Design Mind frogcast Ep.54 – Confronting AI’s Prophecy

On this episode of the Design Mind frogcast take a deep dive into the role of AI in shaping our current and future world. To do this, we’re joined by Thordur Arnason, a Vice President and Global Gen AI GTM Lead at Capgemini Invent. Thordur, a contributor to frog’s recent Futurescape report, is here to help us understand the potential impact this continuously unfolding technology can and will have in all of our lives, but more importantly, our role as designers, technologists, business leads and consumers in shaping it.

Listen to the episode below. You can also find the Design Mind frogcast on Apple Podcasts, Spotify and anywhere you listen to podcasts.

Episode Transcript:

Design Mind frogcast

Episode 54: Confronting AI’s Prophecy

Guest: Thordur Arnason, Vice President, Global Gen AI GTM Lead, Capgemini Invent

[00:00:03] Elizabeth Wood: Welcome to the Design Mind frogcast. Each episode, we go behind the scenes to meet the people designing what’s next in the world of products, services and experiences, both here at frog and far, far outside the pond. I’m Elizabeth Wood.

[00:00:18] Elizabeth Wood: Today on our show, we’re diving deeper into the role of AI in shaping our current and future world. To do this, we’re joined by Thordur Arnason, a Vice President of Capgemini Invent focused on all things AI, and a contributor to frog’s recent Futurescape report. Thordur is here to help us understand the potential impact this continuously unfolding technology can and will have in all of our lives, but more importantly, our role as designers, technologists, business leads and consumers in shaping it. Here’s Thordur now.  

[00:00:50] Thordur Arnason: In my whole long career, ever since uni, and towards today, I’ve had a feeling twice. This feeling is something you feel with your whole body, your whole being, you just know that this is important, and you just happen to be at the right place at the right time. And you need to go all in.    

[00:01:08] Thordur Arnason: First time I felt this was when me and a friend, we snuck into a computer lab at the local uni before we were at the uni, because we knew that one of the labs had internet access. This is in the 80s, and we spent all night there, telnetting, FTP-ing, talking to people online, and we felt like the whole world opened up to us. And that was really when I knew that this… somehow, I would be doing this for the rest of my life.   

[00:01:33] Thordur Arnason: Same feeling I had in ’22, you just know it. You know, with every cell in your being, you just understand that this is huge. It is transformational, not in like, you know, we like to talk about transformational things, but really, you know, this has the power and potency of turning our society upside down. And when you get that feeling, I think you just need to act on it. You know, nothing else matters. You just need to go for it. 

[00:01:56] Thordur Arnason: My name is Thordur Arnason. I’m Icelandic originally, but I grew up in Norway. I work for Capgemini Invent in Norway, but currently I’m filling a global role. and the work I do for global is focused on generative AI and AI in general, and I lead the global effort on go to market, basically meaning I spend all my time talking to clients, figuring out their needs, and co-creating solutions together with them. 

[00:02:24] Elizabeth Wood: While he’s focused today on helping businesses understand applications and use cases for AI, Thordur’s early projects with generative AI started from a more playful, explorative, and, dare I say, tasty place.    

[00:02:37] Thordur Arnason: The engineering part of me was looking at generative AI and what, what was happening in early ‘22 and that’s before ChatGPT and all that stuff. But what we were seeing is there were some really interesting models turning up. So summer of ‘22 I decided, instead of taking a break, I had AI write a cookbook. So that was my summer project. To prove that these models and this technology was, already back then, mature enough to do real stuff, to do real world stuff. Basically over 10 weeks, co-writing a whole cookbook with language models and taking all the food photos to follow the recipes in this cookbook with diffusion models. So image generation models, which was a super fun project. And after doing that project, I decided I can’t focus on anything else than generative AI. 

[00:03:26] Thordur Arnason:  Well, we always have, you know, a playful and joking side to everything we do. So what? What is the weirdest thing you can get a large language model to do? Yeah, well, that is focusing on human nutrition, human foods. Because in theory, this is so far removed from what a computer system should be able to help you have opinion on and create on that’s why we did that. We did it playfully by not allowing any cooks, any humans, anyone to check what the LLM was creating. The only thing we did was editing. This is just out of curiosity to see and understand the power of the then-interesting language models at the time, and we got we were so surprised over what came out of it. We didn’t expect the results we got, and that’s why we actually turned it into a full book that is available as an e-book, and even printed In some copies. 

[00:04:18] Thordur Arnason: The year after, I was doing a keynote at a FinTech conference, and I did a deal with the kitchen that did the catering, and they cooked three of the recipes from that book to serve at one of the dinners, and everyone was okay afterwards. So it turned out well. 

[00:04:35] Elizabeth Wood: As AI has evolved over the last years, Thordur has increasingly been working with teams in exploring new and emerging paradigms within the field, including an emphasis on agentic, or multi-agent systems, and the increasingly complex capabilities these technologies may bring.  

[00:04:52] Thordur Arnason: We have been with a team focusing solely on concluding on what is our definition of agentic AI and AI agents? Because what we started picking up early last year and gradually increasing in volume throughout the year, was that everyone was starting to talk about AI agents and then multi-agent systems, as some say, or agentic. Thing is that if you look at what we are really talking about, we are talking about taking AI from read-only mode to write mode. So we have read/write systems. That is the promise of AI agents not only helping us, providing us with information and making us more efficient, but also starting to do stuff for us.  

[00:05:36] Thordur Arnason: In its simplest form, an AI agent is just a system that can, on our behalf, do things that can be banal things like simple office tasks or more serious things in a bigger system. And once you start talking about the more serious stuff and the more complex stuff, that’s when you’d start to talk about agentic, because then you’re not talking about one AI agent doing one well defined, distinct task. You’re talking about a group of agents, or an army of agents that somehow are orchestrated to do a lot of things for you within a defined system or a network. this is a new topic for most of us, and all the major labs, all the major tech players, are focusing on this. And we see the development is going hyper speed as we are getting accustomed to in the space of AI. So, no surprise there. But right now, we’re not seeing the huge real-world impact of it. But I think by the end of this year, we’ll start to see really interesting things happening, and hopefully mostly for the best. 

[00:06:40] Elizabeth Wood: Naturally, these early explorations of agentic AI require critical and sometimes difficult conversations about the potential impact offloading certain tasks onto AI agents can have on businesses and consumers alike.  

[00:06:54] Thordur Arnason: Right now, the focus is on the technology and developing these agentic systems, but we haven’t really started the serious discussion on how will a human feel in an organization where you’re working alongside AI. 

[00:07:09] Thordur Arnason: When you start to think of the impact this will have on organizations, the impact it will have on human psychology, the social impact this could have, the conversation becomes really interesting, because it’s not about technology. It’s about our role as humans in an organization where potentially half of your colleagues are agentic. And then you have to think about how do we feel about AI and decision-making? How do we feel about AI and agentic pushing the operational efficiency? How do you feel about trusting and collaborating with AI? What is the shift in the workforce structure that follows this? Not to forget, how will this impact human dynamics in an organization? All these things are really big topics that we are just starting to discuss. 

[00:07:59] Thordur Arnason: As long as things are mostly theoretical, you’re not able to fully grasp the impact of it in the near future. Once it becomes tangible, once we start to deploy these systems, then this conversation will happen real fast. So we need to redefine what change management is within organizations when a growing part of the workforce is agentic. What is HR’s role in an organization that is blended agentic-human? What if we start building personalized AI agents for every employee? So, imagine you have every employee in the organization has an AI twin or a digital twin, or call it what you want, that learns from you, and you learn from it. And it gradually understands more and more of you and your role and how you do things. So if you’re off, having a bad day or a bit busy, it can fill in for you. These are things that will start to happen, and that is interesting to think about.  

[00:08:55] Elizabeth Wood: Earlier this year, Thordur contributed to frog’s annual Futurescape report, all about emerging trends and provocations set to shape the next quarter-century and beyond. His trend called ‘Confronting AI’s Prophecy’ explores the real-time shifts our relationship with this emerging technology may affect. 

[00:09:13] Thordur Arnason: So when I started writing and thinking about this, I was really preoccupied with the cognitive limitations of humans coexisting with AIs. So that was my starting point. And I think this is interesting, because I have observed on myself and on others that we as humans, we get tired, we lose focus, we lose our cognitive abilities much faster than these systems do because they just go on and on and on. You have this super powerful helper that is always 100% present, 24/7 ready, always sharp, always ready to help you. But it’s not always easy, as a human, to be in a state of co-working with a perfect entity, so to speak, cognitively speaking, if you get my point. And this led me to the interesting discussion that is going on right now. Is AI actually making us more stupid, or is it making us smarter? 

[00:10:09] Thordur Arnason: Some argue that done right AI will actually make you, as a professional, as a human, much smarter, and it will improve your abilities. Some claim that AI is leveling the playing field. So it may it’s making non-experts quicker, gaining expertise in the field they don’t have expertise in, enabling us to do stuff we couldn’t really do before. Even though we have a little knowledge, we can do a lot more within that field, because AI helps us along. And that is the definitely the thinking of AI being a democratizing force, because it’s broadening our skill sets, so to speak. So if you know a little bit about coding, you can become a much better coder using AI. 

[00:10:46] Thordur Arnason: While others argue that people are lazy and attempt to go with the easy solution, making us dumber when we delegate blindly to AIs. 

[00:10:54] Thordur Arnason: When people are tired or people are overwhelmed, they seek the easiest solution to whatever is in front of them, and then it’s really easy to start using AI in a wrong way, leading to worse results and not better results of the technology.   

[00:11:09] Thordur Arnason: There are strong arguments on both sides, and I think that is something we need to take seriously as we now to an increasing degree, welcome AI into our everyday roles and work. The way we approach it and the way we work together with these systems we need to get right. Because if we get it wrong, it will make us less capable, not more capable. And I think to a large degree, it’s still unexplored as an area of, call it, change management, call it upskilling, call it how do you introduce these technologies in a good way in a workplace? 

[00:11:41] Elizabeth Wood: We’re going to take a short break. When we return, Thordur will be speaking to this very point: how do you effectively introduce these technologies in the workplace? 

[00:12:18] Elizabeth Wood: Now back to Thordur Arnason, a VP at Capgemini Invent.   

[00:12:21] Thordur Arnason: Last year, I was involved in a couple of larger projects rolling out AI to the workforce, to almost everyone in a company. And what we were focusing on in the beginning there was obviously the efficiency gains that we would get out of doing this, which is important, of course. You have to, you have to have a rationale for doing this. Quickly, what we discovered, and we started working a lot more on in these, I call them, change management projects, is measuring joy, measuring employee experience, measuring how much more people enjoy doing their job because using the tools in these tools in a right way allowed them to do away with the most boring part of their job quicker and better. So we started measuring that. And we also started thinking along the lines of, this increases loyalty to the company, because your job is more fun and it’s easier to do, and it attracts people as well. So from the recruitment aspect as well, if you show that we are a workplace, putting humans in the center and augmenting humans and helping them with smart systems that make work more enjoyable, we are able to play with that as well. And that is us finding business rationale for not you know, forgetting about the human experience in it all. 

[00:13:36] Thordur Arnason: An organization that has a positive sentiment make better things. They serve their clients better, they make better products. The more fun and joy you feel in your job, the better the company is. Maybe oversimplifying, but that is definitely my observation. If people are safe and enjoy their work, then quality is good. 

[00:13:55] Elizabeth Wood: During our conversation, Thordur shared why it’s so essential to keep the human experience front of mind when integrating AI into business applications. 

[00:14:04] Thordur Arnason: I’ll give you an example. I worked with an insurance company last year, and in that company, they had a team, this team of four people. So it’s not a big project, but this team of four people, they do a process four times a year, once every quarter. And that process takes the team of four nine weeks to do. Most of their work here goes into doing this process four times a year, and it has been like this for many years. So what we did with that particular team is that we involve them in dissecting, deconstructing their process–the truth that had been standing for years. We broke it up into pieces, and then with the team, we designed a new way of doing the same process. Only we brought in AI tools, we brought in four or five different AI tools. And together with them, we designed a new process using AI where it made sense. And then not using AI where it didn’t make sense at the time. We spent three months with the team doing this, and at the end of that process, the outcome was what used to take nine weeks now takes three, which is super good, because that was what the company paid us to do: to find efficiency. But for me, the most important metric is that I have four people in an insurance company that absolutely love AI, understand how it can help them and benefit them, and they are now our strongest promoters and ambassadors for bringing AI into that company. So we created champions by allowing them to design their own work experience.  

[00:15:30] Thordur Arnason: And I think, you know, approaching this using design thinking, approaching this using the service designer mindset, approaching this using co-creation as a powerful tool—even though it’s technology and AI we’re talking about is a good approach in my experience, and that is where, you know, the design side of me comes in and comes in handy in that experience I have.

[00:15:51] Thordur Arnason:  You know when, when you’re doing work like I did in the early years of, you know, the web and the internet thing, you do everything you know, you don’t, you don’t even have, you know, labels on the stuff you do. You do the coding, you do the service setups, you do the visuals, the front end. You try your best to do everything. And I think, reflecting back on those first years, you are trained from the get-go to have an opinion and understand everything it takes to bring something to life, from the design side, from the structure side, from the coding side, and also from the business side. So we didn’t have the luxury of specialization back then. We just did it all. It’s only later, a few years later, we started to specialize, bringing in, you know, the design practices, bringing the people that handled more of the business dynamics and business side, structuring technology into who did the development, who did the testing, and so on and so on. So it is a result of being allowed to work in the very beginning of something that forms into what we have today.

[00:16:51] Elizabeth Wood: As AI has stepped out of the realm of only engineering and technology, becoming a part of our daily lives in new and sometimes complicated ways, Thordur has some guidance on how to stay exploratory and playful with AI, especially in creative contexts. 

[00:17:07] Thordur Arnason: On a personal level, I think you need to understand how to coexist with this technology in a good way. You need to understand or decide if you are delegating, or if you are enhancing your skills at any given time. So when you use AI right, or as I define right, it is a super amplifier for creativity. It is amplifying your ability to do everything from research to playing with concepts and to play with prototyping and building things and just playing around and experimenting. You can do that at 10 times the speed you could only two years ago. And this, I think, is under-communicated, underutilized, how much you can bring this in to make you so much better in what you do and have so much more fun in what you do.  

[00:17:56] Thordur Arnason: Two years ago almost, we decided to spend 10 days writing an opera with using AI. And why did we decide taking on an opera? Well, we decided to do that because we recognized that this is one of the more complex creative processes you can undertake, because it involves a lot of different skill sets. We did have a starting point with me, with my creative background, and my partner working, as a, with a formal background from theater and scenography, and with that, you know, skill background, we just undertook a 10-day experiment seeing how far can we get in creating a whole opera. 

[00:18:29] Thordur Arnason: And we did that to provoke,  How much can you amplify your skills and how much can you boost your creativity using AI correctly as a collaborator, co-working with it, iterating together with it, bringing it into both design and decision processes, experimenting with a speed of iteration that I have never experienced before, because we went 10 times the speed of what I normally do, which is interesting and exhausting as well. 

[00:18:56] Thordur Arnason: In that experiment, at the end, we actually had a libretto. We had score written, so music written, we had the visual concept ready, and we had the stage design ready. What came out of the process is no masterpiece, but it is a proof of how we can use these tools to amplify a creative process. And I think everyone needs to sit down and really think how I can bring this in to my work, to my role, to the things I do, to amplify my creative skills, to amplify my capacity of doing things. 

[00:19:26] Thordur Arnason: And this is not being done enough today, because most people tend to bring AI into their work and delegate to AIs. Can you please summarize this for me? Can you please find this? Can you please this and that? So you use the AI as an assistant, which is, of course, natural to do, but you do not use AI or AI bots or agents. Call it what you want. You don’t use them as collaborators. You don’t bring them into your role and into your work. So only when you do that you start to really reap the benefits of this.    

[00:19:59] Elizabeth Wood: Yet, not all AI use cases are made equally. During our conversation, Thordur warned why it’s not exactly advised to think bots are always the solution to every problem. 

[00:20:10] Thordur Arnason: I wrote a little thing on something I call botflation last year, and that is just that small example of things we are we haven’t thought through and we were not really ready for. And that is, in an organization, I have observed a few times that we build a bot for this, and we build a bot for less that they’re all powered by AI, so they’re smart and they’re helpful. Thing is that each of these bots, they have their quirks, they have their personality, so to speak, they have their guardrails, they have their do’s and don’ts and a slightly different interface. So in an organization, I’m trying to make work better and people more happy, and then I gradually introduce 11 new bots that will help you. They are distinct, they are not connected to each other. And in reality, what happens is you get what I call botflation. It’s an inflation of bots trying to be helpful for you, actually ending up in deteriorating your ability to do a good job, because all you end up doing throughout your work day is context-switching. Lowering your efficiency, when the goal was to make you more efficient and more happy. We didn’t intend that to happen. We just didn’t think through the totality of the effect of introducing the technology. 

[00:21:15] Elizabeth Wood: Something impacting the speed of change of AI systems is the incorporation of government policy and influence. Different regions have different attitudes toward regulation in this space, which is why Thordur feels it’s so important to have a global perspective and draw from learnings around the world. 

[00:21:31] Thordur Arnason: Only last year, you know, we, in the EU we were dead set on ensuring responsible development of AI systems and ensuring that everyone using AI in their businesses do that in a regulated and responsible manner. It’s hard to disagree with that, because, of course, we want to make sure that things are properly regulated, and we don’t create systems or solutions that spin out of control and can do harm. So intentions are the best. Kind of the same thing was happening in the US and the Chinese have their version of that. So everyone was kind of on board. Let’s make sure we have at least some basic regulation in place so we have somewhat of an oversight of the rapid innovation going on in this industry.  

[00:22:16] Thordur Arnason: And then January came this year, and everything changes. You have the powerhouse in AI innovation and development globally is the US still. And what we are observing is that the US is taking pretty dramatic and quick steps on deregulating, so they’re taking their foot off the brake pedal. Actively removing anything that slows down what they call innovation. Strengthening, then, the position of the US-based companies doing this because they have less regulation to worry about, and then they can go faster. They can go faster in their labs. They can go faster developing and innovating on their models, and they can bring it faster to market. 

[00:22:56] Thordur Arnason: Now, if you then jump over to the EU and what we’re doing, we are slowing down. How does that leave us? If you look at competition, what does that leave us? What situation are we in? If you think about innovation, are we able to compete? Are we able to innovate? Are we able to create beneficial systems that will not be out-competed by actors that are not regulated at all. And this is an interesting discussion happening right now. So everything we believed we were doing right last year with the EU AI Act and regulating things is now coming under tremendous pressure. It’s important to understand that there is a geopolitical play going on right now.  

[00:22:33] Thordur Arnason: You have the Chinese are doing great strides. They’re increasing in importance, and they’re producing PhDs faster than anyone in the world, currently gradually overtaking, or not overtaking, but closing in on the US in this space. And in a few years, if nothing changes, They will be they will be number on, US will be number two, India will probably be number three, and the EU will be number four, in this extremely intense race for intelligence. That is one thing.   

[00:24:02] Thordur Arnason: The other thing that is interesting to talk about on this is, what does this do when you have a geopolitical climate where nationalism is put first. So America first or Europe first, or India first or China first, and global collaboration is falling and we are building walls instead of cooperating and collaborating as we have done for decades now. What will this do to the development of AI and this technology? These are the big questions we are trying to figure out answers on right now.    

[00:24:33] Elizabeth Wood: Ultimately, Thordur feels the intentions behind the integration of AI into business and society are what will keep the technology as rewarding as it is valuable. 

[00:24:42] Thordur Arnason: So what I want to do is to make sure that we, as humanity, harvest the power of the technology we’re playing with for good. I want to work with clients, and I want to work with problems that make some kind of difference.  It could be about making money. It could be a bit about making us more efficient, but it also, and preferably, should be making our societies better somehow. Because these technologies are powerful and they can be used to do a lot of harm. Our job is to ensure it’s used for good. 

[00:25:13] Elizabeth Wood: That’s our show. The Design Mind frogcast was brought to you by frog, a leading global creative consultancy that is part of Capgemini Invent. Check today’s show notes for transcripts and more from our conversation.  

[00:25:24] Elizabeth Wood: We really want to thank Thordur Arnason, VP at Capgemini Invent, for joining us to expand on his trend in frog’s Futurescape report, which you can also find a link to in today’s show notes. 

[00:25:35] Elizabeth Wood: That’s our show, the Design Mind frogcast was brought to you by frog, a leading global creative consultancy that is part of Capgemini invent. We really want to thank our guest, Antonello Crimi, for his insights on the future of our interactions with technology. Find a link in the show notes to learn more in frog’s new Futurescape report. We also want to thank you, dear listener. If you like what you heard, tell your friends. Rate and review to help others find us on Apple Podcasts and  Spotify . And be sure to follow us wherever you listen to podcasts. Find lots more to think about from our global frog team at frog.co/designmind. That’s frog.co. Follow frog on Twitter at @frogdesign and @frog_design on Instagram. And if you have any thoughts about the show, we’d love to hear from you. Reach out at frog.co/contact. Thanks for listening. Now go make your mark.