AI's real impact on jobs - what it means for you, your company and future generations

The business models of today, the talent models of today, the jobs of today are all going to reforge themselves into new jobs, new tasks. So I would say the single biggest shift with AI is that foresight [music] is completely fogged. Welcome to Radio Davos, the podcast from the World Economic Forum that looks at the biggest challenges and how we might solve them. This week, what impact will artificial intelligence really have on our jobs and the education and training systems needed to prepare [music] current and future generations? >> The jobs of the past are moving out very quickly and the jobs of the future are coming at a [music] slower pace. >> There's a minimum of that 11% of the current world's [music] workforce that will need support in a managed workforce transition to probably an adjacent [music] industry if not a wholly different industry. Ravi Kumar, CEO of technology services company Cognizant says while money pours into AI tech, it also needs to be flowing into education and training. >> Reskilling should be a part of the infrastructure stack. It cannot be done on the side. You have to look at it with the same lens as you look at it for compute, for LLM access. You have to make reskilling a part of the infrastructure stack. >> And World Economic Forum Managing Director Sadia Zahedi says no one should doubt that change is happening fast. Things are moving perhaps [music] a little bit slower than we might have all thought a couple of years ago, but then on the other hand, they're still moving a lot faster [music] than our current systems are designed to address. >> For the radio Davos River podcast, I'm Robin Pomeroy with this look at the real impact of AI on all our jobs. >> How do you integrate that into everything you do? >> This is Radio Davos. Artificial intelligence is already having an impact on the way most of us work, but it's still early days and it's hard to predict exactly which jobs will disappear completely, what new ones will be created, and when and where this will happen. It's a subject that's important to all of us. So, my colleague Gail Maravitz spoke to the head of an AI services firm, Ravi Kumar of Cognizant, and to the head of the World Economic Forum's Center for the New Economy and Society, Sadia Zahedi. Both have unique insights into the impact of AI on the economy, on employment, and the implications for education and training. Gail started by asking RaviKumar for his assessment of AI's impact. >> One thing is certain, AI has created a fog on the foresight. Modern businesses, modern finance has been built with this assumption that we're going to have long slow bets. And with the fog around us, that foresight is broken. And um those long slow bets are no longer valid because the models, the business models of today, the talent models of today, the jobs of today are all going to reforge themselves into new jobs, new tasks, new job families. So I would say the single biggest shift with AI is that foresight is completely fogged. uh having said that I think we have this unique opportunity to repivot a future which is more shared prosperity and we could repivot in a direction where the drift of value coming out of AI goes into the front lines. uh over the last 50 years technology has drifted that value upwards and the front lines have actually been kind of uh not actually had that value shift and therefore we have created a a level of divide. I think we have this unique opportunity to repivot and drift that value downwards, capability downwards and we hope that value actually moves downwards as well so that you have more wages for the frontline workers. That's what I'm hoping so and that's what we we believe we have a very nice opportunity to you know reset the whole future of work workplaces and workforce. >> Sia why would you say this is a global problem rather than a company by company issue? I think um what Ravi said is just so spot-on both at a micro level and at a macro level. And maybe just building on that um maybe there's three big reasons that I guess um we tend to think of this as something that is a global challenge. I'm not sure it's fully a problem. I think there's a huge opportunity here as well. Um but one element is of course that we found that the bet that's been placed on AI and the massive amount of investments the time to productivity gains is somewhat slowing down which means that there is a very interesting window in which to make the equivalent workforce transitions if we want those productivity gains but also um be able to do that in a way that people can sort of absorb and manage. I think the second element is that there's society level um concerns that have to be solved for because not every organization will be able to manage this by themselves. Um what we found in our last uh one of our last reports and we'll do an update later this year is that if the the world's workforce was 100 people, 11 of them would not be able to reskill and redeploy in their current organization. Which means there's a minimum of that 11% of the current world's workforce that will need support in a managed workforce transition to probably an adjacent industry if not a wholly different industry. So there's there's a set of support that is needed between the public and the private sector there. And then finally the third element just again at that very big picture level is that AI is not the only trend and it is interacting with a number of other trends that include all of the geopolitical and geoeconomic shifts we see and that include the demographic shifts. So let's take the ge the demographics piece. There is that opportunity here for many of the world's workforces where there actually is a reduction in talent over the long term due to um aging or shrinking workforces. There's an opportunity here to integrate artificial intelligence in a very different way. And equally so there's an opportunity for AI tools to actually support and upskill talent in many parts of the world where education systems have not been able to keep up with the needs of many young populations. So I again it's it's a complex picture but I think there are some broad global implications that go beyond what is happening inside each organization. So Ravi, I I know that Cognizant has recently conducted some new research on AI and jobs and um it's revealed some changes that you hadn't predicted. What's what was the statistic that stood out that really surprised you? >> We did a research in 2023 and we did another research in 2026, early 2026. We also partnered with uh the World Economic Forum to do some joint research as well. Um you know when we did this in 2023 we expected impact on every job and when when I mean impact on every job the job by itself uh changes in form uh the tasks underneath change some of those jobs will remain but they will reforge in a different direction and there'll be new jobs 90% of the jobs we've we actually in 2023 said will actually get impacted by AI at least 10% of those tasks by 2030 and here we are in 2026 93% of those jobs when we did the second survey in 2026 we realized that um they they have actually got impacted. We took um almost 18,000 tasks a thousand occupations and we kind of uh um you know conducted that research on on those occupations and on those tasks. What is fascinating is the velocity of change is significantly changed. You know what what was supposed to happen in 2030 is now happening in 2026. We saw the velocity of change is at around 9% every year. Uh 50% exposure to jobs where there is you know you would actually believe that the jobs have changed tectonically. uh 50% exposure uh 30% of the occupations actually have gone through that 25% exposure where you actually have those jobs reforge in a different direction uh we almost we almost uh saw 56% of the jobs going through that kind of uh change. So most of these jobs we have today those occupations are going to reforge in a different direction. Some are jobs which are going to be futuristic. In fact, we believe AI is going to be in the middle of a flow. They're going to be a lot of jobs on the front and a lot of jobs on the back. The ones on the front are related to authentication, problem finding, creativity. The ones on the back are going to be validation, verification, judgment, accountability, outcomes. So, you're going to see a lot of those jobs. So, the asymmetry is not going to be about expertise and intelligence. The symmetry is going to come from applying that intelligence um and uh in including that intelligence as throughput into your input factors on a job. You know, I'm actually fascinated by the opportunities which this uniquely presents to us. You know, look at it this way. If you're going to create more throughput, more productivity uh on the front lines, you're going to actually shower more wages. Uh per capita wages have not gone up for the last 20 years. if you uh if you adjust it to real inflation. Now if wages go up because throughput goes up and uh you're going to get significantly higher output without inflation it's a good thing for the economies. Uh the point is how do you drift that capability downwards? How do you push more more throughput downwards so that uh there is a incremental wage and higher throughput that's what we are looking for. So this research uh tells us that it's coming at rapid pace. The capability is right out there. The production value in enterprises is way below. And this is what Sadia was referring to. The production value is way below. The capabilities out there. There is a big bridge. So we have a unique opportunity on that bridge to build you know workforce skills and and uh uh reskill our workforce so that the production value can go up. And we need the production value to go up because the technology you know you know this in the last 12 months a trillion dollars has been invested into AI infrastructure. The scaling loss for this infrastructure is only 6 months which means in 6 months it goes obsolete. So the [snorts] faster you could actually drift that value to enterprise production value the better it is. And to drift that production value and bring it to the same level as capability you will have to reskill the workforce. I've actually said this in a uh in the ve uh research we jointly did that reskilling should be a part of the infrastructure stack. It cannot be done on the side. It has to be a part of the infrastructure stack. You have to look at it with the same lens as you look at it for compute for LLM access. You make reskilling a part of the infrastructure stack. >> What does that look like especially for say a nontech organization? What what does that mean in practice? And and and I'm interested also that you use language like bridge because it really is like building something. >> The reason why this production value is significantly lower than the capability I mean the capability will keep going up. So the bridge is actually going to be broadened and you know you have to keep bridging it and I call it the velocity gap. And the reason is very simple you know we can't apply this technology on old stuff in businesses. You have to reinvent flows. You have to reinvent the ability to integrate digital labor with human labor and to re and to reinvent and reimagine business models, operating models and business flows. Uh this technology is very contextual. Uh the classical software we wrote in the last 50 years was very deterministic. It we codified it. A lot of the balance things we do in a workplace in the flows of a business are very contextual. the judgment oriented. You have to ground this technology into the hustle of the company. We call it a science called context engineering where we uh we put the guardrails. We uh you know we we put the harnesses needed for this technology to uh be productive and it has to work in sync with human effort and it has to be integrated. So all that work really means you have to reinvent those businesses and therefore I believe that bridge is very important [snorts] and uh you will then have to redesign redesign jobs you'll have to redesign the flows of jobs you have to redesign tasks which are done by people and how they integrated with uh what is done by digital labor we need a good nice landing spot on this uh and that bridge um I feel um has to be smoothened a little bit because the jobs of the past are moving out very quickly and the jobs of the future are coming at a slower pace. So in between the two we have to create a bridge. In fact I I wrote this thesis that you should tax to a large extent digital labor um which is eliminating tasks versus amplifying human potential. And that's a temporary bridge. That temporary bridge of taxing the capital so that uh there is a little bit of a level playing field between human effort and digital effort. I mean remember capital got a free runway because it actually created more wages and more jobs. Now if capital is creating digital labor and human effort and human labor and human work and jobs of the future are coming at a slower pace, you need a nice landing spot. So, you know, one of the suggestions I had is to look at capital taxation in a slightly different way in a short term and use it as a smooth landing spot. It sounds like a a huge challenge for human beings to to to kind of come up to speed with that with that gap. Is there is there some good news? Is there something I know there's been research Sardia um there have been some misconceptions about the time it takes for example to upskill? >> Yeah. So actually in the research that where we partner with Cognizant um uh there we found that it is not quite as prohibitive as people may think to have a base level of understanding when it comes to AI um and big data. So roughly through 30 hours of study um but of course to become more proficient then you're talking about 137 hours or obviously a lot more depending on the level of depth that you want to get into. But I say that to note that in again in some of the crossindustry surveys that we've done, one of the fastest rising in- demand skills is AI and big data. But no one is suggesting that that needs to be at a level of depth. It's simply the ability to be able to work with and understand technology. And what also comes through over and over is that the organizations that are likely to be the most successful are the ones that will be able to combine artificial intelligence with human judgment. And that means there's still a huge premium on creativity and collaboration and um interpersonal um dynamics and leadership skills and social influence skills. So all of that combined with then the ability to understand. So I think it's increasingly that you won't see two completely different tracks. You will need to bring some of that together. Um so I think just adding to to what Ravia has said that's what's going to be needed. But this is where some of the pain points come in. So um things are moving perhaps a little bit slower than we might have all thought a couple of years ago. But then on the other hand, they're still moving a lot faster than our current systems are designed to address. And so that is where most organizations will need to move forward very quickly in thinking about as Robbie said, workflow redesign, but then very quickly thinking about what are then the consequences for the people that are currently attached to a set of occupations that will go through a lot of change. Um and then at a policymaker level I think something very um similar does need to be done as well and that's where the piece comes in where not there is simply no way for each organization to handle this separately and then there's the other pieces around how do we fund this and that's where there are ideas such as what Ravi has mentioned and there's also some other ways in thinking about how to do this because of course the costs of reskilling and upskilling are also going down because of artificial intelligence and the ability to personalize that learning learning and reskilling and upskilling are also that that ability is just so much higher with artificial intelligence. So there are some ways to turn this technology um and its disruptions into an advantage when it comes to speeding up reskilling and upskilling. >> A follow-on question from that is um given it's so much more fluid to upskill and reskill um do you think uh expertise will be in some ways democratized? It's a question to Ravi. Does it mean that we're all going to become generalists? >> I think what's going to certainly happen is there's going to be diffusion of this technology much deeper downwards. That's because the interface is natural language. Uh unlike in the past when you needed digital skills to access technology uh this is kind of democratizing that process. Uh you know we had this distinction of a producer of software and a consumer of software. that line is blurring. Everybody can be a producer and a consumer which means you could build your technology and allow it to uh amplify yourself. So expertise in some ways is going to be on your fingertips which means the asymmetry we created over the last 50 years based on expertise. We we created a symmetry with individuals we created asymmetry with organizations that isn't asymmetry anymore. The asymmetry will come from interdisciplinary skills. You should be a biologist with the ability to use agentic to improve your uh throughput, improve your uh output. You should be a historian to uh have you know four clock terminals around you to be a futurist. You could be a child accountant uh you know having a bunch of AI agentic uh work [snorts] around you to uh power your insights that is the future we are all looking for which means the ability to absorb this as an interdisciplinary skill is much much easier as sia pointed this out it's much relatively easier in comparison to uh what we did in the past because expertise was really the symmetry the symmetry now is interdicciplinary skills. We do you know we need this intersection between a domain uh a business operations and technology and I think that that is much relatively easier in including the fact that you could also use AI to create a personalized micropersonalized tutor. I mean, we now have this unique opportunity to have a tutor and a nurse for each individual, each each person on the planet at a throwaway price. >> That's the power of this technology. I think we have to pivot this to these meaningful, purposeful use cases which will support this process. You know, over the last 50 years, the drift of value went upwards. We we created layers of white collar jobs. We captured value there and we we created premium on wages. If you're pushing that downwards, a nurse in a hospital, a frontline worker in manufacturing, they would have this capability. But the way you have to design the workflows, the way you have to design organizational structures is you have to drift drifting drifting the capability downwards doesn't necessarily increase wages. You have to drift the value downwards as well. You because value actually follows controls. It doesn't follow access. Once you do that redesign, then the asymmetry will shift to judgment, accountability and outcomes. And once you have judgment, accountability, and outcomes on the front lines, you're obviously going to pay more wages. And so there's going to be more distributed wages in the in the process. So I think this is the repivot we have to do. We have got this unique opportunity to reset our workforce and the work we do and the way we actually distribute value and if we can design this well, this is a unique opportunity for that reset. >> So in that scenario, what happens to the kind of traditional talent pyramid? uh is that no longer that relevant? [clears throat] >> Gail, I've been a big believer. I've written quite a bit about this extensively. It's a contrarian view. Uh I think the pyramids are going to be broader. They're not going to be, you know, the pyramids were like this. They're going to be broader. You'll have more early careers and shorter path to expertise. The entry barriers on the pyramid are going to be disappearing. I've been a big believer of this. At Cognizant, we hired 20,000 school graduates last year. We're going to hire more than 20,000 this year. The year before we hired 12,000. So, entry barriers to jobs are going to be in some ways disappearing. You know, a lot of jobs were STEM related. Now you're going to see STEM and non- STEM because effectively you could be a producer and a consumer and you could intertwine technology in your daily flows which means you know you need a lawyer with agentic skills you need a biologist with agentic skills to do life sciences drug development kind of a thing. So you have broader pyramids shorter pyramids. The middle layers in every company are going to be player coaches. uh we also had roles for coordination orchestration. Those roles will disappear. So there'll be more player coaches roles in in the middle and those nodes are going to be very real and agentic. So you're going to see digital labor doing things which were in the past related to coordination, orchestration and and and and moving information up and down as I call it. >> Those roles will disappear. They will get transitioned to digital labor. The new roles are going to be player coaches in the middle and you're going to see much broader pyramids. That's a phenomenal thing. I mean, if you have much broader pyramids and shorter path to expertise, you're going to see more modular teams, more singular pods or singular squads. I mean, this is brilliant because you, you know, [snorts] to express yourself, you don't need large teams, you need small teams. In fact, to express yourself over the last 50 years, we used institutions to uh to leave a mission and um uh and work with companies who actually have shared mission. Now you could do that in a much more modular democratized way. So I think that's the future of uh how organizational structures are going to be. They're going to be moreorked versus hierarchical. There is this rumbling negative narrative around job losses and we even saw recently with commencements uh some of the graduates booing tech leaders for example what does real augmentation look like and how how can we persuade those graduates that actually this is an exciting time to be entering the workforce but I think in terms of the what does real augmentation look like it is something around that player coach model that Ravi's just mentioned But many organizations haven't quite made that bridge yet. I'll I'll then step back and just refer to what we've found so far. There is an overall net positive. We have found that it is very likely that there is likely to be job growth rather than overall job displacement or reduction over time. And that would point to that healthy growing bigger base pyramid that Ravi is referring to. that is likely based on everything we've [clears throat] heard so far. At the same time though, there's probably sort of three ways that people are thinking about this. There's a set of people that believe this is sort of an early canary in the coal mine situation. You're going to have these large urban rest belts because a lot of entry-level and middle level roles are going to get wiped out. There's a set of people that I think again we just discussed this piece broadening pyramids. actually we're going to need so much more talent not just because of the augmentation piece but because of the wholly new roles and new value add that people can bring as some of this workflow redesign takes place where essentially wholly new products and services are um possible to create some things that are not possible to imagine right now because we're still thinking in the domain of current jobs but if we think four five years from now this would just be wholly new set of jobs and then there's the set of folks that I think would probably say actually none of this is true and we're essentially looking at a number of organizations that are tightening their belts due to the current economic situation and that is why you're seeing a reduction in some of that entry-level work and actually has nothing to do with artificial intelligence. I guess it's really going to depend on industry and organizational readiness and what they're actually absorbing in terms of technology and not every industry is making this leap at the same pace as others. So I think that pyramid and how things go is going to look very different across different organizations. But one point that I think is probably consistently true for over the last 10 years we have found business leaders telling us in one form or the other that for about 60 or 70% of them the lack of entry skilled entry-level workers is one of the major things holding back the transformation of their organizations. Which means that with or without artificial intelligence, what the education and university systems have been producing in terms of talent, while it may do it that while there are many good things about it, it doesn't always equip young people with the new economy skills that they need today. And so if that is the case, then a lot more effort needs to go into building simply those new economy skills inside education systems and as they enter the workforce. And I think that's where we have to build the bridge. That's where a lot of the forums time and effort is going to be going. Ensuring that those crosscutting new economy skills are built up through education systems and as they enter the workforce because that's going to be necessary regardless of the particular shape of an org structure across any industry. >> Do you feel like traditional fouryear, three, four year degrees are still relevant? And do you think education is keeping up? >> That's a great question. Um you know the current system of going to full-time academic intervention for the first 25 years of her life working for the next 50 years and then retiring is a linear template from the industrial revolution >> where the world was running at a much slower pace. with the clock speed we have, I think we have to revisit that template where the K12 schools should kind of focus on building lifelong learners and then you have partnerships, industry partnerships for I would call it digital apprenticeships or AI apprenticeships or whatever you whatever you like and then we draw learning resources all alive [snorts] on a on a continual basis. Um I mean today the alumni associations of schools are actually for networks not really to draw resources all your life. I would think that template should be revisited. You should intertwine work and and learning resources all your life because the change is happening in the middle. It's not happening on the front and at the back. It's actually happening in the middle when you need it the most. So it's there is a certain revisit. Every institution is doing some experiments but it's not as mainstream. Uh I wish we could you know we could intertwine a few years of that undergrad education into apprenticeships in a different form. This is AIEled apprenticeships or digital apprenticeships and then we draw learning resources uh all our life. >> When we look at sort of K through 12 education let's also not forget that that is the place where um young people learn how to be members of society. There is so many other sort of skill sets and traits and characteristics that are built up during that time that are incredibly important. But in many parts of the world, that K- through2 education system is designed for competition and for rank ordering students by the end of a school year, which is very different from the skills that are going to be needed in the future. very few of those systems actually teach some of those interpersonal dynamics that teach what is needed in terms of collaboration. So the earlier that can begin I think the healthier it is for societies as a whole much less for businesses and the and the economy. And then on the university point um absolutely and this is why we've set up at the forum um the first mile sandbox which is all about creating those industry partnerships with universities including in the digital apprenticeship space that Ravi was mentioning. Um and we're beginning with five industries and we plan to roll that out across all the the various um industry groupings that the forum works with exactly for this reason because just this entire methodology has to change and to Ravi's point some of this is right now about that sort of first mile um sandbox and it's really focused on that early part of the career but this needs to be continuous um across across the entire um life cycle and you know there was an interesting stat where um something like1% of the GDP of OECD countries is spent on the lifelong learning piece. The post university learning, retraining and upskilling. Um many large businesses that can afford it spend thousands and thousands per employee in terms of retraining and reskilling. If that entire system which does have enough funding in it could just be better connected into institutions whose job it is and who have really the expertise universities and colleges and community colleges that can really do this at scale and if we could do that throughout entire life cycles that would really pay off and that's going to be necessary I think you know going back to the point that I was making earlier around policy makers that is essentially how policymakers will have to rethink the incentives they create for collaboration between private sector and the education sector. >> I have a teen and uh I wonder so ask asking for a for a friend um Ravi if there was one piece of advice that you would give her uh if she were graduating this year 2026 now as she's looking for a job and then perhaps once she started that job what what would the advice be? >> I have two toddlers at home. I I wish I could tell them this. Uh I would say the future is going to be much more interdisciplinary. You don't need to be a computer science graduate to thrive in the AI era. [snorts] Uh you need to figure out a way to apply the technology this extraordinary technology with [clears throat] you know it's a significant shift in terms of capability from the technologies of the past. How do you integrate that into everything you do and your professional life? One of the policy makers asked me this question saying um what what should K12 schools and undergrad schools do and you know with related to AI? I mean it's funny uh we tell students if you use AI at your work at at your class we're going to fire you. And we're telling employees if you don't use we'll fire you. And you know the dichotomy of dealing with that is you should build native skills >> at class without AI and you should build uh you should do your homework and your evaluations with AI. What then happens is you power your native skills with an amplification with AI. So effectively you have the native skills to do the judgment uh which Sadia was referring to outcomes, accountability and you know uh intellectual curiosity and everything else but you then amplify yourself with AI. So I would actually believe build the native skills without it and use it to amplify it uh and you know try to power this with interdisciplinary opportunities. You could be anyone. You could be a journalist. You could be a biologist. You could be a chemist. You could be a lawyer. You just have to look at this technology and say, you know, it's available on on your fingertips. How do I integrate it into everything I do and create more more productivity, new products, new services? >> What how would how will the economies who are doing this right look different to those who get it wrong? Let me maybe just give a a a quick overview of a scenarios piece that we did. And just in very simplistic terms, think of one vector where it's about how quickly technology is moving forward and being integrated across an economy. And think of the other vector as how quickly people are being skilled, reskilled, upskilled. And essentially the only no regret move available to policymakers is to combine that technology investment with the people-based investment. There is essentially no such thing as getting the returns from the technology investment without the equivalent people investment because these two things have to work together. There there is no other way. And so I'd say the the first thing is economies that understand that that do not think that the people related investments are an afterthought as Ravi said they have to be integrated into that stack to begin with. That has to be number one just that basic understanding. I think the second piece is um around raising the digital floor for everybody. Um because I think a lot of organizations, businesses, governments are thinking about that sort of top end. But what we still have to remember is there's nearly three billion people across the planet that still don't have basic digital connectivity. And so this is just an extremely fast growing chasm between the halves and the have nots. And so the digital floor does have to be raised for everybody. And then the third element is the public private collaboration that is going to be needed to manage this well. And I don't want to boil it down to sort of you know just a basic term like that. It's everything that we've just been talking about. It is that element of yes businesses have to do a lot within yes governments have to think a lot about policym on their own but those two sectors will have to talk much more to each other when it comes to managing this workforce transformation zahidi managing director of the world economic forum you also heard Ravi Kumar CEO of cognizant they were speaking to my colleague Gail Marovitz the world economic forum and its partners do lots of research into the future of jobs and skills [music] and on artificial intelligence more widely. Find that on our website, links in the show notes, and it's one of the subjects we watch closely on Radio Davos. Make sure you're following us wherever you get podcasts, [music] and you can find the forum's three weekly podcasts at w.chmpodcast. Radio Davos will be back next week, [music] but for now, thanks to you for listening and goodbye.