AI Isn’t Just Replacing Work — It’s Rewiring Organizations (Part 2)

This transcript has been generated by AI and may contain slight discrepancies from the audio or video recording.

Nichol: How can HR specifically build this learning and development system that supports this fast-moving adaptation and skill building, especially when traditional career ladders are fading?

Today we're bringing you part two of our crossover series with SHRM's Tomorrowist podcast, which explores the trends shaping the future of work. If you haven't listened to part one yet, you can find the link in the show notes. I'm joined again by my co-host for this series and the host of Tomorrowist, Jerry Won. Jerry, I'm excited to continue the conversation.

Jerry: Nichol, I'm really glad to be here.

Nichol: I am so glad to have you here. I've been wanting to do this for a while. In part one, we examined how AI is reshaping organizational strategy from the top down. In the second half of the discussion, we're exploring what AI means for job design skills, HR operations, and the reshaping of day-to-day work. We're thrilled to continue the conversation with Sangeet Choudary, Senior Fellow at UC Berkeley, Founder of the Platform Thinking Labs and author of Reshuffle: Who Wins When AI Sacks the Knowledge Economy. Sangeet, welcome back.

Sangeet: Thank you, Nichol.

Nichol: Okay, so from an HR perspective, what practical changes should organizations expect as work becomes more modular and AI helps allocate tasks across teams?

Sangeet: I think there are a few different things to think about from an HR perspective. The first is that we need to prioritize the idea of a learning organization. We've not really had a learning organization at the rate at which recommendation systems learn or any other technical system where we use AI learns.

But as AI capabilities improve and as we adopt them in the course of our workflows, we have the ability to constantly capture data from the workflows and understand how work is being performed across the organization. So, to a large extent, HR used to rely on episodic assessment and on episodic learning and development. I think both of those things have to become continuous.

You cannot rely on episodic interventions. By the time you design the intervention and introduce it, the need for it might have changed. You need to constantly sense, intervene, and improve how capabilities are developed across the organization.

So that's the fundamental point that I would talk about. One of the key misconceptions that I see people often have is that they talk about the role of HR as now managing not just humans, but also agents. It's a gross misframing.

What I actually believe is the right framing is that as HR, you have to still manage human capabilities as you always have done, but knowing that as agentic and AI capabilities increase and improve constantly, which human capabilities will be valued and which ones will have to be nurtured will constantly change. That is why we need this learning organization: constant sensing, constant data being captured from actual workflows, and using that to design interventions in real time.

So how we think about learning and development, how we think about matching capabilities to the right needs internally, all of those things have to change in response to this.

Jerry: I've seen some interesting research that people who are good managers of people are often better at managing agents because of the level of questions that they ask and things like that. So that really circles back to what you're saying about human skills. But I want to double-click down into a learning sensing organization and talent mobility. What does that mean for AI recommendation systems and employees knowing or having assistance in navigating what's next for their roles, projects, and skills pathways?

Sangeet: As this learning organization is constantly adapting, AI recommendation systems and managing talent mobility are going to be really effective only if they have a clear sensing mechanism, which is constantly capturing data and constantly capturing patterns about how humans or workers are constantly adapting themselves around new AI capabilities that are coming in.

You mentioned this point about the fact that people who are good people managers are also good at managing agents. I think that happens not just because people who are good people managers only have power skills. Typically, people who are good people managers also have good system design skills. They're able to allocate the right skills internally in the team to the right problems and reallocate and nurture those capabilities internally towards solving the right problems.

So what we really need, if we want to have a model where we effectively align skills to the needs that the organization has, we need really good system designers, system thinkers at every level of the organization, but even more so at those levels where human capability is being managed.

We need a good sensing mechanism by which how well humans or workers are reorienting themselves around increasing not just AI capabilities, but increasing AI adoption and how they are reorienting their own work in response to that. We need a good sensing mechanism towards that, and we need to, on the basis of that, alongside that also have a very good sensing mechanism at an organizational level of which capabilities are likely to be valued and which new needs are likely to emerge in the near future.

This is not a one-time fix because today when we talk about the impact of AI on jobs, there are tons of articles that come out with very pedantic solutions like AI will create new jobs like algorithmic overseer or things of that sort. That's not really what matters.

What matters is you need to have a very clear sensing mechanism as an organization to see how your competitors are changing how they perform work, what customers are paying for, and how products that they are getting access to are changing as competitors innovate and use all of that to determine what kinds of opportunities you'll have to create internally in order to compete effectively in the industry.

So you need all of these three things together, but you need the system design. You need the sensing of capabilities as AI gets adopted, and you need the sensing of needs as the way organizations compete and how they organize themselves keeps changing as companies innovate.

Jerry: Yeah, I love that. We're going to iterate to these new jobs. So I really appreciate how you lay that out. Is that what you mean by ecosystem skills in Reshuffle? You talked about them. What is that and what are the essential capabilities employees need for those?

Sangeet: When I think about the idea of a reshuffle, what I'm essentially saying is that the way we create value is going to change. But the power structures associated with that are also going to change.

So whenever technology changes and when new forms of technology come in, value creation changes, but incumbents can always hold onto traditional power structures and prevent any shift from happening. But with AI, the speed with which these shifts are happening, I believe that we are going to see not just a change in how value is created, because how value is created changes what your job is going to be. It changes what your company does because if the value creation logic is changing, every part of the organization has to change in response, but it'll also impact what power structures exist in the industry today.

So new forms of competitors will come in when knowledge that used to be tied to a certain industry can now be accessed by players in other industries. New forms of competition will emerge where competitors come in who did not look like you in the past at all.

We've seen this happen with structured data, not with AI yet in a big way, but if you look at car manufacturers getting into the insurance industry and insurers getting into automotive services and vehicle driving services, you see that industries have already collided and overlapped, but we're going to see that increasingly happen in knowledge services and knowledge work as well.

This is where the reshuffle comes in because your future competitors will not look like your past competitors. Your previous competitors could become partners in the future. The way companies win or lose is going to completely change.

So the idea of an ecosystem orientation is that you constantly have a view on all of these changes because the nature of the playing field in which you're playing is constantly changing. Traditionally, the playing field was very static. If you were a plastics company, you were a plastics company. If you were an automotive company, you were an automotive company.

But today, those playing fields completely change. Automotive companies also become energy companies because they have to be deeply integrated into batteries and plastics companies also become biology companies because bioplastics are coming in.

So the nature of the playing field in which you're playing is constantly changing. When that happens, the game that will win in that playing field also changes. Your previous game cannot continue winning if the nature of the playing field has changed. That's really what I mean by an ecosystem skill and being prepared for the reshuffle.

Nichol: I really love that word, ecosystem skills. So far we've talked about three north stars essentially. On our previous podcast, we talked about removing friction. On this one, we've talked about understanding where value creation goes and this learning organization. So I really appreciate your framework.

I'm curious for our HR leaders who are listening, how can HR specifically build this learning and development system that supports this fast-moving adaptation and skill building, especially when traditional career ladders are fading?

Sangeet: I think HR cannot be a purely internal-facing function anymore. HR is very good at an internal-facing function when we believe that the nature of an industry is fixed, the way the organization works is fixed, and it's going to be fixed and structured for the near future. In that case, all you're doing is nurturing capabilities within a fixed structure, and you don't need to see outside the structure in order to do that, because the rules of which capabilities are valued and which ones are not, and which ones will still be valued tomorrow, which opportunities are going to arise, all of that is fairly static.

But with AI improving at the rate that it is, as we established, the nature of the playing field is changing. The nature of the game you play changes. Hence how you value capabilities changes and how you organize capabilities changes.

HR is not going to be in a position to nurture and acquire and improve and retain those capabilities if it does not have a clear view on all of these aspects: on how the playing field is changing and how your identity as a company and your game is changing, and accordingly, how which capabilities get valued and which opportunities arise change.

So HR is going to be increasingly an external function. With the rise of digital transformation, a fairly back-office role, like IT, increasingly moved towards becoming a very central strategic role. Yes, there was a part of it that still remains there, but the idea of digital transformation and more importantly, competitiveness in the digital era is not just about managing your data at the backend. It's about reinventing your business model around data.

The same thing is going to happen with HR because the idea of competing in an era where work can be performed by both humans and machines is not a backend internal function anymore. You have to have a clear view on what your organization is doing, how it's evolving, how it'll compete, and accordingly, which opportunities will come up and hence which skills will be valued.

So the single most important test, or rule of thumb, I would say, is that if a year from now your HR function is still primarily internal-facing, you're not really prepared for what's happening.

Nichol: If every AI conversation feels both important and impossible to prioritize, you are not alone. The AI Plus HI Project 2026 cuts through the noise with immersive learning, live demonstrations and peer-driven labs focused on real HR challenges. Whether you're building a foundation or reimagining HR at scale, you'll leave with clarity, confidence, and tools you can use immediately. Space is limited. Register now at SHRM.org/aihipod.

Sangeet, when it comes to companies, as you mentioned, rethinking how you hire, train, and promote, talk to us about how companies can redesign or design internal talent marketplaces and micro-credentialing pathways to identify those people who are better equipped to both function as independent contributors and also as a leader in this new reimagined reshuffled world.

Sangeet: I think that's an important piece because if you think of the idea of micro-credentialing, there are two ideas implicit to it. One is the idea of micro, which is that the credential is no longer this big bundle thing. You can tie a credential to a very specific skill, a very specific need, and hence match the skill to the need.

But then alongside that, if you believe in the fact that both which skills will be valued and which needs will emerge will also change very fast, then credentialing, the second aspect of micro-credentialing cannot be static anymore. It has to constantly evolve.

So increasingly micro-credentialing will have to be intelligent, but it'll have to be implicit. You cannot curate this list of badges the way initially micro-credentialing was envisioned, or at least its physical manifestation is envisioned. We can't rely on that model anymore.

Going from improving the granularity through micro-credentialing is a good step. But relying on static credentialing or static definitions of skills and opportunities is going to break down dramatically. So we'll have to increasingly make micro-credentialing implicit in the system.

Increasingly as AI capabilities improve, not just the AI that does our work, but the AI that coordinates work across the organization, that micro-credentialing will increasingly be implicit to the logic of all of these AI capabilities through which organizational work will be coordinated.

I'll give a simple example. If there's a coder and in the past, a coder would have to then generate documentation about his code so that it could then be reused across other places. With AI as a coordination layer, increasingly the documentation can be auto-generated.

Depending on the reputation of the coder, based on how their previous code and the associated documentation was adopted by other parts of the organization, different levels of scores would be given to the code and to the documentation and where it can be applied and accordingly, where that documentation should be served and into which workflow will be determined on that basis and hence the way that coder could find opportunities internally would get impacted as well.

So we have to think about AI as a coordinating layer as well, because a lot of the things that we are talking about here were things that either information management, knowledge management roles used to perform or HR used to perform, and both of those will increasingly get collapsed and absorbed into AI as a coordination layer as well.

So we need to think about the impact of AI at all these three levels. First, AI changes which capabilities are valued. Second, AI constantly changes the division of work between what humans are doing and what machines or agents or AI is doing. And third, AI as a coordination brain determines which human output to align into which workflow and inform which human input where. That itself is a key determiner of how talent will be allocated across the organization.

Jerry: Well, that sounds like it's going to change career development, it's going to change hiring. How should HR think about both career development and then also rethinking hiring internally with talent mobility or externally? Everything's reshuffling, career development's reshuffling, hiring is reshuffling. How should HR think about these two?

Sangeet: I think that's an important issue to think about because the nature of where you source which talent changes with AI coming in, in three very important ways.

First, to what extent do you trust a particular quantum of work, and I'm saying a quantum of work, it could be a certain workflow, it could be a certain end goal to be achieved, to what extent do you expect that to be achieved with a specific AI tool? And if so, what human capabilities do you need alongside it?

Which essentially means that how you define the human skills required would change based on what your assessment of the AI capability is. Because as AI capabilities change, the people who can perform that particular work will change as well.

Take the Google Maps example again. In the past you needed to pass the test of the knowledge of London in order to be a cabbie. Today, you can just switch on Google Maps and navigate through the city. So the nature of skills and the type of workforce that can handle it or that can be aligned towards a particular quantum of work will constantly change. So that's one thing you need to look at.

The second thing you need to look at is will this work still need to be performed in the organization? When the cost of performing work externally went down because of cloud technologies, because of constant connectivity, we started seeing the rise of freelancer marketplaces. But still a lot of work does not get sent out to freelancer marketplaces because the cost of transferring the context is very high. You have to train them on a lot of organizational context.

But if a lot of organizational context is now available inside an AI assistant, and you provide that AI assistant to a freelancer, you can now perform the work externally. So you will have to constantly think about, is this work even supposed to be performed internally or can I find better ways to perform it externally? So that's the second important shift that happens because of this.

The third thing that you'll really need to think about is that as AI capabilities constantly improve, and in response to that as to some extent you need lesser human intervention on certain forms of work, how do roles or teams constantly unbundle and disintegrate and then reintegrate around new problems? Because that's really what's going to happen as organizations increasingly and successfully adopt AI.

Today, to a large extent, the challenge is not that AI is not good enough. The challenge is that organizations haven't successfully adopted it. The more successfully they adopt it, the more individual internal roles, as well as how teams are structured will constantly have to change.

So one of the key functions of HR will also be to really think about how to measure and benchmark how well the organization is adopting AI towards executing and achieving work, and how that is changing which roles are required and which teams are getting structured. So these things are no longer static, nor are they dissociated. Where work gets to be performed, which kinds of talent can perform it, and how talent and roles and teams constantly reorganize, all of these things are different levers of flux that HR will have to manage.

Jerry: Sangeet, one of the things that I keep thinking about is we're doing what we're doing. We have to learn what's new, and we have to adapt to what's coming. So talk to us through how HR leaders can foster and encourage a culture of learning and relearning while there seems to be a little bit of more reconfiguring, reshuffling. There's a lot of change. We know that change is here to stay and is only going to accelerate, but how do leaders balance this encouragement of continuing to learn while being judged?

To put it in layman's terms, how do we encourage a culture of learning while the target seems to be constantly moving for both leaders and for individual contributors?

Sangeet: This is really dicey because assuming too much change or too little change, both of them are equally problematic. If you assume too little change, you're not prepared for the reshuffling that's happening. But if you assume too much change, it blinds you to which parts of your organizational identity, how your industry works, how your organization is structured, which parts of those are simply not going to change.

For example, no matter how well or how good AI capabilities become, if an organization has certain values which are nonnegotiable, it'll still figure out how to reorganize itself while preserving those values. So you need to be very clear about which factors are going to be constant around which you organize yourself. In some ways, those factors would become the new friction, if you will, the new constraints. But at the same time, that is what helps you preserve your organizational identity.

So I want to start by just making that distinction. Don't be too inflexible. And don't be overly flexible to keep jumping onto the next thing. You have to have a core around which you constantly learn. As you learn, you can keep questioning your core. You can keep questioning if the core still makes sense, but to the extent possible, if the core, if you don't have a reason to change the core logic of certain things you do, any learning should come back and reinforce that core rather than work independent of it completely.

The reason I'm saying that is because understanding what is core, understanding where you need to explore, understanding how you tie that exploration and the learning back to your core, and by core I don't mean a set of core capabilities, I just mean a set of core constraints. Here's who we are, here's how we have organized ourselves, here's what we believe.

With AI coming in, some beliefs, some of those assumptions are going to change. Some are not going to change. Here's what's not going to change. If you have that very clearly laid out, you can be more confident of how to manage things in your day-to-day in the present because you know that these things are certain aspects of your core assumptions that are not going to change and hence any investment towards improving that in the foreseeable future until those assumptions fundamentally change are actually aligned with improving your competitiveness, improving your capabilities, while at the same time you invest in learning in anything that helps you explore and reinform and improve your core even further.

I'll just give a simple example. Think of the materials industry. Material science is changing. AI is changing how the materials industry innovates because you can now come up with new material compounds much faster. So coming up with a compound and protecting it through a patent is no longer necessarily going to be the way to win.

But there are certain core elements of the industry, which will never change. One of the key core elements is that a material should match the end user requirements that it is mapped to. The reason I'm giving this example is that even as AI and material science improves the range of compounds that you can create, a company that is clear about the fact that the material will still have to be validated, will now focus less on building capability than hiring talent around compounding and will focus capabilities and talent around validation and governance.

So if you're clear about what's going to stay, I wouldn't say static, but what's going to retain value, you can keep investing confidently in the present towards that. And where you are not clear, you keep investing in learning towards that. So you need to have this ambidexterity in order to manage this. If you're constantly just running and learning, you might just keep learning yourself into irrelevance because you don't have a direction.

Jerry: Sangeet, very helpful. Okay, so last question. If you're advising a CHRO who's starting their journey today, what is the very first step that they should do after hearing our discussion?

Sangeet: I think the first thing that you should really go back and think about is what are the assumptions about your organizational capabilities that you believe are immovable and around which your current organization is built. What are the assumptions around how you structure your teams? Just think about all those assumptions very deeply and then ask yourself, will those assumptions still hold once AI adoption works out well?

Once AI is well adopted, but also as AI capabilities improve, will those assumptions still hold? Create sort of a future scenario map and keep updating it. If you don't have a map, you won't know where you are navigating. So create some kind of a future scenario map that given certain future scenarios of AI capabilities and the extent to which they're adopted internally, how will those assumptions change? And accordingly, what will that mean in terms of your role as a CHRO?

So start there. You have to be a system designer. You cannot be a system optimizer.

Nichol: Wonderful. And that's it for this week's episode. A big thank you to Sangeet for sharing your experiences and your insights with us. Thanks for joining the conversation and we'll catch you next time.