Dave Drodge on Why Imagination Beats Efficiency in AI Adoption

Most companies treat AI as an efficiency play. Cut costs, save time, do more with less. But what if that mindset is exactly what's holding them back? 

In this episode, I'm joined by Dave Drodge - digital transformation and AI strategist with 25+ years across Roche, Novartis, Sony, and Amadeus - who argues the real AI opportunity isn't about doing things faster, it's about reimagining what your people and business can become. 

We discuss: 

• Why "human in the loop" often means "person who takes the blame"
• How IKEA turned AI-displaced customer service roles into $1.3M in new revenue 
• The difference between experiment, exploit, and enlightenment in AI adoption 
• What the EPOCH framework reveals about the skills AI still can't replace 

Connect with Dave Drodge: https://www.linkedin.com/in/davedrodge/ 

Connect with Valeriya:
LinkedIn: https://www.linkedin.com/in/valeriya-pilkevich
YouTube: https://www.youtube.com/@aimadesimpletalks

Need help building AI capability in your organization? Book a call. 

Transcript

SPEAKER_00 0:00

Most companies stay there adopting AI, but what they really mean is they're using it to cut costs. So what happens when efficiency becomes the ceiling instead of the starting point? Welcome to AI Made Simple, the transformation series. I'm Valerie Velkievich, and I talk with global leaders, innovators, and practitioners who are shaping the future of work in the age of AI. In this episode, I'm joined by Dave Roch, digital transformation and AI strategist with about 25 years of experience across Roche, Novartis, Sony, and Amadeus, now lecturing on AI and digital leadership at Northwestern University in Switzerland. We talk about why most organizations get stuck in AI experimentation and never reach real impact, why human in the loop often means the person who takes the blame and not the person with agency, and which human skills are actually rising in value in the age of AI. Dave, it's great to have you on the show.

SPEAKER_01 0:58

Thank you very much. Really, really appreciate the invite and great to be here.

SPEAKER_00 1:01

Dave, you've built such an usually, unusually broad career across countries, industries, agency life, global organizations, and now also academia. Can you take us a little through that journey? What were some of the defining moments that shaped your perspective on transformation and maybe the role that technology plays in it?

SPEAKER_01 1:21

Yeah, great. So, you know, I've worked in digital for over 25 years. And I started out when the web was just coming of uh, you know, presence of interest. And so at that point in time, myself and a friend started a web agency because we were interested in technology and really believed that it could change the world. You know, looking at that now, it's a case that that's even more so. Uh and we couldn't have imagined where AI has brought us. Well, across that journey, I've really uh been like a digital lighthouse keeper, really kind of guiding organizations through each wave and giving them, like, for instance, some expertise in my side to how we can actually navigate those different waves. So I think for me, one of the big ones as well was uh e-commerce. And I work for a company called Amadeus, which is a background, what they call a global distribution system. And uh at the time it was disrupting, you know, travel agents, because we used to have to go and actually go and speak to someone and say, you know, can you book this travel for me? And on the one side, we had uh travel website where you could actually plan your trip, and then we had checkmanship as well, where you could actually uh go and actually find out your itinerary and download your e-tickets. So, I mean, in those times, it was actually quite revolutionary times uh from that point of view. And over my career as well, I've worked in nonprofits as well. So the World Wildlife Fund, and that was an amazing experience from the point of view of partnerships, really, this idea of you can't do it all by yourself. You have to go with other organizations that are willing to work with you. And uh there we actually had some really great uh partnerships, like for instance, at the time with Twitter, for instance, and we had a Twitter bot that actually looked at people's Twitter feed, saw ones where their emojis were their endangered species, and then we tweeted them and say, hey, do you want to become a kind of a spokesperson for WWF around that? And so that was a really great example of taking the brand um trend, like for instance, emojis at the time, and then bringing together the technology to actually make it happen. And so, as you mentioned, you know, the last thing I've been really focused on is more on the digital transformation side of things. And uh, luckily, my roles recently that's employed employed uh AI and actually bringing that to the whole organization. So, really giving people the skills, giving people the confidence in how to actually use this technology to actually further the goals of the organization. So, yeah, so it's been a it's been a whirlwind tour. And I would say one last thing is that um I think where I'm going to next, it's a case that's uh, you know, lecturing and the like. And, you know, that's really goes back to one of the things I really feel is valuable for everyone in these days is actually this continuous learning. It's a case that the world is changing so quickly. We need to keep up. And that's a way to actually do that, to always be looking at how to learn. And that's not just taking courses, but it's a case, of course, like, you know, like day-to-day looking at our feeds, looking at who's interesting in our field, and actually always then collaborating with people to figure out what the next fixed thing is and how to actually apply it.

SPEAKER_00 4:17

You mentioned this tweet, it was very interesting, and I'm sure it made an impact as well. What I read in one of your articles that interviewed you, that also while working at uh Roche, you also managed with AI to uh cut down time of, I believe it was clinical reports from 12 weeks to was it 10 minutes, uh, 10 hours? You can tell us more about it. So it's really at the end of the day, technology should make our lives better. So, my question to you is that you've described three waves of AI adoption: experimentation, exploit, and enlightening. Uh, the three E's. Can you walk us through what that means and why most organizations seem to get stuck at this experimentation without really moving into the impact or this enlightenment from the technology?

SPEAKER_01 5:00

Yeah, no, that's a great question. So I think that uh one thing is that we all stand on the uh shoulders of giants. So, like for instance, Alex from Strategizer, the first two, uh experiment and exploit, are from his methodology. And then I add enlightenment because I think that one of the key things today is that a lot of companies and on the flip side, a lot of employees really look at this as an efficiency ploy. And having said that, there are lots of good examples of that. You mentioned one. And on the other hand, it's something where I think it's a lack of imagination. And that's one of the things that we still hold that uh, you know, people are better at AI about. I mean, really new things. And, you know, looking at how do we actually, you know, grow the top line, how do we actually look at that? And, you know, I've had a good example of that, two Swedish firms. I mean, one of those Clerna, and they came out and said, we're gonna get rid of our customer service people and replace them with a bot. Uh get rid of 700 jobs, and then they have exactly, exactly. And then the flip side, another Swedish firm, IKEA, actually had a different attitude. They actually used a bot, but then they actually looked at it from the point of view of, hey, we've got customer service people. They know our products in and out. What are we gonna do with them? We're gonna retrain, I think it was 8,500 customer service people, and they put them into a situation where they were doing virtual consultations online. And that actually generated$1.3 million worth of revenue. So that's the kind of thing that we need to see more of. How do we actually do something better? How do we allow people to thrive, not just, you know, get a couple of cents on the dollar from the point of view of efficiency gains? And going back to your point, and you know, where we started this conversation and this question, experiment is really sometimes about that, is dimes on the dollar, and it's not actually looking at the big picture and what we can really do with. So I think it's a good thing from the point of view of a starting point. And uh, but you get to a certain point where you have to say, okay, one of these things is actually going to scale. Where do we put our real focus as an organization on it? And then how do we actually, you know, exploit that to actually get, you know, like I said, better top line, you know, better bottom line as well, of course, but all those things together. And, you know, we really have to be broad and again use our imagination.

SPEAKER_00 7:19

Okay, thank you. You provided a very good example. And I, by the way, had on the podcast on the show also somebody from IKEA talking about skills-based approach to uh yeah, basically training all the floor workers and everyone on using AI and generally like designing trainings that's relevant for them. So also putting human in the center of the business and not like kind of outside of it. You mentioned as well some of the layoffs. So I think it concerns many people right now, right? And we hear also many stories and just a couple of weeks ago that the stock market went down for several SaaS companies because investors predicted they don't need analysts anymore. They might just need instead of 20 analysts, five analysts with a clawed account with an AI, right? Can you tell us your opinion? So shall we be scared about the jobs or do you have a positive outlook? Do you believe AI is going to create more jobs?

SPEAKER_01 8:14

Yeah, yeah, that's a great question. So I definitely lean towards the abundance mentality, but with uh a balanced approach. So it's a case that, you know, if companies are looking at it and say from top down, this is an efficiency gain, then you know the what's gonna come is gonna come, you know, naturally the the knock-on effects will be when AI is actually able to do certain types of work better than people or good enough, then it's gonna be a classic disruption play. Having said that, again, um, I really believe that this is a lack of creativity. This is something where, like, you know, there's not many businesses in the world where they cannot expand. And this is a great way to do that with people that know the business. I mean, this is the thing. I mean, you're gonna need certain types of new roles. And again, like the AKA example, do we have people in the organization that are willing to upskill for those? The other opportunity, which is really great as well, is that in certain circumstances, the roles become unbundled. And then when you look at rebundling them with AI, you don't necessarily need someone to have such a high level of education or experience. And then sometimes the AI can actually help with that as well. So I think there's a lot of opportunity, and and that's the way I like to look at it, but it definitely is something where we need to be wary of the fact that it can go into a very dystopian direction. And I really believe as well that you need to be critical about what you're hearing. If the head of like these AI labs, like for instance, especially in uh the US, are talking about, you know, getting rid of jobs and things like this, that's actually to their benefit. I mean, they're they're actually the ones that are actually going to capture the value from that. So you need to be very critical as well where the source of that is and what their uh alternative motives would be. And that's again another human uh quality, you know, critical thinking.

SPEAKER_00 10:02

Um, Dave, you were talking also about efficiency gains from AI. And we already see, in fact, some efficiency games. Game companies are seeing um some data, sh some data from ADECO, I believe, mentioning that AI saves people two hours a day. But I believe there is also a big expectation gap. So if AI saves me two hours a day, it's at least ten hours a week and maybe more. So what is expected to do with this freed up capacity? And I think there's also sometimes lack of communication from management. Right. On the other hand, we see studies that show that people tend to take more work and more work and more work, and which leads to burnout at some point. There was also a recent HBR study on it. So it's not sus it's not long-term, also, it's not very sustainable. How do you see this play out inside enterprises, inside organization? And what do you think uh should the leadership approach be?

SPEAKER_01 10:56

Yeah. So I think if you look at it from an abundance point point of view, then part of it is actually, you know, you will gain benefits from it. And this should actually put people in the situation of actually having some more time. And with that time, they should be looking at the doing the things that uh people are good at, having relationships, going deeper, asking the right questions. These are key things that us people, we need to keep doing. And also with that freed up time to actually continue learning, to kind of share the learnings that they get from things like AI and make sure that the actual organization is continuously learning and benefiting from that. And I think that's one of the key things as well, is that a lot of organizations haven't upskilled their workforces. And I think that's a a big gap. Um, and because of that, then people are fearful. So management leadership actually has to really come out and say, well, what are we gonna use this for? What are what's our purpose with this? And again, if they're looking at just bottom line, I mean, no, that's really uh not where we're you know, you see the the the best use of that uh resources.

SPEAKER_00 11:58

So, what would be what would you say that timeline should be? So first uh getting these new tools and then upscaling and then communicating, or first communicating or transforming. So, how would you see it uh having this much experience in the transformation areas in big um corporates?

SPEAKER_01 12:16

It's a really tough question, um, and it's difficult to get it right. And I think um unfortunately, I will use the it depends uh uh answer. So, in an organization that really um is open and uh it's a case that uh already, you know, very much you know, a continuous learning organization and everything else, I think that that you know that would be a good thing to actually get the uh upskilling program ready first, and then like for instance, obviously a part of that is the technology as well. Uh, although a lot of these technologies are more or less interchangeable and and always there's this, you know, race between them, but it's a case that, like, for instance, they're doing similar things. And, you know, one of the things um I would say is that it shouldn't be something where it's a knee-jerk reaction either. So, for instance, like, you know, the CEO comes and says, we have to have AI and I want it by the end of this year. That's the wrong way to go about it. It really should be something where it's based upon where the enterprise is today, what are the objectives, basically, you know, how are we going to use this and to have a think about it. It's okay to experiment with lots of different things as long as you follow the current way you go about, you know, looking at these different types of software to make sure that you've got uh the risk covered off and you're looking at the right reward. So I think it's really important that uh, you know, leadership is very much involved with this, role modeling it, talking about it. And therefore, then people and their studies show this, people will come along behind it because they want it uh in a lot of cases. And there's lots of studies saying as well that especially when uh ChatGTP came up first, there was a lot of people just using it in the shadows. So even though it wasn't actually you know approved or this kind of stuff, they were using it, and unfortunately, they were putting company information in there which was not safe.

SPEAKER_00 13:59

Yeah, I recently talked to uh also a leader from from an organization, and he was it was a very um anecdotal story. Uh he said our company rolled out a challenge, AI challenge, and everybody could submit their AI use cases. So we didn't tell them like what use cases or what tools should they use. And then he mentioned they realized after people submitted the use cases that over 50% of the use cases were submitted, not using the company tools.

SPEAKER_01 14:26

Yeah, yeah, exactly.

SPEAKER_00 14:28

Not the enterprise tools, but kind of as you mentioned, like kind of private Chat GPT account or or Claude's account or or any of that matter. So it it was, I guess, a good learning for the company as well, but also I think it's in every company right now the case as well. The shadow IT or shadow AI right now, definitely you're mentioning. When we um think about AI adoption, and in some companies, obviously it's um they have a higher rate of A adoption, some companies are a bit behind. So, what do you think differentiates these two companies from what you mentioned is I heard continuous learning culture, also leadership support, or that leaders, senior leadership is actually driving it. Uh, what else would you say are um are the points?

SPEAKER_01 15:08

Yeah, and I think that you had a really good example there as well, is that then it's like, you know, if you've got the top management on board, if they're role modeling it, then you need catalysts, you need an army of people that are actually saying, hey, this actually works, and I'm gonna share how I'm using it, and I'm very curious to hear how you're using it as well. And that learning culture is actually what will actually make it take root and actually flourish within the company. And again, going back to the three E's, then you want to look at, okay, what are the use cases that we found that are really gonna change our business? Hopefully, more of the top line side of things, but of course there's gonna be some efficiency gains there as well. And look at end-to-end process redesign and you have to have people on board. It's something where if they're not on board, they're gonna resist, they're going to maybe even sabotage what's happening, even. And because of that, it's something where you really want all people going in the same direction.

SPEAKER_00 16:02

How do you make sure that people are, that you get people on board?

SPEAKER_01 16:05

Well, I mean, I I think that one of the key things is that you with these tools, I mean, you do have monitoring available. Like, for instance, so you can put up your dashboard and say, okay, these teams seem to be doing it really well from the point of view that they're following whatever we've put together, for instance, they're using the tools, soft things as well, like for instance, you know, qualitative surveys to see where people are to start the baseline, and then to see how you mash against that going forward. And then you can course correct. Uh it's really critical that it's not something where, you know, you're just like, okay, this is the plan, it's in stone. You really always have to be iterating as well to figure out what works. And again, I think that one of the key things is, and there's a BCG study about this, that like, you know, combine not just, you know, kind of upskilling, like, you know, traditional kind of classroom kind of thing or online training, but actually getting people's hands dirty with the technology and sharing that in the teams. And that's where one of the great things to have, as well as champions that are actually kind of pushing that forward. The example you had there a hackathon, these types of things where people really, you know, you you uncover who the power users are and the ones that believe in it and not only the ones that are technically able to get something out of it. And I think that that builds that force, that kind of army of people that will actually push it forward.

SPEAKER_00 17:23

I think you mentioned many, many good insights. So from continuous learning culture to leadership actually driving adoption to having a team of AI champions or sort of role models who would be inspiring also the rest to use these tools and pro like internal promoters, let's say, of this transformation and of the technology. But also you mentioned that use cases or trying to find use cases that are solving problems. It's not just rolling out engineering training with just one same use case, like how to, I don't know, draft your email replies to all the organizations, though it's important, but really looking at what use cases do the teams have, how can we use AI to solve these problems? And that's where again, champions, for example, from the teams can help, right? They can be the first to identify it and to apply and to share with others, for example. Yeah, and and being hands-on. So that's also what you mentioned. Basically, I think it it goes together the identifying the use cases, but also like trying to implement them right away and see the gains for yourself. Um, Dave, you talked about two models, the Centaur and the Cyborg. The Centaur draws a clear line between human and machine tasks, and the cyborg lends them throughout. From your experience driving this across enterprise teams, uh, which model emerges in practice?

SPEAKER_01 18:39

Yeah, so um, this is actually based upon a uh Harvard Business Review uh study as well. And one of the things that's there is that they have different outcomes. But having said that, for a lot of people, the Centaur is a lot more natural. Like for instance, they know that, okay, can look at so many reports and look at crunching all this data and kind of coming back with a good response. And therefore they just kind of outsource that part of it. Um and the cyborg is much more iterative. So looking maybe at the subtask and saying, oh, okay, this is it's good for this, and then having interaction with it. And I like, I don't know where the source of it was, but there was this really nice analogy where it said, like, you know, Google is like um a dog fetching uh a stone or not a stone, but a book or a branch or something like this. And, you know, like for instance, the large language model chatbots is more like having a conversation with a colleague. At the time it was more like a junior colleague. But I think that that's exactly it. It's something where so the cyborgs are kind of more having that interaction with the actual technology, trying to figure out what are the good things about it, what are the bad things about it. And uh, in this study as well, Ethan Mulks a lot about the jagged edge. And it's not like a kind of a neat curve where you can see this is where AI is good and where the human is better, but it's something where it's jagged, and you have to kind of always be playing with it a little bit, experimenting to figure out what is going on there and see how it's changing over time as well. Because sometimes it gets worse as well. I mean, like for instance, there's some really interesting things right now where people are saying, you know, a year ago I was using this version, it was fine, and this kind of stuff, and now the new version is doing a bad job. So it's something where it's not a consistent thing that we can kind of rely on.

SPEAKER_00 20:21

I think it also um relates to this other aspect of where we actually outsource our thinking to AI and where we augment our thinking with AI. I believe there is also another model which is AI as a co-pilot versus AI as co-thinker. So co-pilot being when you like outsource the job, the full tasks to AI. And co-thinker is where you use AI to spot maybe the flaws in your own thinking or in your own decision, or you use it again to zero as a reflection partner or as a as a mirror, let's say, to mirror your decision to tell you where you might be wrong. So I think it's also an important mental exercise for each of us to try and understand, okay, where can I use AI to actually improve my thinking, not to just, you know, outsource my thinking completely.

SPEAKER_01 21:11

Yeah, exactly. And actually, this study has been updated uh last year, and they actually brought in ex pretty much exactly what you just said, which is actually this idea that, you know, what are the things that we're just outsourcing to AI? And it also said as well, depending on what you're trying to do, it might be fine to do that. Even from a learning point of view, is something where this is a productivity issue, is it something where I have to get this out this second? And therefore, yes, you know, you can actually say, okay, I'm gonna let it do that. But there's also the downside of that, which is cognitive offloading. And it's a case that based upon that, it's something where someone's skills in that area will get worse over time.

SPEAKER_00 21:49

Your motto is be human, not just human in the loop. I think it plays out nicely in this conversation. Can you unpack it even more, especially for leaders who would hear, okay, human in the loop and think They've ticked the box on people's side of AI.

SPEAKER_01 22:02

Yeah, and I and I think there's a really good analogy to that. So, like for instance, if you look at um uh actually I've did some legal studies uh in the last couple of years, and it's a case that part of you know human the loop sometimes is just actually to tick a liability box. So if something goes wrong, this person did it, and they are legally liable. And therefore the company is kind of you know kind of covered. And again, this idea, you know, this third case of actually you know abdicating and outsourcing our thinking, this is something where ultimately, as an you know, people, employees, you know, in companies, we need to take responsibility for what we have actually approved. So having said that, the this third case, like for instance, if you look at the human in the loop, is kind of like uh in a car, the crumple zone. So when you know there's an accident, there's a part of the car that actually compresses to take the damage and in this case to take the liability. And so therefore, like for instance, we need to make sure that when we talk about human the loop, we're talking about true agency, where someone doesn't just say approve, disapprove, but they can actually do something about it. And and then that's exactly why I say be human, because it's a case that first off, it's something where then you have agency to do something. But also, second of all, it's like, for instance, how do we as human beings stay relevant and also again human? And so part of that is like, for instance, putting our fingerprint on the things that we do so that when you see something from a certain person, you say, Oh, okay, I know this came from this person, not from a botless site, for instance. And also, it's also like, for instance, if you look at a really great study that uh came up from MIT about what are the characteristics of people that actually, actually, you know, AI is still having difficulties with, and they call it epoch. And, you know, the first one's emotional, the second one is presence. You look at presence, for instance. I mean, what we're having here right now, a video conversation, is actually sometimes even using AI or at least, you know, kind of acknowledging the fact that this is actually a certain important skill that we need to have, and then trying to increase that. Of course, physical presence as well, even more so. And the other parts of that as well are like, for instance, like opinion, which is around, like, for instance, like, you know, judgment and again, you know, taking responsibility for your work, creativity, and uh they use uh hope, but they also talk about you know leadership and vision in there as well. So that epoch framework is quite good, and they actually prove that, like, for instance, they looked at uh 20s and 2024, and they could show that actually these epoch skills are the ones that actually have gone up in value, even post um GPT.

SPEAKER_00 24:37

Okay, so um emotion, presence, opinion, and creativity. So we we call it, we'll call it the future skills.

SPEAKER_01 24:44

Exactly, and the hope as well. And and that's where more leadership and vision, and again, it's something where you know these are the types of things where AI can emulate some of it, but through epoch skills, it's a case it's still a human uh characteristics.

SPEAKER_00 25:00

When you look at how organizations are approaching AI today, what's the one mindset sheet you believe leaders most urgently need to make?

SPEAKER_01 25:08

Again, I think that the first thing is like going back to this vision. It's like, what are we trying to do? And I hope that they're looking at this more as an opportunity to grow than it is to actually just be efficient. I think that that's the key thing right there and then. And then back to continuous learning. It's something where this is really like if we free people up to have some time, then and reinvest that in actually how we're gonna go forward, but also in our human capabilities. And I think those are the key two things.

SPEAKER_00 25:36

And I have two more fireside, very quick questions for you.

SPEAKER_01 25:39

Go for it.

SPEAKER_00 25:40

First, what's the one AI tool you personally can't live without right now?

SPEAKER_01 25:44

Well, that's a good question. I have to be honest with you. Yeah, yeah, exactly. I was just gonna say that uh I actually am using Comet Browser, which is uh perplexity. But having said that, can't live without, but I don't trust it is the key combination. And I have just have numerous examples where it's kind of gone off. And also it's been lazy as well. It's just like going like it's like, okay, I you've asked me to do these things. That's gonna take a long time. And this is gonna stuff like, oh my goodness. I mean, and then I say just continue and do it.

SPEAKER_00 26:12

Okay. So how do you, if you don't trust it, how do you do you go to incognito mode when you open like, I don't know, bank accounts or some confidential information?

SPEAKER_01 26:22

Well, well, having said that, I'm um I use it for very specific types of tasks. And okay, okay. So it's not your default browser. No, no, no, no, no. But having said that, it it I found it very interesting that you know, some experiments I've done with it, you know, where it's actually gone into like um a copy of uh a spreadsheet that I've done and written over columns, all these types of things. I'm like going, it's not ready. And and having said that, I mean that's that's that's it's it's some concrete examples, but um but also it saves a lot of time as well. So it's it's still a good tool, but you just have to watch it.

SPEAKER_00 26:53

Okay, I got it. And uh the second question, uh, what's one task from your career? I mean, it doesn't have to be now, but maybe if you uh think back, that you'd happily outsourced to AI or automated with AI years ago.

SPEAKER_01 27:06

I would say most of my jobs with reporting. It's a case that and I think that that's something where really, yeah, there's a funny cartoon from ArcTunist, and it's like for instance, this idea of like, for instance, that, oh, I need to write this email and you give it a one one prompt and then it makes something big. And then the person on the other side is actually looking at the email and then going, you know, can you give me a summary of that? And I think this is exactly the thing, like for instance, this is exactly the thing where those type of things that we do that we don't need to do, and reporting is a really good example of that, is something where that's what we should be cutting out, and that's where the end-to-end enterprise view of things is so important.

SPEAKER_00 27:46

Yeah, thank you. So it's also a positive outlook for the listeners that AI can also free up your time for more valuable tasks for more high level and uh for tasks where you really enjoy your work so that you you like it more.

unknown 28:00

Yeah.

SPEAKER_00 28:00

You outsource the things that we dislike to AI. Thank you. Thank you, Dave. It was a great discussion.

SPEAKER_01 28:06

Thank you very much for having me.

SPEAKER_00 28:07

You can find Dave Drawch on LinkedIn. The link is in the show notes, along with all the resources and frameworks we mentioned. If you enjoyed this episode, follow AI Made Simple, the transformation series, for more conversations with practitioners and leaders shaping how AI is actually adopted inside organizations. Thanks for listening.