The Future of Work: A Fireside Chat with Industry Leaders from LinkedIn and OpenAI

Hello, I'm Ronnie Chatterji, the Chief Economist of OpenAI, and welcome to OpenAI Forum. This is an event I've been looking forward to for a long time, because I'm here with Karin Kimbrough, the Chief Economist of LinkedIn.

Karin, a couple of reasons I'm looking forward to speaking with you. One is, you have a wealth of knowledge about the future of work, which is today's topic. But maybe more importantly, I get to ask the tough questions today. Usually people get to ask me really difficult questions about AI from the perspective of OpenAI. So I'm going to ask you all the tough questions today, and I think we're going to learn a lot about where the job market's going and what kinds of indicators people should look at.

Before we jump into that, though, I want to ask you, what does a Chief Economist do, and how do you find your way to LinkedIn?

Yeah, sure. So, first of all, excited to be here. I would say a chief economist at LinkedIn is probably the best job I've ever had, if I'm allowed to say that. It is an opportunity to look at all the data that LinkedIn has. And LinkedIn has just this amazing breadth and scope of data. So, if you think about it, there is a billion plus members worldwide. We're in nearly every country you can think of. We're able to look at, at any given time, what 70 million companies are looking for in terms of job postings.

Every minute, there's some 5 million updates or engagements by our users. So, what it really does is allows us to see both sides of the labor market.

supply and demand, and no one else has this comprehensive look into the global labor market. So it's fascinating to be in this role. It's an honor, a privilege, and an opportunity to ask a lot of questions around what's happening in the labor market.

And if you think about the last five years have been pandemic, and then a lot of tightness labor market, now more slack, and the remote work, the hybrid work, and now AI coming through as a real force of change. So it's been really interesting.

Wow. And when you think about 1 billion users, supply and demand side of the market, how do you prioritize all the questions you could be answering? How do you set your research agenda? How do you make sure that's aligned with LinkedIn and what the world wants to know about what's going on in the platform?

Yeah, I mean, obviously, relevance is key, right? So because if we're able to speak to the most relevant questions, we're going to be, I think, the most helpful. For LinkedIn, our mission is to create economic opportunity for every member of the global workforce. And so I think the first aspects of relevance is just understanding where the labor market is right now, what's going on right now, whether it's a macro cyclical story or it's a structural change story.

And then the second element of relevance is like, and the thing I get asked a lot is like, so what's going to happen in the future? How is, you know, I think it's going to change as a result of whatever impetus or event there is. And right now it's the question is generative AI. But what does the future of work look like? And so.

So we're truly focused on understanding what's happening in the labor market right now. How is it for job seekers? How is it for employers? And then also how is work changing in the future? Where will we end up?

Well, let's start with those two kind of frames of reference in mind. On one hand, you have the right now and trying to create opportunity for people in the workforce right now. And then you have where this is gonna go in the future.

When it comes to Gen AI, what are you seeing right now in the labor market, sitting where we are here in 2025?

Right now, I would say the labor market's actually reflecting a lot of macro cyclical effects. And I know that's not what we should be saying at open AI forum here, but I'm gonna start with the macro and then I'll move it over to AI.

If you think about it, right after the.

pandemic hit, there was actually a lot of hiring in 2021, 2022, globally, hiring just surged. And then everybody sort of froze in place. And since then, there was very little attrition. Quit rates were relatively low, people just sort of held on to what was a really cool, very flexible job opportunity at relatively high wage.

And now we're starting to see the labor market stabilize and come back to normal. So we kind of went from like a up cycle to a down cycle. And I think now we're sort of normalizing out. And what I mean by that is, we're starting to see a labor market that's neither too tight nor too slack. There's more competition than there was before, but not

so much that people can't find a job. So if I speak across all occupations, I would say right now is a labor market that's looking for normalcy, trying to normalize back up after very slow hiring.

If we bring it to generative AI, which is what you asked, the question is really, do we see right now evidence that AI is starting to transform the workplace? And I would say yes and no. I don't know if you want me to kind of elaborate.

I mean, what you're describing here is sort of like a Goldilocks economy, kind of finding it in the kind of the just right. And this point you made about macro factors, and you know, for those of us who live and breathe AI every day, it can sometimes be hard to remember. There's all these other things going on affecting the job market. So I would love you to dive a little deeper into kind of how you're seeing the gen AI.

shape the workforce going forward. And also, you know, you and I have talked before about white collar and blue collar jobs. I think a lot of things you're seeing there also might be pretty surprising. So we'd love for you to talk about that.

Sure, sure. Okay, so one of the things that we're able to look at in our work is actually labor market tightness. This element of like, how much competition is there among job seekers to find a job relative to number of jobs that are available. And when job seekers are competing with each other, obviously that's like a tough story.

And here's what I'll say. If you are in what is considered like the knowledge worker space, and it could be that you're a consultant, you work for professional business services, you are an accountant, an economist, a software engineer, maybe even a professional.

it might be a lot more competitive for you right now to find the role that you want. Right? The labor market is more slack for you. And particularly when we look at the tech sector, it is extremely slack, meaning there's a lot of competition for fewer roles than there were before. That is not the case if you work in retail. That is not the case if you work in healthcare. The world is your oyster if you work in those industries, right?

If you work in construction, it's been pretty good. You need a job, you will find one. So there really is a significant difference in the labor market. And I'm just talking about the US, but I could go globally and tell you the same story. Big, big difference between the jobs right now where you're sitting at a desk with a laptop versus ones where maybe you're out there in the field, in healthcare.

in construction and retail. Yes, and this is not necessarily because of AI, right? Because you see a lot, right? It's for structural factors. It's for structural factors.

So, and how do you know that? Say a little bit more about that, because I think a lot of people will look at those sectors and say, oh, well, one is more AI exposed than the other. Construction, let's say, vis-a-vis professional services. But tell us, tell us why you see these macro structural factors at play. I think it's important.

Sure, so, well, for one, you might anticipate that if, you might anticipate that if it were AI, you would say, well, actually, there should be fewer jobs, right? If retail's being somehow displaced by AI, there would be fewer jobs available for us humans. That is not the case. It is very easy to find a role if you want one. Same for construction, same for healthcare. Although, of course, generative AI and AI.

are in parts of all of those industries, but the other reason I say that is because it really is much more of a response to a backlash from the pandemic.

What we are really seeing in our economy is a kind of resettling after some really big roller coaster waves of highs and lows in the labor market, and a lot of that was hit by construction and retail and the service economy where people weren't willing to come in and be in person, and then of course that came back, but it was harder to find people because they rotated out of those roles, and you may have elements of immigration that's affecting how construction roles are being filled, you may have elements of just people's willingness to work in retail if they have other options.

And healthcare, of course, has been on its own structural kick because of demographics, because of all the things that are going on with the need for more healthcare. We are seeing just a huge demand and increase in roles there. So we're seeing the number of roles open up. It's not just the people are looking for a role and are able to find one. So there are actually more roles. I mean, this is fantastic. These are the insights you can get from having a billion users, people who are logging in and updating their profiles, but also on the employer side.

I want to talk a little bit about that.

Yeah, let's talk about it. Looking forward in terms of how the job market might change, both from the employer and then the potential employee side. On the employer side, what are you seeing?

A lot of people are saying, well, look, large companies are going to have fewer roles. And a bunch of startups that are adopting AI are going to hire a lot of people.

people, is that what you're seeing? What does it mean for the future of finding work in those areas?

Yeah, so I'll tell you a fun fact and then we'll go into that. So when we looked last year at where people were getting hired, there were roles on the platform. Like at any given time on LinkedIn, there's some maybe 14 million roles globally. But what we saw is when people were switching jobs, if they left a role, employers were not backfilling as much.

In the past, hiring was a mix of people being hired into a backfilled role or hired into a new role. And as hiring started to slow in the last year or two, when someone left a company, the employer chose not to backfill as much. They would hire for a new.

role, particularly in AI roles. So if you are working in the AI space as an AI expert, you're building large language models and all things probably people in this forum watching, you know, know how to do with high levels of AI expertise, is what we would call it. They are just in a position of choice because that is one of the fast AI engineer, AI consultant, AI researcher, some of the fastest growing roles on our platform are around AI expertise. That's fantastic.

But if you are in other roles, consultant or, you know, a project manager, it's been a little bit more challenging. And if you left your role, your company often wasn't trying to balance. So that was the past. Now we look forward, what do we do?

What are we expecting? Well, I do think that we're expecting companies to continue to hire a little bit more, but they're also looking for more churn. They're looking for people to move in and out of roles more. If I repeat myself, let me say it one more time. Hiring was very sluggish the last year and a half, and companies would like to accelerate the hiring, but that also means they want to see more turnover.

I see. Now, this is interesting. I should change my occupation to AI economist. It sounds like rather than economist, as we think about it, right? Watch my LinkedIn profile for that.

As we think through the employee side, and let's get really focused. We talked about this. We're near graduation time for our college seniors and many others. People are asking this question of me. They're asking of you. Hey, what do I do with my life? If AI is-

change the economy, if these other macro trends are affecting the economy, what is the right career path for me? What do you see people on the platform doing to advertise their skills or the way they might search for jobs that can give new grads a clue of what they might be undertaking going forward?

Yeah, I think it's really important to acknowledge that it's been materially harder for new grads in the last year or so to find a role. It just is. Partly, again, it's an economy where hiring has been more sluggish than normal, and partly because new grads come in with sort of that bright-eyed enthusiasm that maybe aren't known for as much productivity, forgive me for saying that, but you know, they don't have the expertise yet.

Yeah, yeah. So they're coming in at a time when employers are looking at their talent roles being full. Again, no one has quit. And they're saying, what do I do with this one extra person? Do I really need it right now? They would like to have more churn, but they don't have as much as they like. So I think it's been really challenging for new grads, both last summer, people trying to find summer internships. This year, it is still a challenge. That is not a permanent state. That is just, again, an overhang from all the hiring we had two, three years ago. This is going to settle itself out. It will wash out. There will be better opportunities for new grads.

But how do you position yourself is the question. And what I would say is, one, companies are looking for a lot of agility.

They're looking for people who are agile because they know that they want to bring in AI into their processes and their operations. They don't quite know how, we can talk about that. Sure. And they don't know what it's going to look like because whatever the product is today, whatever the best answer is today, it will be even better, right? In six months. I mean, I've heard people say that like, already AI solutions are 10 times better in the last six months than they were. That's true. You correct me if I've got that wrong, but things are improving so rapidly. So it's really a question of hiring people who are extremely agile, one, willing to learn, and then two, people who are at least have some level of proficiency.

I see. And this is what I wanted to like, I'll close on this point here, but.

One of the things we've seen in the past couple of years in our data has been an incredible increase in demand by employers for AI literacy. What is that? That's just your ability to use AI tools. It's like I am NOT an AI expert in the sense I don't build large language models, but I do know how to use chat GPT. I do use co-pilot and so finding people proficient in using it and willing to kind of learn as those programs get better and iterate and get do more reasoning and get more complex, that's what they want. They don't know what they want for an instate, but they want someone who can be along for the ride and is willing to learn.

So that's not a very helpful answer for new grads. Oh, I'm agile, fantastic.

fantastic. What does that mean? But let me finish here.

What has seemed to work the best for people in times of, if you think about it, in times of a big technological change is to have an ability to sort of lean or over-leverage your human skills.

So as much as it's really important to have like this AI proficiency or this AI literacy, whatever you want to call this, general ability to be able to use AI tools, even if you're not a technical person.

I know this is a technical audience, but I'm talking for the average person.

It's actually even more important to have communication skills.

collaboration skills. One of the fastest growing in-demand skills on the platform by employers is conflict resolution. Wow. So like having tough conversations, resulting conflict, leading teams, setting a vision, collaborating, all of that is some of the fastest growing skills on the platform. And they're rising in tandem with the demand for AI literacy.

So I think they go hand in hand, this need for both human skills and technical skills. And there's a lot of research that supports this too. Research out of Harvard, two separate papers, finds that the returns of soft skills in the economy are increasing. You also have this notion that those who are best at leading human teams are also really good at leading agents. I mean, that's interesting work by David Deming and co-authors.

As you think about that work, it really does support what you're seeing on the platform, which is an interesting angle. When you think about what people are doing on their profiles, how are they demonstrating these skills?

I think what's so amazing about the treasure trove of data that you all have is you can see how people are presenting themselves to prospective employers. And I'd like to know how people show that agility, that neuroplasticity, that AI literacy.

So I'm a big fan in believing that people are irrational, and when I- An economist. So that's a good match. I'm gonna put it out there, I'm a believer. And we actually see it.

So when we think about who is signaling the most that they are AI literate, it's actually folks who are in occupations that we consider most at risk for-

Oh, interesting. Okay. So it's not the AI engineer that needs to really signal It is the person I'd say this but in Communications or who does like records administ medical records administration? folks in marketing are Some of the biggest in terms of uptaking these updates to the profile saying hey, I know how to use the latest Whatever AI tools in my domain to help me increase my efficiency my productivity my innovation and creativity and

That's because they need to signal it because the last thing they want to do is be Fighting over a job either because an AI could already do half of it or because someone else can do the other half.

or half of it with AI. So people are, the fastest uptake in showcasing they have some AI literacy is actually for those occupations that are sitting in what we call the disruptive bucket.

Wow, I mean, so this excites you. Your point has been here that AI is not making sort of a big disruptive dent in the job market today, but people are preparing for that by augmenting their CVs, adding things to their repertoire, and the people who are doing that the most are exactly the people you'd expect who are in the most disrupted or exposed industries, and they're advertising their ability to work with AI and combining that with their soft skills, which makes sense.

Yes, it does make sense, right? And here's another thing, and what we're seeing is, I get asked a lot, like, okay, are you seeing evidence that AI is transforming the workforce?

And I'd say, we see evidence that it is changing it. I don't see evidence that it's like wholesale replacing jobs. What we see is evidence that people who are in roles are rotating from certain tasks to other tasks. So the emphasis on what they're doing in their role is changing because of AI. It's not that their role is going away, it's how they're spending their time in that role that is changing. That's some of the evidence that we're seeing.

And we can see that both through what employers are, when they're job seeking or looking for, and we can see it by how people are advertising what skills on the platform that they think they're best at. And that rotation is happening really quickly. That is fascinating.

It's not elimination so much that we're seeing. I know everybody's like looking for that and maybe we'll see some of that, but the moment we're seeing is task elimination. And this is consistent with the history of technology adoption in markets for a long time, right? We're seeing humans becoming better complements to AI, but not necessarily wholesale substitution of jobs. And look, we will continue to study how that will evolve, but right now just what we're seeing, and this comes straight out of like economic theory in terms of how you think it would work.

I was surprised at how neatly it lined up. I promise you, this is actually what we see in the data, but you know, it's early days, right? These are just a couple of years since generative AI really kind of hit the scene and it takes people a while to react. So I know that things are gonna evolve, but right now what we see is not so much that there is.

massive job elimination. It's more just people are adjusting, desperately trying to up-skill so they can stay relevant, which is a smart thing to do in a time of change. But, you know, it's gonna be a long while before we see the full effects. And people like you and I, our job is to be in the data, to develop the indicators and the forecast, to help people, let the data tell us what's happening, and then to figure out if these things start to change, we'll be able to let people know. You guys have platforms to do that, we have platforms to do that. And this is, I think, the best role for a chief economist. That's what makes our job so fun.

One question I wanna ask you on the career ladders and young people starting up, and then I'd like to go global and talk about some other markets out there, because I think it's interesting. You mentioned that, you know, for some people, last summer was a hard summer to get an internship, and I've seen that.

anecdotally, too, from a bunch of college juniors. You, I think, have a college junior or a high school junior yourself. You know how this is. The internship market was weak. And for those students now coming back from the market, if they don't get that first job, is it hard to then climb to the next rung of the ladder?

I think what a lot of people worry about with AI, but also just disruptions to the job market generally is those junior roles where people are learning the skills. They're getting trained. It's almost an apprenticeship. If you miss your chance on that, how do you get back on the ladder?

Do you see sort of dynamics like that playing out on the platform in terms of how people build their careers? And, you know, what might you say to those folks who are now, maybe they missed that summer internship, but they're trying to get back on that rung of the ladder?

Yeah, definitely hear that all the time. Also had someone looking for, and family looking for an internship last summer.

struggle.

One is in the past, so pre-generative AI's big launch a few years back, we took a look at how grads fared when they graduated into what you might call a slower hiring market.

What we found, generally speaking, was that people catch up. They do start a little bit behind the starting line.

They do have a disadvantage. In the first couple of years of their career, maybe they don't progress as fast, but eventually they catch up.

Five, six years down the road, they're kind of where they would be for peers that had graduated equivalently earlier at a different point in a better market.

So one is like, if you're in a cyclical downswing, don't despair, you will catch up. I personally had friends who graduated from Stanford and one of them literally went to be a sheep herder in Colorado, and his parents were like, what did we just pay for? You finished at Stanford and now you are herding sheep in Colorado, but it all worked out, it all worked out. He gained playing point.

The second thing I would say, it was really bringing it back to AI, is that the other thing we're seeing in the market, and I think this is how AI actually is changing the job market too. I did mention this earlier. It's making things much less linear. So your job opportunity, your career pathway, it's not, obviously it' not the latter it was before.

It is far more organic, it is far more skills-based. And because AI is so powerful, it's gonna allow you to pivot into many more options for roles. Does that make sense?

It does, yeah. So you might come out of school with economics training or data science or, you know, in this case, folks are probably maths and physics and computer science, but you're not limited to that occupational ladder. And you can use AI to compliment you, to let you jump onto a, it's not even a ladder, I think of it like a net or something, but you know, you can jump elsewhere in the market, different industries, different types of occupations, different levels, and what we're seeing, and I'll just.

I'll finish here and let you ask me the next question. What we're seeing is actually even at the C-suite level now versus, say, five years ago, folks who rise to the C-suite roles are coming from a much broader set of experiences, more industries, more functions than they did, say, five years ago. So it's okay to have a broader base.

So going back to early new grads, if the thing you have your heart set on isn't working, go do something else and just get started. And I think that there's no career ladder. It's a net. You can just kind of make your way back over to where you wanna be.

you know, in a couple of years. Because there's not one direct path, and just like you're finding with the CEOs, it's not like everyone who's a CEO had to work in sales, or in finance, or in marketing. You're seeing a broader base of functions that they come out of, and AI potentially, I think implicit in your answer is, AI might help you get up to speed in these new areas more quickly.

And you see this, I mean, so many people on our team are teaching themselves things using ChatGPT, and that is an angle on the job market that we haven't been talking about. And access to knowledge that, you know, you wouldn't have had 15 years ago. If you wanted to rise to be like the CFO, you had like one path, but there might be people who ultimately can rise there, coming from totally different functions, and start to build their knowledge.

I know, and I don't want to, you know, gas dispersions on CFOs are super qualified and built their way up, but I'm just saying there's probably more avenues to the same spot now than there were before because of AI.

Let's go global for a second. We talked to each other beforehand about how much travel we do and the demand across the world for the kinds of insights that you guys are generating, that we're generating.

When you go to other countries, there's a different conversation about the job market and AI, and reflecting on two places, I think, that have become, you know, sort of central in my mind, you know, India and Nigeria. Two countries that are going through rapid demographic transitions, and you think about those parts of the world where they're graduating hundreds of thousands, in some cases, of STEM grads each year.

Their populations are very young.

When they look at AI and the job market, they might have a very different view than you would in the U.S. or Japan or Germany. What are you seeing on the extent you have visibility of that on your platform in some of these global frontier markets related to the themes that we've talked about?

Yeah, so I mean, India's one of the markets that's really key in our LinkedIn data. We see a lot there. And it's just incredibly dynamic. Yeah. Do you know what I mean? And I think it's, if I could say, I think India's just really well-placed because they're graduating. You said so many people every month. In contrast with the U.S., where there's 10,000 people retiring every day. Right. Hello? So you've got a very young, well-educated, and very dynamic, large domestic economy that could really.

there's a lot of potential there right they don't even need to look outside they've got a lot of domestic potential

what I think is going to be interesting and I'd almost like if I could I'd turn the tables on on you to help me think about this is how do we ensure that AI products are relevant in all these different locales yes and how do we make sure that what's going to be right for Nigeria you mentioned is also is will it also work for Indonesia which also has a young and fast growing population but just totally different cultures

and so I think that's going to be the trick for how quickly we see adoption across industries across occupations across

the newest, youngest generation is going to be whether the relevance is there, the accessibility of the digital divide and the relevance. And those are things like, I don't, LinkedIn doesn't have an answer for, but I think are really important to consider.

Yeah, I mean, just on my end, I think, no, and I'll just do it once because I want to ask you more hard questions, but this is an interesting area I've been thinking a lot about. You know, OpenAI has also a mission to benefit all of humanity. And when we think about that and we take it very seriously, it's going to include in large part of the countries that you mentioned. I'm headed to Jakarta, actually, in June.

When you think about these countries and you think about previous technological revolutions, you see tremendous divides in terms of adoption of firms and people. And even when you compare, let's say, the US and the UK during the IT revolution.

A research by Nick Bloom find that, and co-authors, that US firms adopted IT much more quickly and beneficially than UK firms, even though they had access to plausibly the same stack.

And so we're looking for those kind of divides where we see them across the world, trying to make sure that doesn't happen again, because it's our core belief that the populations that adopt AI, the firms that adopt AI, they're going to be able to do what you're talking about on the LinkedIn profile, which is say, look, I have these skills, right? I can do these business models. I can be more productive.

And that's going to be a signal, whether you're an individual employee or a big employer, that you're on the cutting edge and you're out to win. And I think if you see firms adopting AI at higher rates, only let's say in the United States and not in markets like India.

India, Nigeria, Indonesia, around the world, we're going to see that digital divide play out again. And so one of my big jobs when I travel to those countries is to explain, you know, here's what the tools can do. There's a lot of uncertainty about, hey, what is JATCPT even for? And how can it benefit sort of individuals and organizations and governments also is a big part of that.

And when you think about that, then we're hoping to unlock early adopters, sort of use and learning from that. And of course, in the end, customization for the local market to your point. And that's why, you know, one of the big priorities of JATCPT on the engineering side is language accessibility and making sure that we're relevant, like in the top 10 languages in India, for example. So those are the ways we think about it at OpenAI.

And you just made me think of something else too.

So one of the things that we can do in LinkedIn's data, and in your comments you'll see how this relates, is look at company size. And so I think that the hurdles for a large company to sort of bring in AI and start to create training programs for everyone who works there so they can kind of, you know, just roll up their sleeves and get their hands dirty with it and start to learn and play with it, that's very different than if you're a small firm where you don't have that opportunity.

So in a lot of, you know, the countries you mentioned, there are many, many small firms that are gonna also need to adopt AI, either find that AI expertise talent, the AI literacy talent, or just.

bring those systems in-house. How did the, you know, the question I have, and you don't have to answer it, I'm just sort of positing in here, is how did we make sure that they do it too? So, not just large companies, but small.

Yeah, I mean, this is such an important question for economic development more broadly, right? In these countries you're talking about, you have this idea of like a missing middle, where you have lots of small businesses, often you might call them main street businesses, mom and pop businesses, that really, for reasons of lack of access to credit and capital, sort of regulatory barriers, they aren't able to grow into large businesses. And that creates a lot of distortions in the economy.

And if you could have levers to unlock growth for those SMEs in countries like Indonesia, India, Nigeria, around the world.

you could create a lot of opportunity, right? Serving both of the missions actually of LinkedIn and OpenAI.

I think one of the things I've noticed, one of the angles I think that AI can be really useful is one of the most useful things for entrepreneurs trying to grow is mentoring advice from peers, from people who've done it before. But that's actually in pretty short supply. And it's actually hard to get enough humans to provide all that advice.

And there are consulting firms out there that charge way too much money for a small business but can help a large business. What if we found ways to basically provide business intelligence and consulting and advice and coaching and mentoring at scale to small businesses all around the world? To me, that's like an opportunity where the ROI is gonna be really high. And if we can find a playbook to deliver that.

and the tools that we have, I think, are a big part of that. I think that's one way those businesses can grow and start to hire more people.

Yeah, and you know, I 100% agree, and you just said something that I hear all the time lately, which is the ROI. Like, everybody's on the hunt for how do I implement these new AI solutions into my business process, into my operations, and then how do I measure it? And that is sort of the question. It's not a question for you to have to answer, but I think that's one of the things I would say to everybody who's here listening, that's one of the biggest challenges I hear when I talk to business leaders, is trying to understand how do they identify the ROI. They may have like three or four pilots running, and then they just don't know.

We'll see if we have the right case study, the right question.

us work, how we know that we can then scale this across the entire company, even if it's a big company. And I think that's something that I just want to share that. That's what I hear.
I think that's a challenge as well, even for large companies, is figuring out how to measure your ROI on these things.
I mean, yeah, I think so. And this is the parallel pilots.
In general, I think we learned from the lean startup revolution, A-B testing, that people are now much more willing to experiment inside organizations. That's great.
But what do I do after the experiment? And how do I set up metrics that allow me to quickly make a decision about scaling an experiment organization-wide?
Because getting buy-in for that experiment and the metrics that you would track is very different than doing it at an organizational-wide implementation.
And I think that a lot of folks, when they're measuring micro-billing and looking at the micro-billing, they don't have a lot of confidence that they're actually going to scale.
And I think that's a challenge. And I think that's something that I want to share with you.
I've been working on this for five years now. I'm still working on it. I'm still working on it.

productivity, like how much time you might save on an email, at least for the products that we're producing, intelligence to really leverage it to change your business, you're often gonna re-engineer more sophisticated workflows. You're gonna have to do a organizational change kind of blueprint to actually make that happen.

If you're measuring the time, this time it takes someone to write email, you might see some productivity increases, but I think you're not gonna unlock the true value, and that is a much harder thing to demonstrate against a traditional ROI framework, depending on how you're measuring it.

And if you think about spreading it across functions within an organization, every function has their own risk guardrails and metrics. Yes. And so you bring in a new technology, it needs to kind of be meeting all of those different guardrails and metrics, whether you're talking about HR or sales.

sales, or finance, or engineering, and that's where the complexity comes in. Oh my goodness, that last mile of across the different functions, and then across sectors, right? So if you think about finance, or healthcare, or energy, these sectors all have very different characteristics, and so adoption of AI and how it's used is going to be different in all those sectors. And we see that, actually, in our data, too.

Yeah, yeah, on our platform, we actually see that the adoption rates, which we measure in terms of how much talent are you hiring that has AI expertise, or is, you know, we consider AI illiterate, they may have worked in an AI role. And what we see is it's radically different. So I don't know if this comes as a surprise, or no surprise to you, but like, one of the industries that's lagging the most in terms of adoption.

adoption is the education industry. And I think I can understand it to some degree because they're probably more cautious. It's an older institution, right? It doesn't change its model very frequently. But that, even so, we see them growing at really rapid rates in terms of taking on new AI talent. It's just not at the same rate as, say, finance is, professional services, or think healthcare. You know, all of those are like really leaping forward. And it's funny to see education lagging. So there's a big difference between industries globally. And then, of course, a big difference between countries.

For sure. So, you know, India, of course, is one of the leaders. Singapore, a leader. U.S., a leader.

Maybe not so much. Interesting to see this variation. I mean, and if we are not seeing this in the education sector, how will students get the skills they need to be able to demonstrate those skills on LinkedIn? That is the key question as we think through.

Well, Karin, I have one last question for you, I think, given where we're at, and I'll make it a fun one, which is if we were gonna look at your LinkedIn profile, what would be the most surprising thing we'd learn about you? I think if you were to, something that's on my profile that might surprise you, probably not that much. That tells us more about your profile. I'm a secretive person, but what I will say is I'm on the board of a company and it's like in a parallel,

of my career, I am a board member, and I find it incredibly rewarding. Well, we found this to be an incredibly rewarding conversation. I want to thank Karin Kimbrough, the chief economist of LinkedIn, for joining us at OpenAI Forum. We look forward to working with you and continue to talk with you, and thanks for sharing your insights with our members. Thanks for having me. Awesome. Thank you.

All right. Hi.

Hey, everybody.

Good to see you.

This is amazing.

I hope you enjoyed the conversation.

I mean, it's just — it's such an honor to have a conversation like that with another chief economist.

I've learned so much, and I know that Cassandra and I, afterwards, we reflected on a bunch of things.

that we can continue to talk about with Karin and the LinkedIn team, just given how amazing she is, but also what kind of data they have in terms of what they're able to say about the economy. So anyway, hope you enjoyed it, and thanks for being with that. Now, happy to answer questions.

Cassandra, I think you're gonna interview me, or I'm gonna interview you, either way. We'll have a good conversation.

Yeah, we actually had some really good questions come up in the chat during the live stream, so I think we'll start with those, and then if we have some time after, we'll go to the audience.

I think it's awesome. This one was from MJ Ratz, and this is a really excellent question, I think, for you as a parent. So the question is, we wanna prepare our students for this rapidly changing market. What is the best way to prepare these awesome students for the future?

they have lots to worry about, even if it relates to their personal growth. Yeah. No, MJ, thank you for the question. And Cassandra's right.

I've been thinking about this a lot because of my own kids, but also just as a professor. I think most of you know that long time professor have taught MBA students at Duke for a long time. And I think a lot about as young people are coming up and thinking about the job market, like how to prepare them for that.

And talking to my own parents and reflecting on kind of what they knew about the job market and how they tried to advise me. One thing I've realized is like, well, I wish I had more information or I wish it was the good old days. A lot of times they didn't know either, right? Some of the advice you should get from your parents, especially if you're like the son of immigrants, like I was, might be, hey, go to medicine, go to engineering.

engineering and go to these very predictable professions. But those two were actually professions that changed a lot since the time I've come up. So I do think, Cassandra, it's like, there's this element of, you know, for parents like me, I'm always like, I don't know if I know that much, but maybe it was always a sense where we didn't have perfect information on where the market's gonna go.

I do think as a professor, what I try to do is give my students advice, and I do the same for my kids, I don't know, they might not listen, about kind of where does the data show that the job market is expanding and when does it show it's contracting? Try to be data-driven when you're giving that advice, and working with someone like Karin at LinkedIn is a way to do that.

Second thing is, are there general sets of skills that you can develop that'll be flexible and evergreen no matter what the market is? You know, we think about people who...

have a lot of neuroplasticity, the ability to adjust to lots of different situations and learn new things. That's a set of skills or an attribute I think is gonna be really useful as the market changes.

I also think that some of the soft skills like EQ, this is a really fertile area of research, but I feel like these are things that are becoming more important. We talked, I think, before about the study out of Harvard that Dave Deming has done where those of us who are really good at managing human teams are also really good at managing agents.

And that was kind of counterintuitive, although lots of people told me they totally expected that result, which is the best kind of study is the one where half the people say it's intuitive and the other half say it's counterintuitive. But what David Deming found with his co-authors is that those who are best at leading human teams are also the best at leading agentic teams.

And I think that sort of leads me to think that there are going to be some set of software skills that are going to be useful for us to impart to students.

And the last thing I think is critical thinking. You know, critical thinking is something that there's a lot of discussions around the impact of AI on that sort of ability. I think it's just really important that we double down in schools in terms of how to teach that.

And I'm not an expert in how to do that. I just know that in my classes I used to teach, that's probably the most important skill. And I don't think that LLMs or the rise of generative AI are going to absolve us from that being really, really important.

So those are some of the things I try to tell my kids. I also tell them to think about sort of the kinds of problems they want to work on, not necessarily the job all the time. And I know that can be, you know, it's a luxury to have that opportunity.

But I really think the ability to work on the problems you care about with the tools that are available That's like the greatest privilege at work rather than being a specific job or career. Yeah Actually being motivation based instead of being like so yeah interesting

I've also thought about like you know people often say I want to be like this person when I grew up I want to do this job, and that's a really high bar You know and I always thought the better way to think about it was here the problems I want to work on you might you know not be the one who cures cancer or wins a Nobel Prize But if you're the person who's working on You're part of it. You can actually be really satisfied with that. You know versus only having this kind of unrealistic Aspiration to be the person so that's at least how I've been thinking about my kids. Yeah, absolutely awesome

Okay, we're thinking more about the future now, so this is from Abbas

And Abbas asked, what does meaningful human AI collaboration look like at work beyond using it as just a productivity tool?

Oh, you know, it's so interesting. And that's Abbas you said? Okay, hi, boss. Thanks for that question. Super cool.

I think that one of the most important applications of AI, and it's surprising coming from an economist, is emotional support. Because as you know, as I talk about this all the time, I have lots of stories that I regale the team with that they're not sick of hearing about how awesome it is to get decision support, emotional support from Chats with BT.

And there are limits, and there are real challenges with this too. For someone like me, though, when it comes to trying to stick to a program of diet and exercise, or navigating different questions about how to deal with

team members or lead a team, Chatsuite has been really invaluable in terms of asking those questions. And I start to think, gosh, human AI interaction is going beyond just a simple productivity tool, you know, writing a better email or an RFP or, you know, optimizing your work routine. It's like, it can give you the support to get things done in your personal life that you couldn't do otherwise.

And we have to be careful. We have to make sure and, you know, there's widely sort of known results around like model personalities and how we might need to calibrate them more effectively. And we deal with that in open AI all the time. But to me, the ability to provide like a trusted advisor to a lot of key decisions in your life, that to me of us is like the frontier of human AI.

interaction. And it's one that I actually think economists may be undervalue. You think about all the people who just need a little bit of help to get out of the bed in the morning and go to the job and stick with it. There's a lot of people who can be more productive if they have a little support. And some of those things are under-provided in the market. You don't have the humans that you need to do those kinds of things. All people who just need someone to talk to, right?

And I think that, again, we have to be careful, we have to do it the right way. But that, Abbas, is what I think is really interesting about it. As an economist, coming here to OpenAI has made me more interested in studying those topics. And Kassandra knows I'm really interested in research designs about that.

Yeah, absolutely. Okay. This one is from Michael. So, historically, it has been the case that when tech augments labor, productivity...

increases jobs in the aggregate. Where would you see more jobs shift if AI displaces but doesn't replace all human work?

Yeah, it's a great question. I think that's a great comment about history too in terms of looking at previous technological revolutions. I do think that, you know, the places where we can scale human ingenuity are gonna be places where you see a lot more opportunities.

We talked about this earlier today in another context. There's so many places where we just don't have enough humans to do particular jobs. And I talked to a CEO of a leading company recently about the rate limiting factor for him is really not enough engineers to do new product development. And we are able then to use intelligence to really effectively like.

pause the most important problems in your organization and put the maximum amount of intelligence to them, you can unleash limitless potential to develop new products and ideas and commercialize innovations. And so I do feel in R&D-intensive industries, you're going to see a lot of opportunity for that kind of work.

I mean, we often think about the substitution hypothesis. But if you could complement R&D workers and extend what they do, that could create lots of opportunities for other people to build on that work. Think about the way that science develops. It's often cumulative. I'm building on Cassandra's work, and she's building on someone else's. Well, if we can keep those advances going forward, we'll have more people building on them. And I think that's actually a metric where we'll increase the number of people working in particular jobs.

So that's my hope, I think, is a.

big question. I think you're right with the analysis of the general suite of technological history. We're still in the early days, I think, when it comes to AI. Absolutely.

Okay, this one is a bit of a trickier question. I think you can be imaginative with this. So this is from Carlos. So what's the most optimistic scenario for AI in the global workforce that you see and are excited about in the next 10 or so years?

Oh, thank you, Carlos. It's a really creative question. You know, I feel really lucky every day to really enjoy my job, get fulfillment out of it. Like, it's really a privilege. And I know that many people don't have that not, you know, many, I don't know if it's most, but like many people don't have that. And I

feel like if AI can make our jobs more satisfying, that is going to be the best win we could have. So I don't know. Maybe there's a way for me to enjoy my job even more. But the way I wake up looking forward to the job I have, if that's something that we can expand to more people, gosh, that would be a huge win.

Imagine if AI lets you work on the things you really enjoy and maybe jettison the things that are lower margin or lower value to you. That could be a game changer. Now, I know it might not work that neatly. And I'm sure Carlos was polite enough not to ask about the worst case scenarios. But I do think, when I think about the best case scenarios, it's me taking the things I don't enjoy about the job that need to be done and having AI do them, free me up to do the other things I love.

And I think about, if I could accelerate the things I love, do more of it.

that just plug in. I mean, even this is just a small example, but just showing up today that the platform is already ready, speaking to a group of people, I don't have to worry about the tech setup. I don't have to worry about the fact that the, they already saw the video with me and Karin. All these things are taken care of and allows me to just be in the moment in this conversation. That is sort of a great way to think about like a division of labor. And the more we can do that with AI, maybe we can have more satisfying work. Of course, right?

How the technology unfolds and adopted, there's lots of challenges in that. And some of that will be up to us and some of that may happen in ways that we can't completely control. So we need to think carefully about it.

Yeah, absolutely. Okay. This one is, I think maybe our last question.

it taps into your past as a public servant. Oh, actually, there are two. Maybe we'll do this one, actually. Well, I'll do two. I'll talk fast. Ronnie can speak very quickly. Yes, yes. I'll try to be brief. Sorry, folks. Yeah, okay. So this one is from Cesari, and I apologize if I mispronounced that. But the question is, I work in AI adoption and public service, and what types of messages would you use to calm those who worry about job loss due to AI? Thanks for the question. I think the first is empathy. I've talked about this a little bit earlier today, too, in this other context, which is, I do think we need to meet people where they are and be like, look, this anxiety is real.

When I took this opportunity, and I never want to be the person who said, oh, don't worry about that, that's crazy. No, people are worried because they're seeing this new technology that's really useful, but also seeing some of the things that they've been able to do themselves for a long time wondering, oh gosh, if AI can do that, then what's left for me? That's a completely natural feeling.

Sometimes it's not even such a zero-sum mentality. It's more like, oh, AI is doing the things I like to do. For me, writing is a big part of it. We've talked a lot about writing. I enjoy writing. To think that in a world, the time I spend writing is not going to be very valuable anymore if such a world exists, that's depressing for me.

I think first is empathy is really important. Having worked, to your point, in public service for a long time, I think for public servants who are...

you know, by definition, mission driven. I think one of the people I worked with in government, they were there, you know, not for the salary, not for the prestige. They were there because they really cared about the mission. And if you care about the mission, all of a sudden, this new technology is there and you're thinking, oh, the things, the reason I'm here is going to change. That can really create anxiety. So first is empathy and understanding.

I do think, and you know this in the context of your organization better than I, but like, you need to do the hard work, the listening tour and the focus groups and the kind of implementation plans to make sure you bring people along. That's probably something that people know, but it's easy to pass by that and try to speed it up or make radical change happen. I get that.

And then you have to understand and be as specific as you can about like, why are we adopting AI in this?

agency, or in a government context in terms of public service. What are we trying to accomplish? How are we going to track our goals?

The other thing I think is cool about government service is a lot of people are interested in metrics of outcome. A lot of us think about impact. And so if you can articulate those things, you might bring more people along.

And three, you've got to be willing to admit your mistakes. When you get it wrong, or it's not having the lift that you thought it was. And so that takes humility as a leader.

So I think sort of empathy, and then providing some metrics, and then being able to go back and say, hey, did we reach our goals, is one way to start. And AI implementation in the government context is just very complicated anyway. So you're going to need a lot of experience to really get it right.

Yeah, absolutely. Okay. This is a fun question.

Oh, good. Yeah, yeah, yeah. The other ones are...

fun too. But no, this will be even more fun. Okay. So this is like a question of how you got here. This is another one from MJ, but I think it's a good one. So you asked the question to Karin, what's the most interesting part of your LinkedIn to her? And so now he's, or they're spending the question back to you. So how did you get to this spot with AI and like thinking about your pathway here? I would love to hear more.

Yeah. I think if you look at my LinkedIn, what probably is unique, it's just like I've worked across a couple, three, at least three different contexts that are interesting to people. You know, I have this academic background and published papers and thought a lot about like what good research is and done hopefully some good research or my best attempt to do good research. And I really do.

love research, and I never plan to leave that behind. At the same time, I've taken these opportunities to go serve in government and do pretty hands-on applied work, like channeling my research into public service. And there's a handful of people who have done that and done it with the intensity I've done it. And that small group in academia, particularly, there's fewer, let's say, at business schools, where I come from, than policy schools or econ departments.

That's a group, I think, that has an interesting understanding between the intersection of business and public policy. And now you take this third leg of the stool, which I think makes me a little more unique. Maybe the Venn diagram, I don't even know. But it's now three bubbles, where I'm working in an organization that produces products. And not just a product, but one of the most important products of our time.

And that gives me a perspective of connecting the role of business, academia, and policy together in one place. And when you think about AI, it comes from a research foundation. You have to understand the research underpinning what we do. It has been operationalized in a commercial entity, and you have to understand why people are purchasing intelligence, how they're valuing it, what the economic dynamics are going to be.

And then government's going to play a huge role in shaping the trajectory of AI because of the significant impact it's going to have on a communist society. And so I do feel like if I look back at my career path, I would say, oh my gosh, these are all these random things I've done that don't really connect or make sense. And maybe you'd think that if you saw my LinkedIn profile. That's probably the most interesting thing about my LinkedIn profile is how diverse it is.

is. When you put them all together, the three legs of the stool are what I'm sitting on today, with a perspective of the way AI is going to shape the world. So that's what I'm excited about. That's why, hopefully, I'm in the right place, sitting on the right stool at the right time.

Awesome. Thank you so much. Oh, Kassandra, thank you. And thanks, everyone, for being part of the forum. I hope you liked the interview. Send us some feedback. Also, let us know if you want us to interview more people like Karin or in similar positions. I think Kassandra and I will be excited to keep doing this. And I appreciate everyone's time.

Yeah. Thank you all. Take care. Bye.