Episode 7 - AI Adoption or AI Transformation?

Brian Milner (00:23) Welcome in everyone. This is the People Over Prompts Podcast. This is the podcast where we talk about the future of teamwork here in the age of AI. I am Brian Milner. I am the host here for the show. And I'm excited to be with you today. we've got a good topic to to bring to your attention. before we dive really too too much into that topic though, I just wanted to make one quick announcement here before we we go any further. I I I like to take the opportunity to have some face-to-face time with people who are listening to the podcast and try to find opportunities to do that sort of thing. So I I just want to make sure that this is on your radar. if there's there's a chance this could could be something you're interested in, then I I hope you'll consider it. But I was invited by the wonderful people who are putting on the the regional scrum gathering for Central America this year to be a keynote speaker for them. And I gladly accepted that. I'm going to be in Guatemala November fourth through sixth this year. It's a a really amazing place. It's in Antigua Guatemala. It's a place called the Porta Ho Hotel. It's a resort kind of hotel. so it's a really nice place. It if it's something that you're considering, I'll I'll tell you you won't be disappointed in the in the environment. It's you know, resort type hotel. And there's gonna be a lot of great agile thinkers and leaders there that are talking about this new age that we're entering into. the in fact the the theme for the the conference itself is called Human by Design. So I I'm really excited about that talk, and I just wanted to put that in front of you before I dive into this week's conversation. in case that might be something you'd be interested in, just check it out if you're you're interested in that. You can go to the Scrum Alliance website and look for their regional scrum gatherings and you'll find that listed there on on upcoming gatherings. but I wanted to dive into something this week because there's a there's a report that I saw. a survey, a good piece of data that came across my desk that was put together by a little company that you may have heard of called Deloitte. they put together this this research paper that's called the the 2026 global human capital trends research. and it's it's a a really interesting collection of information that I I thought was really informative for some of the things we talk about here on the show. And wanted to bring it to your attention. they they did this in partnership with Oxford Economics, and they surveyed more than 9,000 businesses and HR leaders. They did that across 89 countries. so it's it's a widespread pool, a data pool there that they pulled together for this. And there's some really interesting things that that came out of this. Now I I titled the episode AI adoption or AI transformation because I'm I'm trying to draw a distinction here a little bit. that's kind of what I'm taking away from this on a very large level. So let me let me hit you with a few of the stats here that came out of this and and just kind of work work my way through it with you. first first one out of the gate, this one shouldn't be too surprising, but they found six nearly 60% of workers. intentionally use AI at work in some way, shape, or form. Not too surprising. I I might be a little surprised it's that low in today's world. Maybe, you know, maybe there's a lag a little bit here in when they took the data to to present day. I would expect it's even more if it was a snapshot from today. the another kind of interesting portion of that most organizations, 59% almost that same number Are taking a tech focused approach to AI. Now, this is a very clear distinction they're trying to make. They're saying that they layer AI what they're saying by taking a tech foc focused approach to AI is that they layer AI onto legacy systems and processes rather than reimagining how humans and AI interact, how they collaborate, and how they make decisions. And that's obviously something that we're trying to focus on here in this podcast is really that difference in in approach. the stat that kind of went along with that was only 14% of leaders say their organizations are good at shaping how humans and AI interact. And this is the thing that I think is going to become its own cottage industry, if you will. because I I think we've barely even scratched the surface on how this changes the the whole ball game. not only from the the perspective of our processes and workflows, but even from the perspective of organizational culture and trust and leadership and all those things that come along with it. I think it's it's gonna change kind of the foundation underneath everything that we're doing. this research also Deloitte it suggests here that organizations that are getting the most from their AI spending are not simply just buying technology, not taking that tech approach, but they're actually actively redesigning roles, workflows, decisions, and the relationships. That are around it. so I I I want to explore that question in this episode a little bit. Is there a difference between an AI adoption and really an AI transformation? Is this semantics? Am I using just consultant speak here and and making a distinction between these two? Or is there there really a significant difference between these two things? Let me give you my explanation for it. just in reading through their research and and my own kind of approach to it. I I would define AI adoption as merely adding the technology and the tools. giving your team access to s to a certain set of technologies or tools. That would be more of an AI adoption. Whereas an AI transformation is the full-on reimagining of your process, with AI as a key component of it. But still reimagining everything, looking at every part of it to say, does it still apply? Is it still valid? Do we still need to do things in that way? Or is now there a new paradigm that we need to fully reexamine here? an AI adoption is going to ask things like how how can AI perform this existing process? Or can it do this thing that we currently do? So it's it's almost even more task-based as can it do this thing that we do currently? And the the transformation kind of approach to it, an AI transformation, is why does this process exist? what's the goal of this process? what should it become now that AI is available? Because just automating Part of a broken process, that's not gonna fix the process, right? That's just going to automate something that's already broken in the first place. so there's an interesting kind of use case that they they bring up here that I wanna kind of go through with you as well, that that kind of really highlights the difference in these two approaches. So they they refer back to a McKenzie consulting page. research that that came out in twenty twenty four about a European telecommunications company. They don't name the company, they keep it anonymous, but it's they just refer to it as a large communications, European telecommunications company. And what they they said was this this company initially added AI expertise to an existing customer service workflow. So they took that AI adoption approach of saying Here's the task we currently do. Which task can we replace with AI? And here's the kind of interesting result of it. That they found that they only received a 5% productivity improvement when they did that. 5%. Right? So we we see these amazing stats that people are putting out there for AI adoptions and what AI can do for places. But this one example. They only saw 5% productivity improvement when they were just looking at the task and saying, Can we replace this task? Luckily enough for us, that same company did a broader rollout later. And during this broader rollout, they took the transformation approach. They invested heavily in the redesigning of the human-AI relationship. They changed their workflows. They changed their training approach. They they looked at their escalation paths and and the points where people should trust or question AI. I've heard this referred to as sort of the trust boundary. trying to identify that trust boundary line to say, here's where we can trust AI up to this point, but beyond this point, no, we really can't trust AI. It doesn't have the expertise to make that that call. And when they redesigned using that systems approach to it, that produced the 30% productivity improvement. So five percent versus 30%. And it's the same basic category of technology here. We're not talking about two different systems or two different types of technology. It's the same kind of technology that we're implementing in both cases, but the difference was not simply a smarter model. it was it was a difference in design of the work that was around it. now now to be clear that doesn't mean that every company is going to see that difference. It's not going to see a you know, every company's not automatically going to get a 30% improvement. It is, you know, anecdotal. It is one company, but it is a a use case you can look at to to l to kind of examine the difference in approaches in that. same type of environment. And I I think it also does a really good job of illustrating the opportunity that I think a lot of companies are missing today because they don't really take that full on transformation approach. it's it's not about just training your company, your your your employees to prompt better, right? It's it's it's about addressing some more fundamental questions like what should the the human handle, what should the AI handle, when should it be checked? When should the person trust it or or when does it need to be verified? who owns the final outcome? W when an organization prioritizes work design, Deloitte actually found in their research that they're twice as likely to exceed their expected return on their AI investment. And that's huge, right? If they're so again, prioritizing the work design when they're implementing AI. Twice as likely to exceed their expected return on AI investment. People are spending a lot of money on this. And if my company is spending money, I yeah, I want that 2x button. Right? I want to take that that approach that's that's gonna make it twice as effective. And you know that that could be the difference between success and failure for a company right there. so redesigning the work or transformation, I I know as I said, that that can sound a little bit like. consultant language. So let's let's try to take a closer look at what that means at a more practical level in practice. What it means is is not asking only what task can AI perform, but asking what relationship should exist between the human and the AI. A AI can take on a lot of different roles. And I think it's important to Understand what type of role we're talking about for AI depending on the scenario. AI can play the part of being a tool. You know, a really smart calculator, if you will. It does that kind of thing really well because it handles large data, it can crunch data, it can help us figure out things faster than a human could do that. So that's just one of the flavors of AI. AI could be serving as an assistant where they they're maybe preparing a a a morning brief or a meeting brief, drafting plans for us or monitoring routine information. that kind of ongoing activity. that could be the the the role that the AI is playing. It could be playing the role of a coach. Where it's observing human performance and it's it's offering real-time guidance from what it's observing. there's a example that Deloitte gives of this in this paper. they highlight how MetLife has used AI to coach employees during emotionally difficult customer calls, which, yeah, I know none of us have those, right? but that what they found was the company reported a 13% improvement in customer satisfaction. Along with shorter calls and lower employee stress. But understand it's not the AI doing it for them. Right? This is AI as a coach. So it's allowing those employees to practice before they're put into the pressure, the pressure cooker of being on that call with a an I rate customer. Right? If they're practiced in that, then they they handle it. much more calmly and like like this shows, right? Thirteen percent improvement in customer satisfaction by doing that. AI could be a a decision advisor for us. that's a role it can play. It can analyze a a large amount of information. It can recommend courses of action. the human re remains responsible for that decision in that scenario. But AI can be sort of that trusted advisor to help us make the decisions. that's another role the AI might play. And then the big one that I think is becoming more and more prevalent is, you know, agentic AI, AI as an agent of some kind. When when AI receives some kind of a a larger goal, an objective. And the AI then is actually determining steps, it's it's using specific tools and skills, it's performing work with you know some limited version of of of supervision and probably less and less as it grows and gets better that could be something that the ai is doing so design questions in that scenario become a lot more serious about how we're working what authority does the agent have what what systems can it access what what What does it have the power to approve? w when must it be stopped? Right? When when when must there be a gate, a stopping point to to make sure that it's not just spinning off and doing something endlessly, right? What happens when the situation falls outside of its instructions? that's that's something important to consider. When when someone says then we use AI, that that really tells us almost nothing. And it it it's you know, we have to come back and say, are they using it as a calculator, as a coach, advisor, agent? those are completely different relationships. And if we're in the the the mode of trying to look at our systems of work, then it's really important to understand how that tool is being applied and how it's needed and how it's being used. So here's a here's a practical framework that you can use. There's four things that you need to think about when it comes to this kind of decision. One is human responsibility. What must remain human? Because it requires empathy, it requires judgment, accountability, values, or or even relationships, right? AI is not going to have a relationship. so sometimes there's a need for the human because it requires a relationship. The second thing is the AI responsibility in this scenario. What can AI do because it benefits from the speed, the scale, pattern recognition, repetition, or or even constant availability that an AI would provide. The third thing is shared responsibility. Human AI shared responsibility. Where do the human and AI need to work together? Where does there need to be overlap between these two where they actually do collaborate? And then the fourth one is escalation. What conditions require control to move back to a person? As these agentic systems get better and better, we're going to be trusting them with more and more. But it's important to keep our our eyes on that gate to say when it trips through this, this is the point at which a human needs to get involved. The the future of work is not going to be designed designed by deciding whether humans or AI win that scenario. It's going to be designed by deciding who does what, who decides what, and who remains accountable when something goes wrong. That's going to be an important question. And that last one who who decides and who is accountable? that that may be where organizations today are are least prepared. So that leads to kind of another point here. There's a stat that came out from this as well that says 60% of executives regularly use AI to support their decision process. 60% of executives. But only 5% say their organizations manage AI-supported decision making well. Meaning that they have systems and structures in place to assist to properly take AI into account when making decisions. 60% of them are doing it, right? But only 5% say their their organization is actually equipped to have that be effective. And that's it. That's an important thing for us to kind of take hold of. Six out of ten executives are regularly using AI to help make vital important business decisions. But only one organization in 20 believes it it manages that process well. And I I think that's really indicative of the gap. That's kind of what I'm trying to point out. my finger at is that the the speed of the technology is going so fast that it's outpacing our process ability to keep up. And I think we're having this this kind of gap that's that that's developing. human review by the way does not automatically create accountability. It it it needs to be really looked at carefully because a person may simply just approve whatever the AI recommends, right? They could be rubber stamping something. And we're humans, so we have this psychological condition where we ascribe trust to something like AI. And if it's gotten it right forty nine out of fifty times, then yeah, the the fiftieth time we're more likely to trust it. So we might rubber stamp it that fiftieth time. Right? Talked about this in previous shows. i it's about the system. It's about the approval box at the end of the automated process. And it it it it just putting the human there does not necessarily create human human control. We we have to redesign not merely rearrange task. so we have to think about things like decision rights, authority, accountability, feedback loops. And like I said before, escalation points. That's that's the important one there. that gap that I think is there, I think Deloitte is actually trying to coin a term for it, which I I I really like. They introduced this idea of culture debt, kind of like tech debt, right? Tech debt builds when teams take shortcuts in software and they they leave future problems behind that we have to come and clean up. I always use the example in class of saying that if I have a bakery and we get slammed with an order, you know, we're scrambling to get that order out of the bakery. We're baking, you know, I don't know, twenty, thirty dozen cupcakes trying to get it out the door. Well, at the end of that, what does the kitchen look like? It's messy. Right? We've got to put things back, we've got to clean up. And that's kind of the idea of tech debt and software. But culture debt is what they're trying to refer to here as culture debt it is a similar kind of analogy. Because what they're saying is organizations introduce AI so quickly today that they they tend to postpone hard questions about trust, about fairness, about credit, expectations, and even sometimes acceptable use. And those unresolved kind of questions, those are gonna accumulate. so this is this is what Deloitte's kind of talking about here as as being more of a culture debt, right? they they found here that forty pr forty two percent of the w of workers say organizations rarely evaluate AI's effect on people. And here's another big area, right? So now we're getting into psychology, sociology, but these are some big questions that have to be answered for an organization's culture. things like is using AI considered smart or lazy here? That's a big one, right? Am I looked at as being ahead of the curve if I'm using AI, or do I do I am I seen here as being he's a lazy person because he he uses AI. Does an employee have to disclose when they've used AI? Is it important or does it matter? do people care? maybe even who gets credit for using it. Does does AI use help or hurt someone's career? Kind of back to the question about does it make me look smart or lazy? That's a culture question, right? That's for the organization because I I can very easily see how popping in and out of different organizations as I do as a consultant, there's different environments. In some places they are seen as, hey, you're on the head of the the the curve here. you're on the bleeding edge and we like that. We want you to push us and get us, you know, moving into the the future. Another place is people see it as a crutch. And yeah, if you you need to use AI to do that, well, you couldn't just do it by yourself. That's a culture question. another one, what what happens to collaboration when more work occurs between one person and an AI? And that's kind of really the heart of something I've been trying to dive into is What what are those team dynamics? How does those team dynamics actually shift and change? here's a they brought up another interesting point that I thought was really I I had not heard this before, I had not really thought about this too much before, but we we always talk about how AI is going to remove the routine work. But routine work does have benefits. it it provides skill practice, it it gives entry level benefits. experience. It is easier to to kind of see progress. So you you feel a sense of progress when you're doing more routine work. Maybe even some social interaction or even dare I say mental recovery a little bit. If we're only doing the hard things all the time because AI is taking all the routine things, we We might be losing out on some of the benefits there of doing some of those routine things. Which, you know, we keep saying that AI is gonna remove the boring work, but but a job that's made up entirely of hard problems, that's not necessarily a better job. we we're human, we may need to get some of those benefits of doing some more routine things on occasion as well. it it it it could make it not a better job, it could make it just a more exhausting one, right? and I think that's something that we're gonna have to to wrestle with culturally in our organizations. So work work redesign here, when we talk about doing a an AI transformation, it cannot mean designing only for speed and cost. It's gotta account for learning. It's gotta account for meaning, for judgment, for trust. for sustainable performance not just burst because otherwise the organization may gain efficiency but it could be quietly kind of damaging that system that it that it produces that produces the work so he I want to leave you with this kind of seven questions to ask yourself before you add ai to doing something one is what outcome are we actually trying to improve Right? What's what's the the intended outcome from this thing that we're doing? The second one is why is the work currently organized this way? Is it organized this way because we had a scarcity before of skill? Is it organized this way because we had a scarcity in the people that did this work? Maybe that's no longer the case if we're using AI to do that thing. the third one is what parts exist because of limitations AI has now removed for us? were there bottlenecks that we were trying to get our way around? That now there's no longer a bottleneck there because AI doesn't have the same bottlenecks. The fourth one is what should AI do and what must remain human? For the reasons we talked about earlier in the show, but things like relationships and other things? What are the the things judgment calls? What are the things that must stay human? must re what pieces of work must remain with a human? The fifth one is who has the right to make each decision? not only human or AI, but which human? The sixth one is when must AI stop and escal escalate to a person? I think this is going to be a really important part of how we redesign our systems is those escalation paths. When does it flag something? When does it trigger something? When does a human need to see this? Because if we do that on everything, then we're going to lose the benefit of the speed. But if we don't do it enough, then we're going to get AI slob. So we need to have some balance between that. And the seventh one was. How will we know the work became better, not merely faster? And I think that's a really important one to consider, right? How will you know the work is better, not just faster? And if you ask yours yourself those questions, I think that you can start to wrap your head around are we really transforming this? Is is this a transformation with AI, or is it just adopting certain tools and skills into our organization? So it's a really interesting study and research paper that I'll I'll make available to you here on the website at agility evolve.comslash podcast. That's where you can go for our show notes and anything else here from our show. if if you like this episode and this topic is interesting to you, then a couple of things I'd ask from you. One is like and subscribe to this, that really helps us get found. We're still a a very new podcast and trying to grow our our audience here. So if this is something that you appreciate, then please, you know, tell a friend about us. That's really the only way that we grow. And then also email us. If you there's something that you liked or don't like about this, if there's someone you want me to have on as a guest or a topic you want to hear me talk about here, if you you want to ask me a question and want me to dive into some research on something I'm more than happy to do that. this is this is a labor of love for me. yeah, I I'm not I'm not making a dime off of this. so this is just something I'm doing because this topic really, really interests me. And you know, I I I want to talk about the things that you guys want to hear. So if there's a topic you want me to go into, please, please email me. You can email me at Brian, and that's with an I, B-R-I-A-N, at agilityvolved.com Again, that's Brian at agilityvolved.com. And I I will respond to every single email that you send me. so I really appreciate any any feedback that you give me there at our email address. And that'll wrap us up for this week. So appreciate you joining us and we'll talk to you next time on another episode of People Over Prompts Podcast.