The Biggest Challenge of AI at Work Is Not What You Think
Ask about Knowledge at Wharton
Episode Description
ABOUT THE EPISODE AI is transforming the workplace, but are companies actually getting the results they expected? The biggest challenge may be changing how people work, not simply adopting new technology.
Prasanna (Sonny) Tambe, professor of operations, information and decisions at the Wharton School, joins us to discuss the latest research on AI adoption and workforce transformation.
Learn why AI agents raise new questions about accountability, how AI is affecting hiring, why verifying AI generated content is becoming a bottleneck, and what businesses need to do to turn AI investments into real results.
ABOUT THE PODCAST This Week in Business features interviews with Wharton faculty about the latest news, fascinating trends, and issues impacting both consumers and the business world. Episodes are recorded at the Wharton School and published twice per week.
----- The Wharton School of the University of Pennsylvania is committed to sharing its intellectual capital through Knowledge at Wharton, the school’s online business analysis journal.
Launched in 1999, Knowledge at Wharton offers free access to:
- Articles, podcasts, and videos highlighting Wharton faculty research and analysis of current business trends - Interviews with book authors from Wharton School Press - In-depth series of curated content like Wharton Executive Education’s Nano Tools for Leaders - A searchable database of more than 10,000 articles covering all aspects of business - A weekly newsletter that delivers Knowledge at Wharton insights directly to your inbox
Visit the Knowledge at Wharton homepage: https://whr.tn/3qUpSBE Subscribe to the Knowledge at Wharton YouTube channel: https://whr.tn/3DEAgAU
Read the full transcriptHide the full transcript
- Well, as artificial intelligence continues as a growing component of our society and our economy,
- it's important to understand just how much it is being used and where it's being used. Plus,
- how will this impact the future? A partnership between the Wharton School and GBK Collective
- has looked at these components of the last four years to better understand the level of usage
- around AI. Sonny Tambe is a professor of operations information and decisions at the Wharton School
- and he joins us to discuss the research. Sonny, great to have you with us today. Thanks very much
- for your time. Thank you for having me. It's great to be here. So in regards to this report,
- what are the important areas of focus that you're looking at from year to year? Because this is
- the fourth year you've been doing this report. That's right, exactly. We've been tracking this
- now for a number of years. I think this year, one of the central themes that's emerging is
- workforce transformation. I think the technology is obviously amazing. I think now managers are
- turning their attention, though, to we have to make it work, we have to integrate it,
- and that's raising some hard questions around how to transform work around it. So we're seeing a lot
- more signals from that kind of arising in the report. So when you think about the office and
- the setup and what happens every day, where is that transformation occurring? A number of places.
- I think, for instance, decision making. We have decisions that are made in organizations,
- and we assign responsibility, we assign authority. Now we have a new kind of intelligence. We have
- agents, we have AI. We're saying that we can use this new kind of intelligence to make decisions.
- But that raises the question, well, where does responsibility for that decision lie?
- Not always an easy question. Yeah, AI agents are seemingly the latest hot button topic
- that's out there around AI. What kind of role do you see them having, at least right now, but also moving forward? I think this is a grand experiment, and I think
- organizations are right in the midst of it. It's amazing to see how they're able to ingest all this
- context and data within an organization and make decisions, but those decisions are sometimes
- cutting across different parts of the organization, across different groups, and that raises
- challenging questions, of course, around who's going to be accountable for that. So I think in
- the longer term, I do see agents playing a big role in a lot of the tasks that we're doing at scale,
- but all of these things have to get taken care of first. And does the regulatory side of it still
- being figured out complicate some of these components here as well? Yeah, I think so.
- That's a great point. I think in some of these sectors, for example, consumer finance,
- healthcare, HR, where regulation has always played a big role, have to figure out first exactly how
- some of these workflow decisions that we're making, how some of this transformation intersects with
- those important components. A lot of conversation about the level of spend that we're seeing
- in terms of the build-out and how the software is being developed, and obviously the chips as well.
- Is the level of spend that we're seeing play out, maximizing, is it reaching the rate of return
- that everybody expects that we should see, or has it not reached that yet? Yeah, it's on top of
- everyone's mind. I think there's two positive signals coming from the report. One is that
- managers appear to be in a much better position when it comes to tracking AI spend than they have
- been historically. And the second one is that they're in a better position, I think, to tie
- some of these use cases to performance indicators. That being said, I think the challenging thing
- here is the economics or the tokenomics, as people call it, is changing so rapidly. So the cost of
- tokens is changing so rapidly. The agents are much more token hungry than prior use cases.
- So when we look at all of those things combined, I think that's what makes the answer to your
- question particularly hard right now, and something that managers are going to have to keep an eye on
- for a couple of, yeah, the near future. For a while, it seemed like one of the important questions
- being asked was how this was going to impact workforces. It seems like the belief is that
- it's not going to impact them as much as a lot of people believed at the first. Are you seeing that
- in the research? Yeah, I think so. I think there are, like, as with many things, the answer is
- doesn't seem to be lying in the extremes. It's, you know, there have been claims that it's going to,
- AI is going to generate an enormous number of jobs. There have been claims that it's going to wipe out
- a large number of jobs. I think, at least so far, we're seeing something in the middle where there's
- a lot of job, there's job growth in some areas. There's also cooling of entry-level hiring in some
- areas. The net effect of that, we're going to have to watch and see. But I think a more nuanced
- interpretation of what's happening is probably the right one. What about the workflow within
- the office? Because that's a different dynamic than actually just the hiring component. Yeah,
- for sure. And I think there's some really interesting signals here as well. For instance,
- one of the things with AI for agents is we're seeing a shift from producing content to verifying
- content, right? So a lot of managers are saying we're spending more time verifying content than
- we were before. And that is a, it's a profound shift in some ways. When you have that much
- content being generated, who is supposed to be verifying it? What expertise do you need to verify
- it? Whose job it is to verify it and be responsible for it? All of those require workflow changes.
- Otherwise, they just produce bottlenecks somewhere else in the workflow. So that is, I think, part of
- what's going on right now as organizations adjust to these AI tools. How do you determine then the
- difference between, the difference between strong returns and weak returns right now?
- Yeah, one of the things we're seeing, I think, is that this, as this transformation process goes on,
- right now we're seeing that if you are, for firms that are, for organizations that are investing in
- agents, the middle is a bad place to be, right? So kind of adopting agents and not doing a lot
- necessarily on the transformation side puts you in a relatively bad place. What's working is people
- who have gone all in on one side or the other, right? So you can imagine that you're maybe
- waiting insane, or you've gone all in, we're going to adopt agents, but then we're going to go
- ahead and commit to work for the kind of workflow transformation that's going to support that.
- So being on either side of that decision, I think, is the right place to be right now.
- But if your company hasn't gone into this point, is it relatively easy? I mean, you're going to
- have to invest, but is it relatively easy to catch up? Or are you going to be in a constant state of
- catching up as we move forward? Yeah, I think, I think one thing that we do have to accept, which is
- maybe an uncomfortable place to be, is that this is going to keep evolving for a little while. We're
- not close to the end here. I think for most organizations, one thing they have to absolutely
- be able to do is develop kind of a muscle to sense and track what's going on in this space.
- Now the question of when and where to commit is an organizational question, but developing that
- muscle of internally having people who can understand the landscape and what it means for them.
- Not all organizations have it, but I think it's an important capability to build for companies
- because of the pace of this technology, which is unusual. Yeah, and of course, that evidence is
- correlational, but there's a couple of things it brings to mind. One is that maybe it signals that
- those firms are the ones that are fully committing to the use of AI. Another possibility is that
- one of the challenging things about AI is that it's inherently cross-functional. It tends to
- cut across legal, sometimes marketing, sometimes tech. And when you have a technology like that,
- sometimes you need a person who's in charge of the technology and its use cases, not just
- individual silo owners. And I think maybe the chief AI officer sort of signals that kind of
- commitment that there's one person who we can point to to say, get this done for accountability,
- adoption, and all of those kinds of things. Is then off of that, what you just said,
- is there an expectation that the cross-functionality that we are seeing will continue to grow? Like that is the be-all, end-all goal of the companies producing the
- software and producing these platforms is to make it available and make it work for
- so many different sectors out there. If I had to make a prediction, I would say that that is a
- direction we're going to continue to go in. I'll give an example of, you know, one hot topic right
- now is solopreneurs. These are people that can use the technology to do all different kinds of
- things without having to rely on other functional experts. Even within organizations, we see some of
- that kind of activity where people in one occupational discipline are starting to do work
- in a whole lot of others. And so I think this cross-functionality is going to continue to grow
- for a while where that technology is in wide use. So since this is an annual report, as we look at
- the next year, what are some of the things that you are focused on that you want to see potentially
- play out, areas of interest? Yeah, I think it's the space is evolving so quickly.
- I think some other things that we talked about we'd certainly love to be able to continue to track
- are the verification piece of this, which is bottlenecking, I think, a lot of organizations,
- how you transform work around it. Another area was AI sentiment. Generally, we see how the public
- and employees and the public are reacting and thinking about AI, especially with the political
- cycle coming up. I think a lot of that is interesting to track as well. So a little early
- to say exactly what we'll focus on, but some of those have been interesting areas to talk about.
- Sonny, appreciate the time today. Thanks very much. Thank you. Sonny Tambe, who's a professor of operations information and decisions here at the Wharton School.
Brief Summary
Sonny Tambe explains how AI agents are changing workplace decisions, job patterns, spending, accountability, and the growing need to verify generated content.
Key points
- AI’s accountability problem. AI agents can draw on organizational data and make decisions across teams. That raises a difficult question: who is accountable for what they decide?
- AI’s impact on jobs is nuanced. The research points away from extreme predictions: some areas are seeing job growth, while entry-level hiring is cooling in others. The overall effect remains uncertain.
- The new work of checking AI. As AI generates more content, managers say they spend more time verifying it. That shift creates new questions about expertise, responsibility, and workflow bottlenecks.
- The middle is a bad place to be. Organizations adopting agents without changing how work gets done may struggle. The stronger paths are to wait or commit to the workflow transformation that supports AI.
- A capability for an evolving AI landscape. Rather than expecting to catch up once and be done, organizations need the internal ability to track AI and understand what it means for them.
Episode highlights
- AI’s accountability problemAI agents can draw on organizational data and make decisions across teams. That raises a difficult question: who is accountable for what they decide?“Where does responsibility for that decision lie?”1:36Oct 21, 2026
- AI’s impact on jobs is nuancedThe research points away from extreme predictions: some areas are seeing job growth, while entry-level hiring is cooling in others. The overall effect remains uncertain.“The answer is, doesn't seem to be lying in the extremes.”4:31Oct 21, 2026
- The new work of checking AIAs AI generates more content, managers say they spend more time verifying it. That shift creates new questions about expertise, responsibility, and workflow bottlenecks.“We're spending more time verifying content than we were before.”5:18Oct 21, 2026
- The middle is a bad place to beOrganizations adopting agents without changing how work gets done may struggle. The stronger paths are to wait or commit to the workflow transformation that supports AI.“The middle is a bad place to be.”6:07Oct 21, 2026
- A capability for an evolving AI landscapeRather than expecting to catch up once and be done, organizations need the internal ability to track AI and understand what it means for them.“We're not close to the end here.”6:58Oct 21, 2026





