AI Adoption Is Up. So Why Is Work Getting Harder?

[music] You're in the AI Leadership Studio, an ISMG podcast hosted by Dan Burton, Vice President of Content Intelligence and AI Innovation. Now, here's Dan. [music] Here's a number worth sitting with. 13 minutes. That's the average length of a focused, uninterrupted work session inside the American Enterprise right now. 13 minutes. Down 9% from two years ago. And it's still falling. Now here's the number most leaders in this audience are actually looking at. 80%. That's AI adoption inside their organizations. Up 52% in two years, with monthly retention averaging 92% by every conventional measure it's working. The story looks good until you look at what's underneath it. Because the same data shows that after AI adoption, time spent in email went up 104%. Chat and messaging up 145%. Every single work category measured, every one increased after AI adoption. Our guest today helped build that data set. Gabriella Mauch is the Chief Customer Officer and Head of the Active Track Productivity Lab. The research operation that just published what may be the most consequential workforce data set of 2026. Not because the headline numbers are surprising, but because the numbers underneath them are. She oversees one of the largest behavioral studies of how AI is actually changing work inside real organizations. Not what leaders say is happening, but what 163,000 employees across more than 1,100 companies are actually doing with their time. And that distinction matters. Gabriella, welcome to the AI Leadership Studio. And thanks so much for having me. So let's start with the data itself in this report, because I want to make sure we're talking about the same report that I read. Not the sort of summarized version in the press release, the one that took three years to build. The headline from your report is that AI adoption hit 80% and work days are getting shorter. That's the version that's going to travel, but you spent three years building this data set. What's the finding in there that you think is most likely to get Mr overlooked? And you know, what does that particular finding really matter? Yeah, I mean the headline numbers are genuinely good to your point, right? Burnout is down, work days are shorter, productivity hours are up. But the piece that I think leaders really need to pay attention to, you know, the data point that doesn't feel as validating to them as the ones that I just said are the data points around focus, focus efficiency and focus blocks. You know what we're seeing in this data is that focus sessions are getting shorter and that's actually happening at a faster rate for our AI adopters versus their the AI non adopters just yet. Or the laggers in terms of that adoption. And so it's something that I think really needs to be grappled with by leaders and by organizations because it's not necessarily bad, but it is a flag and concerning. Should we see that that most prize level of work that we've spent years trying to further cultivate is beginning to happen in a it is beginning to shorten over time. So your data shows something that cuts directly against the core assumption that I think most leaders made when they greenlit AI adoption or AI programs that it would reduce workload instead. After AI adoption every single work category you measured seemed to go up so email was up 104% chat was up 145% nothing's really decreased when you saw that pattern emerge what did it tell you about how organizations were actually deploying AI versus how they thought they were deploying it. What it tells us is that AI layers on top of work it doesn't substitute it right to your point absolutely nothing decreased and what that says is that our high performing employees are the ones that are adopting AI and they're using it to do more not to do the same amount more efficiently. And I think one of the common things that you'll hear for me as we have this conversation is that I'm really not in a position to tell you what's good or bad just yet, but I other than the fact that organizations inability to understand some of these pieces right now means that they are also growing their the the they're also building upon their knowledge gap of what that impact is of the AI solution or the AI solutions that they're deploying. Right and so you know when you go ahead and you deploy AI if you're doing that without a strategic conversation of are we deploying this so that they can do more or we have deploying this so we can free up time that lack of strategy at the onset of the purchase and of the deployment really leads to sort of this this leak if you will of drifting work that comes out of this. So let's talk a little bit about what this means for leaders in the audience you know those folks who are actually making decisions about their AI programs and you know they're listening because they have to make these decisions your data identifies what you call the AI measurement gap the distance between knowing your adoption rate and knowing what AI is actually doing to your workforce so. For a let's say a VP of AI or chief AI officer sitting inside a mid size enterprise right now what decision does that gap force them to make that most of them are currently trying to avoid yeah the decision that we see leaders avoiding time and time again is the impact metric right and so current dashboards today will show logins but they won't show outcomes and we. And we celebrate adoption but we can inflate those data points really really simply right you did training last Thursday on an AI tool well snapshot that adoption on Friday and it's skyrocketed right but adoption is not the same as impact and when we make a purchase for any sort of AI solution we did that not with adoption in mind but with impact in mind and so I think the piece that we. Strongly support leaders and we we support not only through our product but also in the guidance that our productivity lab gives to our customers is how do you shift away from the adoption conversation towards the impact conversation once you have that adoption in place and so. 95% of organizations have adopted AI in some form but the ones that pull ahead will be the ones that ultimately can measure that impact piece you know I guess also something I was thinking about while you were answering that question is the feedback loop is so important to the whole process because I've gone through exactly what you just outlined which is you know measuring logins. And I so I know x number of people are using x number of tools but without the feedback I really don't know what the impact is right and i'm going to call it what it is vendors love to tell you login information right but login information isn't impact information and so it's just it's a. It's not an important metric because if you don't get people to change their behavior and adopt something you never get to the impact piece but I want us to think about it more as a linear progression and what I see is that leaders get stuck at stage one it's still a stage it's still important and you want to drive people into changing their behavior and trying a new thing but if you don't push to the next question which is again what is the impact and how can we measure it. Then you're not maximizing the impact and you're not ensuring that you're giving people the appropriate training and enablement to get to that impact metric that's going to drive the difference in your business. So maybe you can offer a few examples of you know what it looks like when impact is not being measured. Sure so if you take an AI solution where the intent was to drastically cut down the time it took to process a count executive sales notes right you went ahead you purchase a solution and it's going to consolidate the transcript they had and spit out a summary. If you've never initially measured the amount of time a sales rep spends on processing their sales notes then you have very limited insight into the impact of the purchase that ultimately is now completing those sales notes and so this becomes a very classic conversation of pre post a b testing pre post testing baseline testing but. But it's a good example of, hey, my goal was to stop my account executives from spending four hours a day typing notes. So, I'm going to snapshot where I am pre-purchase on time spent in their CRM. And I'm going to snapshot the time spent in their CRM after the fact and ask myself, if it's the same amount of time, are they at least not on the notes page and are they now prospecting or doing something different? That's a perfect example of really basic type of impact measurement that with any sort of work intelligence type solution, you'd be able to capture that just becomes critical to these types of insights. So your data shows that only 3% of employees are in the, I guess, what it calls the usage range, 7 to 10% of total work hours in AI that produce measurable productivity gains. 27% are spending less than 1% of their time in AI tools. You have 80% adoption and 92% monthly retention. So, by every adoption metric, this is working, but by the impact metric that actually appear, that actually matters, 97% of users are outside the zone that produces results. So I guess to your point, that's not an adoption problem, that's a governance failure or governance challenge. Who owns that? And why aren't they being held accountable for it in most enterprises yet? It almost sounds like a trick question because my answer is no one owns it. And that's a little bit of a problem that's that we're facing right now. And again, I think that really accepting, I listen to a podcast myself a few weeks ago, that really stuck with me. And what it said was in this age of AI, humility is just one of the most important qualities you can possess because you don't really know what's happening right around that corner. And so my answer in that no one owns it isn't to infer that nobody cares or we're not designing organizations right away. It's that we're figuring it out. And one of the things that we haven't figured out is where the ownership of AI effectiveness actually sits. We've got function leaders that are using their own budget, where if it comes in under a threshold, they're buying an AI tool. You have a CIO who just came back from a conference and she's chosen to buy an AI tool. Then you've got a CEO that just met with their friend on Sunday morning and they're off proposing a new AI tool. And it's a very dispersed set of strategy right now. And that's the stage of maturity we're in. But eventually organizations will have to decide is it a C-O-O that owns AI effectiveness? Is it a C-T-O? I think I don't know the answer. My preference has always been go with the person who owns the outcome of the solution, not the person who owns the deployment. But that varies from company to company in terms of scope of work between a C-O-O and a C-T-O or a C-I-O rather, but the answer to your question, Dan, is no one owns it right now and that's part of why we're seeing this problem. This next data point is I find it truly fascinating. Your report flags disengagement risk as the successor crisis to burnout. Up 23% now affecting nearly one in four employees. These are people whose capacity was freed up by AI and that is now drifting. So what does that cost an organization that doesn't catch it in the specific terms maybe a CFO or a C-O-O would recognize? For a C-O-O that's unused workforce capacity that you're already paying for. And that matters a lot to a C-O-O. It also matters to a CFO. For a CFO, it's productivity loss compounding and compounding quietly. It's not showing up in a single quarter. It's not showing up in attrition data. When someone's burning out, you see people talk with their feet. They walk out the door. They're getting another role where the expectations are and as high or they're more balanced. It erodes your competitive output over time and that's something that sticks with the CFO. The burnout crisis was visible for the reasons that I just mentioned but this one isn't and it makes it harder to act on and it's more expensive to ignore. So let's get a little practical at this level. So the leaders that are going to be listening to this episode are going to walk away from the episode and right into a meeting and I want them to hopefully leave with something that they can use. You identified a productivity sweet spot in your report, 7 to 10% of total work hours in AI tools. 3% of users are there. The other 97% aren't. So if you were advising the executive team at a mid-size enterprise, not a Fortune 500 with a dedicated AI governance office obviously. What's the first sort of concrete step toward closing that gap? The very first step is to stop fixating on adoption and start focusing the conversation on intensity and impact and you'll notice that I didn't go straight to measure it. Measurement takes work. It takes time. It takes the appropriate solutions in place. It takes the visibility and the telemetry elements that are required to do that. Just start shifting the conversation. When someone sits down at a meeting and says, I can't tell you enough good things about tool X where at 90% adoption say that's really exciting. What's the impact we expect to see? Start asking the people around the table questions that shift. It's not minimizing the adoption, but it's moving you on to the next stage to start having the conversation about impact. Build visibility into where employees fall on the usage distribution. Understand is that an adoption metric that is purely a function of them logging in once the day after training? Or is that an adoption metric because they are consistently using it? One of the metrics that we have here at ActiveTrack that I believe really changes this conversation is the ability to understand, did you leverage an AI solution once or did you leverage it on a non-going basis? Was it episodic or did it become a routine? That begins to let you know just the quality of that adoption to help informing the impact. The last piece that I would say that I think is just so important is to feel as if you've made all these purchases and you haven't set up a way to measure impact. Then I did it wrong. It's absolutely not the takeaway from this conversation. The takeaway is that if you look at the numbers, which is that you have a huge mass of people that have adopted it and haven't fully gotten the impact or the value. What that means is you still have the ability to measure something similar to a pre-post. You have the ability to say, "What are those 3% of the people doing? What does their day look like? How do those habits look? How can I compare it to the other 97% of the people in that work?" Even though you might not have the clean before the purchase and after the purchase measurement, you do have a clean, "Here are my optimal workers that are leveraging the solution and improving their results and here are my suboptimal workers who are not." How can I change the mix shift so that there are more people over here in the category of improving their work and doing it more efficiently? Let's talk now. We touched on this earlier, which is the focus efficiency issue, which is at a three-year low according to your data. The average focus session is 13 minutes and apparently still declining. Your data shows AI users are experiencing more focused degradation than non-users, which really is a surprise. Maybe not so much of a surprise if you think about there's this argument that we're no longer exercising our ability to think because the AI is providing so much of the data to us. If a leader accepts that this is a systems failure rather than an employee discipline problem, what does fixing this system actually require? What has to change that most organizations haven't touched yet? I love this question and I can't help but speak to my own personal experience to emphasize a lot of what you're including in your question, Dan, which is the leaders who get these types of challenges or hurdles in the way of the workforce, the ones that actually get them right are the ones that do exactly what you just said, which is they see it not as an individual discipline problem. That is a laziness problem, not as a, you know, for lack of trying, but it's a design problem. The environment is designed to pull away from their focus and bring them into the next activity, whether that means 19 different Slack channels and a bouncing icon on an ongoing basis or a ticking phone or whatever that is, that is a function of the way that you've designed the workplace. And so really to get this right is to have the harder discussion, which is what is our meeting culture, what are our notification norms, how do we give our workforce protected time. How do we drive it? expectations around protected time. How do we communicate better asynchronously? Those conversations, actually the conversations the easiest part for anybody who's tried to really change this in their organization. Those conversations followed by the discomfort of changing the way you come to work every day and the expectations you drive are what ultimately helps improve focus. And so it is not easy. It requires a whole lot of consistency and it requires keeping people accountable to really driving something different as it pertains to how the organization works together in an ongoing basis, not one leader, but the entire leadership team across that organization. - Yeah, this reminds me of a conversation I had a couple of episodes ago with a security professional, Jeff Williams. He, we actually talked about how often AI will miss a lot of data when users are asking questions about corporate data. And his comment was that AI is very lazy. AI just wants to look good and will take shortcuts to get you the answers you need as opposed to, we mentioned is this a lazy employee problem? Well, in a lot of cases it's a lazy AI problem. - It's funny you said that because I, I've been thinking about this a lot in the past week, which is I would just kill for a typo. And I would have never said that six months ago, but I would really appreciate a good old typo that told me you were so excited to get your idea out that you typed it wrong, which is actually more work and more passion than what AI is going to spit back at me. And so I couldn't agree more and anyone who's ever worked with me will know it's just, it's as if I, you know, there's, I've become a different person because I would never say that I would live for a typo, but gosh, I could use a good typo and some real thought behind some of the work that versus some of the work that AI is pushing out right now. - Yeah, you know, a lot of people are purposely putting typos into their social post to avoid people criticizing their use of AI. - How unfortunate, how unfortunate. Last question, you've been watching obviously this data accumulate for several years. You know what the leading indicators look like before they show up in the headlines. 90 days from now, 18 months from now, what's the thing this data is pointing toward that most enterprise leaders aren't yet watching, not where they hope AI takes them, right? Where the behavioral data says it's actually heading. - The most top of my one for me is that disengagement curve. That's accelerating. That could hit more than one in four employees in your workforce. And if you take the same principle that we just laid out as it pertained to focus, I take the same one on disengagement. This is not necessarily a for lack of effort. It's a, I don't know how else to spend my time because some of this work has been automated or has been made easier based on the tools that are now at my disposal. I think that if organizations aren't having conversations and upskilling their managers to think about or design the right way, then this is gonna continue to be a compounding problem. - Gabriella Mouk is the chief customer officer and head of the Active Track Productivity Lab. Gabriella, thanks so much for spending some time with us today in the AI Leadership Studio. - Thank you so much, Dan. (upbeat music) [BLANK_AUDIO]