Inside the 9 Future of Work Trends for 2025

(upbeat music) Are you trying to navigate the hype of AI? Now is the time to capitalize. Investment in AI has reached a new high with three quarters of CEOs already personally using AI tools. Gartner is here to help, providing executive leaders with expert insights, strategic guidance, and actionable advice to make informed decisions and stay ahead of industry trends. Visit Gartner.com or click the link in the show notes to download our AI Action Plan and learn how to build right AI strategy for your organization. (upbeat music) Welcome to Gartner ThinkCast. I'm Alexis Waringa. Today we're discussing the top trends shaping the future of work from navigating expertise gaps to tackling loneliness as a business risk. Each year, Gartner's future of work research pinpoints the most urgent developments shaping how organizations operate, innovate, and grow. We're joined by Gartner's Senior Director Analyst, Emily Rose McRae, to help break down these trends, how they're already playing out, and what leaders can do to respond. Emily Rose, welcome in. Thanks so much for having me, Alexis. Yeah, excited to talk with you today. Yeah, this is gonna be fun. Yes, exactly. Let's start with a big picture. When we're talking about the future of work, what exactly do we mean? And what's different this year compared to past years? So technically speaking, we talk about our definition. We say it is changes to who does work, or what does work, which is definitely coming up a lot lately. Where work is done, when or how much work is done in terms of schedule and hours, and how work is done of course, but also even what we even consider work. So how is work itself as a concept changing? More broadly, what are the different social, technological, and business decision drivers that are changing the way work is being done, and what it means to do work? Yeah, and I think we're definitely feeling that disruption seems to be a constant across the board this year. Are there trends in particular that you're seeing be accelerated by, or maybe even contributing to the overall sense of disruption that we're feeling? Sure, so one of the big things is definitely AI. I mean, AI has been a major part of our trends for the last two years. It was not a part of our 2023 trends, and we regretted that choice. We, I made a mistake there, but this year, AI is presence in the feeling where it's really shifted from the, in the Gertner hype cycle, what we call the peak of inflated expectations down into the trough of disillusionment. We're headed to that trough. People are getting really frustrated because they're not seeing the returns they expected, and the hype was so big when gender to the AI was being initially rolled out. And even now, we're still seeing really high expectations in some cases being shared that aren't necessarily playing out in reality, and that's creating a lot of challenges and a lot of pressure to somehow innovate. And then we also have a workforce that is simultaneously bringing in more and more of Gen Z at the same time that we also have a lot of folks reaching retirement age, whether or not they choose to retire, they're definitely getting closer to that age, and that creates new sets of risks around knowledge sharing and knowledge transfer. Yeah, it's interesting to think about how it plays out across those different generations of people entering or preparing to exit the workforce. I'm curious in your conversations with leaders about this, about disruption, what are they seeing kind of from their point of view for themselves and for their teams? So when I talk with folks in the C-suite about this in particular, there's a lot of feeling, last year I would have said it was, how do we get ahead of what's coming? And this year's a sensation of perhaps being behind. And that's an interesting and challenging space to be, but I think especially with all of the reactions that folks are having to changes that might be happening in the broader sort of political realm. There's quite a lot of disruption happening with new elections and the results of those that are creating a level of uncertainty and stress and desire to at least get everything else right because we know we can't control that bit. Yeah, yeah, there's a lot of moving parts to manage, a lot of differing views, probably within an organization, people with different things that they're concerned about. I'm curious, what else is like standing out to them? What are the things that are like keeping them up at night? The big thing that people bring man for is, we invested a ton in AI and bought some sort of license for an enterprise-wide tool that everyone has access to and we are not seeing returns. So now what do we do? It's either senior leadership bringing me in just to talk about it from a strategic broader perspective or sometimes what happens is that IT has recommended training as a solution and what I end up talking to, if I'm talking to CHROs or to people officers, what does that mean in terms of not just training, but actually really thinking about enablement and also working to reset expectations around productivity because in many cases providing access to a generic tool just is never actually going to lead to the results that people expect, despite what they're hoping. Yes, yes, it's part of that implementation and everything as well. You know, somewhat related, you mentioned the increasing expertise gap as the first trend in the research. So how organizations responding to this theme of talent scarcity? It's interesting with the changing economic climate, in particular, there's sometimes a bit of a wish that folks could return to the world where there was enough unemployment that employers felt quite confident that talent was going to need to accept what was offered. And the reality is that since the early days of 2020, that calculus has changed because one of the things that came out of the COVID-19 pandemic and that first year of it is people changed their expectations for the role that work has in their lives and they're not just going to say, yeah, okay, I'll take it. So that talent shortage has not gone away even as unemployment has shifted. Instead, organizations are at the same time as it being challenging to get critical skills. And they're seeing high turnover, particularly in some of the higher volume roles that include skilled talent. We are also facing the reality that some of our most skilled workers are reaching retirement age and certainly have the right and ability to retire if they are so inclined. How do you prepare for that? How do you address? And then we add in another layer that's creating like a longer term issue. So organizations have known that the wave of retirements, but some folks call the silver tsunami was coming. That was a known demographic reality, but a lot of folks did not address it. I'm kind of hoping that it would magically resolve itself before it became a crisis. We have another level to this crisis coming, which is that the AI tools that we're investing in, the use cases that are producing the most substantive value at the moment tend to often be in spaces where you're taking work that used to be a training ground for entry-level talent. So you're taking big parts of an entry-level role that were used to skill someone up to do more advanced work and automating parts of them. Well, that entry-level talent can't be doing the validation because they don't have the expertise. So maybe they can do the prompt engineering or some different activities, but they can't actually do any of the validation. So there's no training ground for them anymore, which means at the same time that we are losing as our most expert talent, we're also losing our potential pipeline of expert talent. And for some organizations, the solution is gonna be a pretty straightforward. Okay, we're just gonna hire less entry-level talent, but we plan to poach mid-level talent aggressively, and we're gonna invest a lot in having very competitive salaries and benefits and offerings there. But for other organizations, that's not gonna be on the table as an option. So then the question becomes, what do we do when we have such a growing expertise gap that's gonna be not just a demographic phenomenon, but really embedded into what we're trying to deal with here? Yeah, it's recognizing the importance of developing people to the point where they can step into those roles that are going to be opening up with the retirement surge up ahead. So as you noted, AI is a big part of this. With the continued push for AI integration, what are the risks of pushing too quickly? And is there a misconception from leaders about how employees feel about this, how their responding is there a mismatch of expectations? There really is. So one of the things that's kind of interesting, and you've seen this in some of the Duolingo and Amazon announcements is employees aren't using AI enough, and we think it's 'cause they're choosing not to. So like if all else was the same, they would use AI. That doesn't really reflect reality, because what the challenge is is that a lot of the use cases that are available for those generic, non-specific AI tools are not necessarily that high value for folks. The average experienced employee doesn't need help writing emails. That's not a scalable impact. It could be super useful for them to be able to get summaries of long documents or summarize patterns in a bank of text that they're able to put in, whether that's research or past writings or something similar, but the challenge is that, in many cases, what people are being given or having highlighted for them are very generic use cases. And in many cases, in addition to the use cases, not being clearly high value, to even use the tool people have to go outside of their workflows. And one of the things that can be a real mental block for folks in heavy adoption, and in fact what we see usually is when these tools release, people are excited they're willing to use them. This is supposed to change the way I do my job, great. We'll see, close to 90% adoption of a tool, efforts released or after there's a training that comes out, but then it plummets to like 20, 30%. It's not because people didn't want to use it. It's because they sat down to use it and some of them discovered it was actually kind of hard. I don't know how much you've tried to interact with some of those more generic tools like it. Internal GPT, that kind of thing. It requires a fair bit of iteration. And that's not set up in the expectations, because if what you hear is this is a natural language tool, natural language processing is a category of AI. It doesn't mean that it's actually intuitive or natural to use. And that's a big difference, which means that people get really frustrated when they sit down to use it and it doesn't really work. So they have to practice. But we haven't given them any time to practice. In fact, we have some best practice case studies about clients who have created dedicated time to practice. But a lot of us are just expecting employees to start picking it up in the middle of their workflows. I don't know of any client I've talked to who said, "Oh, yeah, we've actually got a workforce that has plenty of time on their hands. They can go play with this. It won't be an issue." People are trying to do the work you already want them to do. And you want them to find time a couple hours a day to experiment with AI. That doesn't seem super realistic. So if you wanted that to happen, you would need to provide higher value use cases or simply don't make it optional. What we're really seeing here, and the reason these AI first organizations are struggling or in many of them, such as we sell this with Clarnand IBM, are actually hiring back a lot of the people that they originally let go of, is it's not that we don't need people, is that we need them in different roles. So instead of saying, here's a generic tool somehow, it's a little bit like that meme, where it's like step one, step two, question mark, step three. It's step one, by a GNI tool, step two, question mark, step three, cut your workforce. Success, right, massive productivity all of a sudden, except that's not what happens. And one of the reasons it doesn't happen is that if you think about like the reality here, Microsoft's own data shows that access to co-pilot saves an average of 14 minutes a day. That was nice, but that's not world changing. That's not at a scale where suddenly cutting tons of headcount. If you want to get scale for many investments you're making in GNI, you have to actually be changing roles and workflows, where AI is not optional, it's part of how the work is done. It's part of the workflow. There are dedicated roles for prompt engineering, there are dedicated roles for validation and review and editing if you aren't doing those things. And also once agentic becomes more widely adopted, dedicated roles around decision reviews and was this decision made correctly and depending on the level of independence we've given it, do we agree or disagree? Do we need to override? If you aren't making changes to roles and workflows, there's just no way to see that level of productivity. Your employees might actually really enjoy using the tool. They might find it super useful for certain tasks that were less enjoyable. And if you source use cases from groups, you may find some great use cases that people are enthusiastic about using. But you're not going to see massive returns on investment or productivity there. And then there's another side of this, which is what happens if you are so enthusiastic about AI that you adopt it heavily and you don't necessarily leave space for feedback. So the worst case version of this is an example from a hospital system in California where they implemented several different forms of AI. So like an ambient scribe, that was taking notes, a couple other things that were making recommendations for patient care, a couple other tools. And their nursing staff started to raise the alarm and say, I have concerns. We are concerned about not our jobs because if there's any job that is secure now and in the future it is nursing for which there is severe shortage globally. But we're concerned about actually the impact on patient care. Okay, there was no mechanism for them to provide feedback that was being listened to. They raised their concerns, they raised their concerns, crickets, no reaction, and no one really with the authority to respond or an assignment to listen to that feedback. They ended up staging a very public protest outside of the hospitals. - Oh wow. That's not ideal. - Yeah. - That is a worst case scenario because now they're talking to the press about how the AI decisions with this hospital system were in their view, harming patient care. Now it might not be something as critical of an output as patient care, which is literally in many cases with life or death. We're also talking about things that might just be, I think this exposes us to legal risk or creates a negative client experience or our candidates aren't going to enjoy this. I've seen some proposals for a fully automated recruiting process with various AI agents and bots. That sounds great except that usually candidates want to talk to someone who actually works at the organization at some point before they accept a job. I don't know how successfully are going to be if you eliminate that opportunity. That seems like it might be an issue from a conversion standpoint. - Yes. If we don't create spaces and mechanisms, not just like an intake form or something we're collecting, but actually actively seeking out employee feedback on our AI use cases and what we're trying to do. And also how it's actually being applied. We're not just risking lower productivity or not seeing a return on investment. We're also actually risking major significant moral implications. Another example I have is an organization that did civil engineering where the architects for bridges did not want to use the AI tool that the company wanted them to use. And when the call was initially sent to me, it was how do we get our employees to use AI? It's like, okay, well, let's talk about motivation, ease of use, skill development, what can we do here? But then as the client began to explain to me that these architects are actually legally responsible for anything that might happen on the bridges they build, it's like, oh, no, no, no, no, don't push them to do this, actually. - Yeah. - Because the tool may make mistakes that a human would never make and that they would never think to look for. And it's actually a pretty good line to create of no, I'm not comfortable with that. If they are held personally and legally and financially responsible, but also just from a moral standpoint, yeah, actually, maybe this is not a space to use AI. - Hi, I'm Mary Masalio, a Vice President Distinguished Analyst at Gardner. I've liked several sessions at many Gardner conferences through the years. And if there's one thing I can say about them, is that attending them is the best way to stay ahead of the competition. Each conference I've been to provides attendees with invaluable insights and ideas. And the content is always relevant and tailored to key issues being faced by leaders in every four business, be it finance, HR, sales, IT supply chain or marketing. Join me at a Gardner conference to learn about emerging trends and gain new perspectives that you won't be able to find anywhere else. Leave our conferences ready to contribute more at work. And most importantly, be equipped with the skills to turn new ideas into function. Visit Gardner.com/conferences or the link in the description of this podcast to learn more. - Yeah, definitely. You're opening the door to like large brain risk as you mentioned. And a lot of cases where you need to make sure you have the time, the guard rails in place and to the right, ensuring your employees have the right understanding as well. But I love your point that it's about rethinking what those jobs are. It's not just cutting jobs and saying, AI can do all of this. But it's also saying, the way this job was done before doesn't necessarily mean it should be still be done that way. So there's still that need for that human touch, that human interaction, like you mentioned in recruiting, but you need to rethink exactly how you define that role, what they do, and then how AI is embedded into the processes, as you noted. So that was a really great point. And in addition to that, to follow on this from the employee side, one trend that I thought was really interesting, mentions workers embracing bots over bosses. And so I'd love to dive more into you with that about what does that mean? How was that already playing out in practice for people choosing bots over their boss? - Sure, so this is one that took me completely by surprise because if I had done this, and actually we've written about this in the past, people don't necessarily actually want an AI manager, right? But instead of asking how happy would you be with an algorithm providing performance management feedback, back, blank and getting not a very high number, last year we did a survey of about 3,500 employees globally, and we asked them to what extent they agreed or disagreed with the statement, an algorithm would provide fairer feedback than my manager. Now what I love about this is it is a real comparison. Instead of it being I'd be happy with an algorithm, which the answer might not be that high, it's how does that compare to our current state with who you have right now? Because it's not even an algorithm versus a person, it's the manager you have right now. Only 13% of employees disagreed with the idea that an algorithm would provide fairer feedback than their manager. - Wow, what? That is concerning, yes. 35% agreed and then about half were neutral. I was a little shocked. I definitely had someone rerun these numbers, but then it started to make sense because the reality is performance management is hard. Feedback is hard. Employees, especially those who do jobs that involve a lot of interaction with technology, whether it's in the field and they're using a handheld device or at their desks, they know that there's a lot of data available about their performance and at least their activity, but that doesn't mean that data is necessarily translating into what feels like fair feedback from their manager. And so people are very interested in potentially getting feedback that is algorithm driven, which is really interesting. From my perspective, it's more of a statement about how happy people are with the current state performance management from their managers than it is enthusiasm for algorithms and technology, but it does create an interesting scenario. For instance, we know that most of the large language model tools that we interact with as the public, like the chat GPT being Gemini, co-pilot. They've all been trained to make the user happy. So does that mean that they produce like a nicer performance review than we would get otherwise and we're enjoying that? That's a little, I mean, eventually that system we control for that, but that's something to keep in mind. Yeah. Or is it just that any feedback that feels like it's database is going to feel more objective? It puts us in an interesting position because then instead of asking what is the ideal state, we can say what we really want to ask here, which is the better result optimizing install. The things we care about cost, validity, performance recognition, high performance recognition, all of that, what's the best outcome for us here? What do we think it could be? Is it using a tool instead of a manager and freeing up some manager time? And do we think the tools are actually going to be accurate? Or is it investing even more in manager training and trying to help them be better? Or maybe it's embedding some AI in the tools the managers use for performance management. There's not one answer here, but it highlights an opportunity and a chance to ask some questions. Yeah, as you know, to definitely signalling something there about people saying that there's an issue there if people are thinking like this AI is going to be the better, more fair judge for them. And also, as you noted, that has its own bias in that people are beginning more positive feedback sometimes of AI agreeing with them. So if you've been in a situation where you feel like your boss is always giving you like critique or something and AI is always telling you how great you are and how smart that idea was and peeking it through with you could lead to some people choosing that over their specific manager in that situation. Tied to that. And something else I found really, really interesting reading this was employee well-being in connection and that the loneliness was called out in particular, not just as a personal challenge, but as a business risk. And I think when you first hear that, you might think, oh, it's remote work. It's people working globally. But when you step back and think about it, loneliness can really be heightened in those in-office experiences as well. So I was curious to hear more about that from you and why is this such an important shift to address? - So this is one of those that the natural response is, well, if people are lonely, let's just bring them back to the office assuming they're working remotely. And everyone will be together, it'll be great. But it turns out when we look at the data on-site and hybrid workers are not happier than fully remote workers. They do not feel more connection. They do not feel more camaraderie. So proximity is not a cure for loneliness. All right, proximity is not a cure for loneliness. What is? What sort of connections will make a difference here? And the reason it matters for employers is not just, oh, if our employees are lonely, there's greater mental health strain. They might be a little sadder. We actually see significant performance declines when people respond that they are lonely when they rate their specifically at work. So when they rate their co-worker quality and camaraderie quite lowly, their performance goes down. They are not able to be as effective as they want to be. And this is also reflected in the academic studies as well. Loneliness actually hurts people's ability to do their job. There's some interesting stuff that has just come out recently and the press around AI tools and people will coming emotionally attached to them. So I'm sure there are some organizations that are like, well, we'll just give everyone an AI companion, which has its own set of HR risks. My colleague, LJ Justice has just written about, do our fraternization policies apply to our AI chatbots that people have access? - Oh, wow, yeah. - Great question and a terrain that I did not ever imagine us having to have conversations on. But yeah, this is a real problem. And it's not one that can be as easily fixed by RTO as people might think. It actually requires real investments and helping people build connections. - Yeah, and it made me think of, when I was starting out in my career, you feel that sometimes of that high and awkwardness at lunch time when you don't quite know a lot of people yet and the few people you do know might be in meetings at that time. And so I think for me, it tied back to what we were saying about those entry level jobs being harder too of people entering the workforce are starting with evolving roles of what opportunities. And then at the same time of just because you're in office next to somebody doesn't mean you've formed a connection with them. And so it's about providing those really intentional opportunities for connection in like having collaborative meetings, not just like social, you know, hangouts and whatnot, but just making sure you're in office. You feel a sense of like, oh, it was worth that I came in here. I like connected with some people. So that's a really critical part of it. And that's more just how do you make any sort of RTO strategy whether it's hybrid or fully onsite work is the time that people are spending in the office should be different than the time they aren't spending in the office. It should be clear that there is a point and a goal and it's something we couldn't achieve separately. And often that's going to be social or connection based. But it's also, can we create social groups that are global if it's a global company that are around shared interests instead of being something else? So there's some interesting opportunities here, but it's not like a cheap and easy fix. Yes. So we've talked a lot about disruptions, the risks with AI to your brand, but from an optimistic side, is there a trend in particular that you're most excited about or closely keeping an eye on that you think like signal something positive for the future? Always live a good positive route to kind of wrap us up and end out on. So I think I'm actually very excited by some of the stuff around the employee activism because I think it helps redefine our understanding of what is responsible AI into something that's much more tangible and realistic. And it also is a mechanism for doing what we already know is best practice with any sort of decision making which is to put it as close as you can to where the decision is actually going to have an impact. And that we tend to see issues when decision making is further away from where the actual impact and business results will be seen. So I think that that is a space that I'm sure is very stressful for a lot of organizations but also has incredible potential. Yeah. No, that is like an optimistic, you know, making sure that like the people involved in it, like you even noted of making sure if they get those tools, they feel equipped to use them and that they feel empowered and that makes sense to them. If they're coming to the office, it makes sense. To tie it all together, what's the most important action that leaders should take by the next quarter for a short term action to take now and then going beyond it to the next year or even longer? So what I would recommend within their own teams is that everyone get their team together and have a conversation if they haven't already. That is maybe bring in someone from HR to facilitate but what's hard about our work? Where do our processes really frustrate you? What do you hate about your job? What do you hate about the work we do? Where is something really slow, where it's really difficult, just boring? And then have HR go to IT and say, so which of these could we fix with tech, whether it is AI or something else but also bring in the folks who are function-specific technology experts because a lot of times it's going to be some sort of embedded tool but make it a two-stage conversation where our pain points and then could tech do it. Instead of asking can tech do it and then trying to figure out where it goes in your workflows 'cause that later one tends to actually just mean that we've picked tech for use cases that are nice but for instance, it's relatively easy to use AI agents to automate address changes from employees across all the different systems that they might need to change their address in. However, that is not a frequently done activity. So they've returned on the investment and doing that, it's not bad, hi, so you can do it. It's a great example of something that a Genetic AI could do very well. It just doesn't really necessarily actually give you that much in the way it returns. Yeah, think about the ROI of it all. As we wrap up, is there any final point, anything we didn't address within the future of work that you wanna mention? I think the thing I would say that is usually a little bit of a surprise for folks is that with generative AI, there is no first mover advantage. It's totally fine to go slow and wait and see what others are doing. That doesn't mean there's a first mover penalty. It's just this is a new terrain, but it's also terrain that outside of some very specific spaces, few in legal, few in R&D, there aren't business cases for investing before your peers. It's not a differentiator in that way because you would have to be using it really differently. And if what you're planning to do is implement an enterprise wide tool, it's fine if you wait and see if that actually is gonna be the best practice for you because that's not a game changer. It's a nice to have, but it's not necessarily really gonna change how you operate in substantive ways. I'm sure many legal teams across enterprises would also appreciate this message as you mentioned. It's okay to wait. So ask your legal teams before you launch anything because they may have concerns and the worst thing you can do from an adoption standpoint is roll something out and then have to cancel. - Yes, that just kills enthusiasm for the tool. - Yes, wait, the momentum, the progress you can make with it totally makes sense. - Well, this has been a fantastic conversation. Thank you so much for everything you shared. And thank you to everyone for listening to this latest episode of ThinkCast featuring Gartner Senior Director Analyst Emily Rose McCray. - Thanks Emily. - Thanks a lot, exos. - Thanks for listening to this latest episode of ThinkCast featuring Gartner Senior Director Analyst Emily Rose McCray. If you'd like to learn more about our topic today, visit gartner.com and click the links in the notes. ThinkCast will be back wherever you listen to podcasts a week from today. In the meantime, please rate, review and subscribe and share with a colleague. So neither of you will miss it. - ThinkCast is a production of Gartner. This podcast may not be reproduced or distributed in any form without Gartner's permission. It consists of the opinions of Gartner's research organization which should not be construed as statements of fact. Content provided by other speakers is expressly the views of the speaker and/or their organization. While the information contained in this podcast has been obtained from sources believed to be reliable, Gartner disclaims all warranties as to the accuracy, completeness or adequacy of such information. Although Gartner research may address legal and financial issues, Gartner does not provide legal or investment advice and its research should not be construed or used as such.