The Incentive Problem With AI | Dr. Corina Taban
Transcribed automatically by PodView
So the biggest driver of AI impact is organizational. It's not technological and it's not individual mindset and it's not individual behavior. If you want to change behavior, change how you reinforce it. It's as simple as that. If you take one thing from me, it should be this.
Hi, I am your host, Henry Clavel, CEO and founder of Orlando Insight Group. Each episode, we dive into real conversations with technology leaders, innovators, and experts, influencers who are shaping the future of IT, data, and cybersecurity. Our goal is simple, to give you practical insights, real-world strategies, and expert perspectives that you can use to drive smarter decisions and stronger results for your business. Today, I'm joined by Dr. Karina Taban, an organizational behavior researcher, keynote speaker, and CEO of 934 Advisors, where she advises and trains Fortune 500 companies and other leading organizations on human-centered AI adoption.
Before founding 934 Advisors, Karina led strategic partnerships across international markets at Meta and Microsoft. At Microsoft, she managed more than 100 partner relationships and helped expand Microsoft news into new markets. At Meta, she worked with senior executives and cross-functional teams across product engineering, policy, and trust and safety while helping the company navigate complex licensing and regulatory challenges. Karina is also a guest lecturer at Cambridge Judge Business School, where she teaches examiners how to adopt AI within organizations.
She was recognized by McKinsey as a Next Generation Woman Leader, and her research examines leadership, the evolving relationship between employers and employees, and how organizations adapt in rapidly changing environments. In our conversation, we'll explore on why AI initiatives have installed, even when the technology works, how leaders can build trust without ignoring employees, legitimate concerns, and how organizations can embrace AI while preserving human judgment, creativity, collaborations, and motivation.
So Dr. Karina Taran, welcome to the Orlando Insights Podcast. Thank you for having me. Let's get started. Before we discuss artificial intelligence, tell us about your professional journey and what ultimately led you to establish 934 advisors?
Well, my professional journey.
I used to be, as you mentioned, a tech executive. I used to negotiate deals for Microsoft and for Meta. And I had and have a passion for behavior and went back to school to study organizational behavior as a doctorate and more specifically employee-employer relationships or something called psychological contracts. And what led me to 934 is that I've had the luck and opportunity to work for companies that put a very big emphasis on employee centricity, which ultimately led to engaged employees, committed employees and impacted the bottom line.
And I thought that was how things worked in general, but very clearly they don't. Nonetheless, I think they should. And I think that companies can do better. One, because, you know, it's the right thing to do when this is 2026. But that's just my opinion. Companies don't need to listen to me. They can just look at the different bodies of research, which show very clearly that employee satisfaction and well-being is tied to higher productivity, stronger customer outcomes, in terms of satisfaction, in terms of loyalty, lower absenteeism, lower turnover, and a higher overall profitability.
In other words, companies where people are happy to work outperform their peers. So, you know, don't listen to me and my two cents morality rant. Look at, if you look at the research, it will show you that it pays off. So this was something that I was interested in before. But while I was studying, while I was studying psychological contracts for my doctoral thesis, I had a first row seat to the AI change and to the AI transformation. And I saw how these psychological contracts and the relationships between employers and employees are changing.
And that makes a big difference as well, because if you look at most companies today, they've made massive investments in AI, but they aren't necessarily seeing any gains or any measurable return on investment. And oftentimes it's because, you know, people and processes are more important than technology. If the organization isn't moving, having access to tools doesn't change anything. You need people to use them and people to use them in meaningful ways. So that's the big hairy problem that I am solving working with companies.
companies. That's the big hairy problem that led me to build 934 today. Yeah, I think you should mention that. Yeah, the recent headlines have discussed that there's issues how to measure productivity or AI using AI. What is a workflow? What is a product? And then token maxing, is that a reliable way of measuring productivity for developers or people using AI to get most out of the AI usage in their current lineup? So, yeah, those are very interesting conversations to have. 934, where did that name come from?
Was that the time where you had the epiphany? Oh, I mean, Starman Company. me and 924 but that's a good that's a good point so in in organizational well and psychology in general you have this idea of liminal spaces so in in the normal world the limit a liminal space is a place of transition right like a waiting room like a platform like a corridor and in psychology it's a place where you transition from place a to place b which normally entails a lot of growth that they're very uncomfortable places because we don't know exactly what's going to happen and where we're going but usually if we sit in these liminal spaces enough there's a lot of magic that can happen because we figured it out we figure it out We find new options, new ways of doing things, and it forces us to get creative in a way with our options.
934 actually comes from Harry Potter, from platform nine and three quarters, which is a famous liminal space in literature that takes you from the normal world to the world of magic. Wow, that's very inspirational, very original. This is a first. Thank you for sharing.
You have held leadership roles at major technology companies, including Meta and Microsoft. What did those experiences teach you about how large organizations handle change?
Well, you are in luck, my friend, because that gives me the chance to give you a good example and a bad example. So the good example for change is Microsoft. When Satya Nadella took charge of the company, Microsoft was in a place where you had a lot of toxic management, toxic procedures, and it also was a period where, you know, Microsoft was slowly sliding off of the map. When Satya took charge, he wanted to turn the organization from a know-it-all organization to a learn-it-all organization. And he also put a lot of emphasis on leadership, on learning, on learning from each other, on how leadership treats people, how managers treat people, on diversity, on inclusion, all those things.
And you can really see the difference in the Microsoft culture from the before Satya era to the after Satya era. And that was obviously a long process to put everything in place. But it was a new vision of things. And even if you look at Microsoft's stock price, it grew a lot in that period. So Microsoft really was put back on the map, at least in my opinion, by Satya's leadership.
Meta, however, used to have a very good company culture. And I mean internally for employees.
But with all the changes, right, that went then through layoffs, it still lays off people constantly. People kind of leave employees right now in this sort of limbo where they don't know if they're going to have a job tomorrow or not. Apparently, people just fly off the internal chat. And you can very much see how that influences. I mean, I can very much see because I'm still in contact with former colleagues. But that really does something to how people engage and how they commit to the work because they don't know if there's a future for them there.
And they don't necessarily abide by this new sort of vision that Zuckerberg is now putting in place. So if you, I'm going to, I mean, if you ask me for a prediction, I think that in the medium term to long term, Meta is going to suffer a lot because of all these people policies that are being put in place. And I mean, I've got skin in the game. I still have a lot of stock, so I really don't want that to happen. But, yeah, I'm curious to see where this leads, and I don't think it's to good things and good places.
Yeah, I've seen the decline. Even though they returned with LLMs and their own models, not with news coming out, if they don't get their house in order, they'll lose a lot of talent, especially if they're offered more incentives to leave and if the culture promotes, oh, you can be replaceable. So it is a culture that usually doesn't last long. The replaceable piece is interesting because everybody is replaceable That not ultimately the issue Even I would say if you have a sort of if people are afraid of losing their jobs they try to make themselves as irreplaceable as possible and have a tendency to hoard information, not share with others, and then you start working in silos and you're not going in the same direction.
It's not even a question of that. People are wondering if they're going to have a job or not. And if that's the only thing that's on your mind, you'd have no interest in building amazing products and services. Good point. The fear of hoarding. Yeah, that's not conducive to a productive company. Was there a particular experience that convinced you that organizational transformation is fundamentally a people challenge and not simply a technology challenge? I have been convinced of this for a long time but it's maybe because I am not somebody who's working in technology and you know just by but by pure ontology I believe that people are the ones who lead transformation and that systems, at least social systems, rely on people.
And organizations are a social system, ultimately. So I've been quite convinced of this. I've been going on and on and on about it. Now there's more and more reports, interviews, information, research that's coming out that's supporting this. And even if you look at all the different major technologies from the past, you didn't see any productivity gains, significant productivity gains when the technology was adopted. You saw it when the organization adapted to it. Let me give you an example. If you look at as far as the 19th century, the factories were powered by steam engines.
And fundamentally how that worked is that you had a big steam engine as the source of power. And then all the machines had to be tied to it through shafts and bolts. And that was the way it worked. Now, you had the electric motor that was invented towards the end of the 19th century. And it replaced the steam engine, but initially, factories just replaced the steam engine with an electric motor, meaning that everything else worked the same. It didn't change anything fundamentally. It took them about 40 years.
It took us, humanity, about 40 years to say, hey, you know what? Maybe not everything needs to be tied to this one source of power. Maybe we can have small electric motors in every machine and we can have a different layout and we can have a different workflow. And that's when you started seeing big, important productivity changes in how factories actually worked. And this also goes back to PCs. When PCs first arrived, companies started buying, buying, buying, but there was nothing specifically to do, right?
Because the workflows weren't adapted yet. So you had this economist called Robert Solow, I think. Yeah, exactly. Yeah, who said, you see computers everywhere except in the productivity measures. And it took another 15 to 20 years to actually see a difference in how productivity worked work because the whole workflow was adapted and the people adapted and the organizations adapted and they learned all these new skills so that's the case for ai as well right adopting the technology and we've seen this because everybody's now using ai it's like the magic word everybody needs it everybody's afraid of missing out so everybody's using ai in one one way or another but while we While we see improvements in personal productivity measures, there are no big differences for companies in general.
And that's because everything else didn't adapt yet, and everything else is related to organizations and people. Yeah, that makes sense. If they take, not many companies have a perfect workflow or they're not using it to the best of their abilities to get the technology used. So right now they're early stages and that productivity has not, or that investment has not shown that growth yet. but still a lot of money on data centers and other aspects of it. It's still betting that it will, in the future, see an AI workforce or maybe agents being responsible for those workflows, but we'll see.
How has your definition of effective leadership changed throughout your career, Corita? My definition of effective leadership, I think for a long time in the past, I thought that an effective leader was somebody who had all the answers. And it was a sort of spotlight effect where you need to have the answers. You need to know what to do and tell everybody what to do. You need to look smart. So it was very egocentric.
Today, my definition is completely different and it is very other focused in the sense that as a leader, I believe today, you are only there to offer support. So you're only as good a leader as the team that you support. it's completely turned towards how can I make sure that they have everything they need in order to grow and to have all the supports that they need?
I guess I could give you the most obvious example where when people is, when one of the workers are promoted to leadership, they have to learn a different skill.
Have you seen, is that the case where if a company doesn't provide training, they are not prepared to transition from individual contributor to a leadership position? And if so, So how common is that? Yeah. And I mean, let's call it to management. Very often when you go to first line, you are a manager. You are not necessarily a leader. But yes, absolutely. So let's take this example. when you are a salesperson and you do amazing work at sales, you're a top performer. And obviously they need to promote somebody, so you're going to get promoted.
But then in order to be a manager, it's not at all the same skill sets that you need to use as you would if you continue to do sales. So you get to a point where you need to use completely different skills just because now you've been promoted and you have no other ways of advancing. And most companies don't even provide training or ongoing training or proper training. So they just throw you in a role where you potentially have the skills naturally, but that's quite rare. Or you don't have the skills at all and you don't exactly know what you're doing.
So you're just improvising based on what you've seen or what you think you should do.
In a team, engagement is dependent on direct manager for about 70%. so what you do next is very important and i think a lot of companies are losing a lot of engagement from their ics because not necessarily because they're promoting the wrong people into management but in a way if it's the only way to progress potentially you know people skills are not your zone of genius but also because they're not training them because many people might be better managers if they learned it as a skill, as an ongoing skill, and they were able to practice it afterwards.
So I think I'm getting to the same point as you were alluding. You advanced your position that you potentially aren't good for because it's not the same skill set. And then additionally, you very oftentimes don't get the right training in order to develop those skills that you are missing.
Yeah, it's usually the case where either you are grouping the next generation or you just want someone to love leadership but don't have a direction. Usually that's more of a ladder, but that makes sense. It is a ladder. And I mean, to be fair, don't we all go to our level of incompetence? Because you're very good at something, you get promoted, and then you stop being promoted when you don't excel performances. So finally, at the end of the day, companies are just a set of people who are stuck at the levels where they're not competent enough to proceed.
I have stories, but moving on to understanding AI transformations, Karina. Many organizations say that they are undergoing AI transformations. What does genuine AI transformation look like to you? What does genuine AI transformation look like to me? Genuine transformation looks like to me, you don't just adopt the technology as we're seeing today. you have one a very clear vision of where you are going you have all your leadership who is aligned on where this is going and are pulling in the same direction and not in in 20 different directions you have people who have a common narrative about what the future means for them and with the company if they play the game right.
And they are on board with this transformation. They have access to training. They have a court differently, according to their different needs of course You have new incentive structures because today using AI in meaningful ways is not incentivized at all And given that you have all these things, normally you should have people who are innovating and who are reorganizing the work and redesigning workflows meaningfully. and again you don't need to have everybody redesigning the work you can have some people who are redesigning the work and who are excelling and who are incentivized to do it and to disclose it and then everybody else can just you know follow their lead yeah Yeah.
When an AI initiative fails, how often is the underlying problem of cultural organization rather than technical?
I think in most cases it's cultural or organizational. It's not. So the biggest driver of AI impact is organizational. It's not technological and it's not individual mindset and it's not individual behavior. So I'd say it's a case by case basis, but in most cases it probably fails because of organizational reasons. It is new. And obviously, in my current, let's say, workplace, they're still taking courses in cloud skills. They're still learning the process. So does those help advance or does it need a company-wide solutions in order for them to be ready for AI initiatives?
um so you mean if it's a if it's individual if you can proceed individually i mean you have you have individuals who are much further along than their organizations and when that's the case if they're not treated well they'll potentially leave when they when they get the chance so i This needs to happen and this needs to move organization wide, but not everybody needs to be an expert. Everybody needs to be literate and the systems to be put in place in order to promote people to work differently, to learn differently, to experiment.
But that's not necessarily happening today because today people mostly hold their same engagements as before. They have the same targets, the same KPIs, everything stays the same. But additionally, they are being asked to experiment with AI, learn AI, upskill significantly and reinvent their work in meaningful ways. But nothing is being given to them in return by their company. So there is no incentive. there is no incentive for them to actually move and change in meaningful ways. So I'd say having the license is not enough, and we're seeing that.
You can have people who do some amazing stuff and who reinvent meaningfully, but if there's two of them in an organization of 5,000, then you aren't going to see meaningful ROI regardless. Yeah, I've seen companies where they just adopted technologies, but they have no, I guess, agenda or idea how to train or how to teach. so they are left to experiment and then see what sticks usually that and then later within the process they'll start adopting the learnings and the trainings and academies in order to understand what is out there in order to provide value to the organization that's perfect that's my job they should call me yeah they should they should uh uh we should i provide your your your info so they can contact you right away uh why do companies invest heavily in ai but struggle to get employees to adopt it uh is it psychological or is it just uh uh their idea of how using it so okay so that's for the second part Now, why do companies invest massively in AI is because they think that it's going to generate massive gains because everybody was talking about it and they were afraid of missing the bandwagon.
So very oftentimes it was from FOMO one. And then the second one that I really like is instead of leaving from an actual need that they have and something that they're actually measuring, they read something or they see something and they go like, oh, this is cool. And then they're going to experiment with it. But obviously, if it's not tied to the actual business or there's no actual need behind it, it's not going to generate meaningful value. So companies are adopting AI. I think that, I mean, they're investing in AI.
I think that's cool. I think that's important. But I think they should get clear on what they're investing in, why, or what's the expected result, what's the need behind it, and be much more intentional about those investments. Now, your second part of your question, why are employees not adopting or adapting to AI?
Here's an example that I give quite often because I think it illustrates things quite well. If you have Mary, who's an analyst, and Mary uses AI in a meaningful way, and she finds a way to do a report that used to take her three days in three hours. Great. Now, what does Mary do? She has three options. One, she goes and she tells her manager. And you can betcha that by noon, she will have 20 more reports on her desk because now she has a ton of time. Number two is she tells no one. She does the report in three hours and she chills for two and a half days.
And then she has her third option, which is maybe she tells the whole company and maybe the company just decides it doesn't need as many marries anymore. Which would you choose? the the optimal choice would be not to tell anybody and and do the today's exactly in in today's incentive system that becomes the rational choice and people always always act within their incentive structures and the only incentive today to to use ai in a meaningful way is getting more work uh getting less stability or both that doesn't sound like very good incentive does it so there's no there is no reason for for people to adopt the eye in meaningful ways i think there was this they were scared at the beginning and now they're like yo you know whatever we'll see what we'll we'll see what happens but just to work hard on it especially given that they have all their other engagements, same KPIs and targets, same everything.
It doesn't necessarily make sense. And you have this quote that I like, which says, you can't get a man to understand something when his salary depends on him not understanding it.
That's hilarious.
Interesting. Where did the data come from? where the data came from none other than the quote that was oh the quote is upton i just used it in a presentation so let me find it for you upton yeah sinclair upton sinclair oh upton sinclair who wrote the book on oil industry what interesting uh yeah he pretty much understood the business and how to extract profits during the 1920s. So yeah, he's a perfect... Smart guy. Yeah, smart guy. But yeah, going back, yeah, it goes back to
the organization as in how they reward, I guess, productivity. And government, if you do well, they penalize you and they do more work and in the workplace yeah if you finish your work early you get more work or you get someone else who is not productive as the first person and they take up more on their work so it's usually very structured for the person to incentivize to look out for their own for their own interest. So that makes sense why the adoption is so low or not inclusive to improve the organization.
If you want to change behavior, change how you reinforce it. It's as simple as that. If you take one thing from me, it should be this.
Moving on to leadership, ownership, and trust. should responsibility for AI adoption belong to technology team, HR, business leadership, or partnership among them? There's no one size fits all. I mean, I think the one, I believe the one thing that needs to happen before you decide ownership is getting your leaders aligned. there's only one in four AI users who says that leadership is consistently and clearly aligned. That's not a lot, right? And if you have five horses pulling in five different directions, you won't get as far and as fast as five horses pulling in the same direction.
So I'd say once everybody in leadership is on board with where this actually is going, then you can divide who knows who does what and who owns what because things become so much clearer as they trickle down. Yeah, and especially if the organization is new to the technology or they never had an AI agent or an adoption of this tool within their workflows, it makes sense. I don't think it matters. I think wherever you are in the journey, if you want to move in the right direction, the organization needs to be aligned on where this is going.
And you would be surprised how rarely that actually happens. Why is it rare?
One, it goes to because they don't get in the room to discuss it, honestly. I mean, if you get everybody in a two-day seminar, and these are the kinds of seminars I run, you get everybody into the room, explain why it's so important with research and data, why it's so important to get everybody on the same page. And then if everybody has even if they don agree on everything right having the same vision and the same direction towards where you going and then making commitments in a room full of your peers that this is what this is what going to happen that changes the entire dynamic can you give us an example where you believe let's say someone's taking your seminar and company A is people that want to adopt AI adoption and responsibilities, what are the characteristics you see that works for, let's say, for a company and what doesn't work when you provide the training?
So we just want to understand, I guess, the signals that, okay, we're taking this training, but it's not going to stick because they're not ready. So have you seen that within your workshops and your governance structure? I've seen that. So I do two types of interventions. One will be just training. So I don't necessarily follow them to see afterwards what they actually implement or not. But I also do it as advisory where it goes through a whole diagnostics phase around the different bottlenecks in the organization at different levels through surveys and interviews, you know, a mix of qual and quanta depending on the engagement.
And then based on that, I actually do the training for the leadership team and we decide on KPIs that we're going to follow and see if things work and things move. and not everything is going to move that's the thing you can you know say okay this is amazing when you're in the room and then business happens to you and some things get postponed but that doesn't change the fact that you'll still have some of those kpis that move and that brings you further than where you were before got it got it yeah it is yeah it really depends on who is taking it and the individual that will make a difference oh it does but I mean look that's this is the chance for the leadership team to have somebody from outside coming to say okay look this is the research this is what needs to happen this is the problem this is and one I think it's The issue with heads of people and CHROs repeating that again and again and again internally, it gets easier to say, yeah, okay, whatever.
You know, we have business. We have our stuff. So just having somebody from the outside coming in and forcing you to think about these things and to discuss these things with your peers, by definition, makes the wheel advance a little bit further. But again, as you say, not everything will change, but that's okay. because we've started thinking about it and you have some things that change and then that trickles down into further effects. Yeah, that's why you mentioned surveys. So you have metrics, KPIs, you track that score.
And over time, your company will have those annual meetings and they'll discuss the employee's score, where they are at and how can they improve more. and then you come in again to track it again.
Human judgment and accountability. Who should be accountable when an employee follows an AI recommendation that leads to a harmful or incorrect decision?
It, again, depends on what the indications were. If the indications were here's your AI, do what the AI says, then it's your leadership's team or whoever told you that this is your ai and you should follow it blindly because you can't you can't hold an ai responsible accountable can you so if that was your if that was the instruction follow this blindly then the person who gave the instruction should be accountable but if it's here's ai you can use it to develop to to do your work and isn't that the whole point of having human judgment and here's where that's where it becomes it becomes complicated right because if you have people you can always have them accountable for what they do who do you hold accountable if something happens because of an agent's actions that's a question that's again it's not necessarily my my area of expertise but that's a question that needs to be defined at an organization level from the very beginning so that when shit goes to fan you know not necessarily to look for the answers to have somebody to blame but you need somebody who's accountable and you need somebody who has ownership so as to to solve the problem.
Yeah, if you saw the recent news, Scott Besson said the companies should be held liable for their agents, their technologies. So it is, yeah, he said it directly. No room for to misinterpret his view on companies who are responsible for that process. Looking ahead, and we're going to close, which human capabilities will become more valuable as AI handles a greater share of analytical and administrative? Everything that's related to nuance, context, accountability, taste, connection, communication.
That's all going to stay human. At least that's my prediction. We'll see where this takes us. good good uh and what does a successful first year of ai adoption look like from the perspective of the business and employee what does a successful first year look like first year of ai adoption look like from the perspective of both the business and its employees or something to aspire to. I mean, is anybody still in their first year? No, I believe it's not perfect. We call it shadow IT. Some are using their own AI.
Well, you have structures that provide enterprise AI transformations, but for the most part, it's just a process where people take their technologies and build their own process. Yeah. I mean, is it the sort of spying on people to see what they're doing?
Yeah, and then that's the ethical question. I think Meta has that technology, I believe. And they are tracking their employees' keystrokes, logins, and processes, which I found interesting. I think another layer of issues that employees might find troublesome. Yeah, I mean, I had already jumped the boat. and I don't think that would be legal to do in France. So there's always been like some restrictions and things that they could do in the US and couldn't necessarily do in certain European places.
But look, I mean, is it, again, you have the question of ethics and morality, But you also have the question of if employee well-being and satisfaction is tied to the bottom line and to profitability as a strong body of research indicates, as many companies see. Even if you look at, so even if you analyze different glass doors metrics based on satisfaction, you can actually correlate that to how companies perform. Yeah, I mean, would you be happy with that? I would not be. I mean, I would not be okay with that.
I think the EU is more restrictive. I don't think they'll allow that. But here in America, I don't know if you work here in the US. We're at will. We do a nine to five. And yeah, whatever policies they adopt, they will use, unlike in Germany or any parts of the EU continent. But I mean, you as an employee, if that happened, if your company said, we want to check your keystrokes and we want to check what you're doing at every time of, I mean, you think every adult would be like, even children would probably be like, what are you talking about?
so i mean i'm just saying i don't i don't i think that beyond beyond the ethical implications uh i think you can also look at what that does to to engagement and commitment and it probably works for a while because it pays well and people need a job right now the market isn't that good But like most such changes and adaptations, it usually comes with a sort of disruption and a fall in gains and productivity, but then it tends to go up again. So what will happen to your top talent, to your talent?
And honestly, Mata can tell you from being there has excellent talent. Really people who work very, very well, who are super smart. Yeah, I mean, I'm assuming they will probably eventually leave. Maybe they're also waiting to be laid off with a good severance package. I don't see how this is something that is going to last and I don't see how this is not going to impact the company in a negative way. Interesting. Very interesting perspective. To wrap up, what are you working on through 934 Advisors and where can our audience learn more about you and where can they find you?
I am very present on LinkedIn. You can find me there. You can find a lot of my work there. And then, you know, everything I built that's public is usually announced on there. So you can find all the things. And then for 934, you have a site, 934advisors.com and a LinkedIn page. but follow us wherever you want and wherever you can. Awesome. Well, Dr. Karina Todd, thank you so much for this informative session. This wraps up another episode of the Orlando Insights Podcast. To catch more conversations like with Karina Todd, please visit our website, orlandainsightspodcast.com.
Please like, subscribe, and comment, and we'll look forward to the next episode. So take care. Bye.
you
Apple Podcasts
Castbox
Castro
Goodpods
Moon FM
Overcast
Pocket Casts
Podbean
Podcast Addict
Podcast Republic
Spotify