeing video that looks like the person 3D video, not flat file. Now the technology allows me to take a flat picture, turn that into a 3D image and get you to say anything. So this is something that's very hard to regulate. This is something that's very hard to prove.

Kieran: 13:56

What would I recommend people do? Well, if I'm in business, rather than being invited into meeting Banas to transfer $25 million, I might put protocols in place that says two of the executive team need to be physically in a building. That's increasingly difficult when we work in a distributed digital world. But what about a code word? And for those parents who may be getting those phone calls or video files or audio files now to their mobile phone from the person, their significant other or related person? Because, again, if I'm on Instagram or Twitter, I can see who you're connected to and I can take their voice and their video and everything else.

Kieran: 14:31

I might be recommended. You know to come up with a code word of some sort that only they and their family know, but, relying on technology, you know to signal when there's something fake. I hope we see vendors like Google, microsoft and all the others start to help us, but I suspect that is not primary. I suspect we're not going to see a lot of that and I really suspect that those who are up to nefarious purposes are going to fake this better than probably the regulators can, or the regulators are going to be running to try and catch up because to date they've proven to be behind, because those who seek profit by harming others are willing to invest huge sums of money in this technology, because what they're getting out of it, unfortunately, is huge sums of money from me, you, banks and everyone else. So there's work to be done and I'm not sure we're going to see that filtering through anytime soon, which is really the scary part.

Miahael: 15:26

And just lastly, the difference between AI and AGI, which is artificial, general intelligence In your words, or, I suppose, in layman's terms for someone like myself, what's your take on that?

Kieran: 15:43

Yeah, it's an interesting debate because we see AI in everything that we do. So, for example, if you buy something on Amazon, it's got a recommendation engine in the background that's looking for buying patterns or signals that you have given. It Thumbs up, thumbs down the things that you have bought, and then what it's using is AI to crunch a lot of data to look at what other people have bought, who have bought similar things to you, and then it recommends those to you. Netflix do the same thing for movies, and then it recommends those to you. Netflix do the same thing for movies.

Kieran: 16:14

Agi is the latest what you might call trend or the latest aim for the large tech companies, and AGI is a company's running to try and recreate every facet of humankind, and what they're trying to do is create AI that is not only as intelligent as humans, but actually more intelligent. And we talk about intelligent. We're talking about looking, seeing, judging, working, working things out and given an instruction, and the AI goes off and does it itself. Now it's theoretically possible, if you think of any of the experts in the world who are giving you the timelines to do this. We have to be careful, because we always overestimate AI's ability in the short run, or any tech in the short run, we always underestimate its ability to be in the long run. The latest figures are well, maybe by 2030, we'll get AI that is more intelligent than us. That could, again, be very exciting. In other words, we can get AI to do some amazing things. Go and seek how to solve climate change. Be that digital twin that we've always wanted, that can give us the relationship, the financial advice. Go and do our job for us so that we can stay at home and do the things that we really want to do in life unless you don't want to work or you enjoy it.

Kieran: 17:23

But the scary part is well, hold on a second here. If we have got technology that can do absolutely everything that we can do, then the question goes well, what is humanity going to do and are we needed anymore? Now, those are two extremes from not needing to work because AI is going to do it for us and we're all going to earn money regardless to. Ai is going to take over and we won't need humanity anymore.

Kieran: 17:44

I think there's a lot of debate needs to be had and this goes back to your question earlier on about risks and governance and ethics and the responsible use of AI and getting governments and society globally in business. Just to have these conversations now, before we end up in a place where we might be frightened, why don't we just sit down, look at what we're doing, understand and this is difficult for even the AI researchers what's coming ahead, but why don't we understand what's coming ahead? So, at worst or at best, at least we've got a break that allows us to slow things down without killing innovation, but doesn't allow us to hurtle off the cliff, before we hand over all our agency and everything else to an AI that we don't understand.

Miahael: 18:25

Kieran Gilmurray, three-time author and the multiple AI award winner. The CEO and founder of the Kieran Gilmurray and Company Limited. Thanks so much for your time.

Kieran: 18:36

My pleasure.