The following transcript was generated using automated transcription software for the accessibility and convenience of our audience. While we strive for accuracy, the automated process may introduce errors, omissions, or misinterpretations. This transcript is intended as a helpful companion to the original audio and should not be considered a verbatim record. For the most accurate representation, please refer to the audio recording.
Michael Dunne: I’m Michael Dunne. It’s arguable that the Venn diagram of employee feelings about artificial intelligence, or AI, right now contains a good deal of intersection between fear and inevitability: fear that AI might take their jobs, but inevitability about AI’s continued use and probable ubiquity. But what about the employees who aren’t going to be replaced by AI and will instead have to work with it to better the company and, hopefully, their employment? Are business leaders having the right conversations with employees before they implement AI? Today on the show, you’ll hear from a national business consultant and bestselling author who works with company leaders to better explain AI and prepare employees for its use. Sometimes these are difficult conversations, but according to our guest, they’re absolutely vital. Dr. Gleb Tsipursky is CEO of Disaster Avoidance Experts and a behavioral scientist, with a new book out about AI. Doctor, welcome to the program, and tell us about the new book that you have coming out.
Gleb Tsipursky: Happy to. It’s a peer-reviewed book out with Georgetown University Press called “The Psychology of AI Adoption at Work: From Resistance to Results.” It argues that companies are not adopting AI in the right way, which is why we have a new study from MIT showing that 95% of AI pilots are failing when they’re scaled up, and that’s a big problem. We can talk about why. That’s what my book is about. But the key is that they’re approaching AI as a typical technology adoption tool, but it’s not. With AI, it’s much less about the technology. It’s much more about the psychology.
Dunne: And I want to get to that, because I would imagine that for business leaders and business owners, a word that might best be associated with AI could be something like innovation and efficiency. But it seems like at this moment, right now, for the general public, the word most associated with AI is fear. As somebody who studies the behavior of the human mind, how can we overcome this fear? And I’ll even pull that string further: It isn’t just necessarily about job replacement. I think there’s a lot of fear about what AI can do to us as a society.
Tsipursky: You’re absolutely right, Michael. And here’s the thing: AI is the first technology that challenges our identity. It is the first technology to do so. When you think about typical technology adoption, let’s say a CRM or an ERP, so customer relationship management or enterprise resource planning software, these are typical database adoptions. They are top-down, standardized adoption processes, and it’s a hassle to learn them. There’s a lot of difficulty learning them and changing your work habits, but they’re not challenging people’s identities. You just have to work within a new system instead of your various spreadsheets. You have to work within a CRM. With AI, it’s so different. AI is much easier to learn, but it’s much more threatening to people’s identities because it can do pretty much anything they can do with a computer, and so it can replace things that they hold dear. If you think about work, so many things are key to who we are as professionals: the ability to write a great sales outreach email, to create a great marketing brief, to do in-depth financial analysis, to create a great image, even to make great videos. These are things that AI is increasingly good at, and we get a sense of meaning and purpose from them, at least those of us who don’t just work 9-to-5, for whom our jobs are more meaningful. So it’s a threat to our sense of meaning in a way that previous technology was not. It’s also a threat to our jobs, because if AI can do this work, then it’s going to be a real threat to our economic livelihood. We’re already seeing these impacts, and we can talk about that. But leaders are approaching AI adoption as though it’s a typical technology adoption. They’re approaching it as though it’s a CRM and leaving it to IT. IT is leading this process, and it’s a top-down approach that’s a little bit ham-fisted. They don’t really know how humans think. HR is not involved in the process, and so this is a big problem, because the real sources of resistance are emotional, fundamentally about economic anxiety and identity threat, and companies are not addressing that.
Dunne: So, OK, how do they do that? What’s a better way for a business leader who wants to introduce AI to roll it out? Or maybe, even before rolling it out, are there steps to be taken, again, to satisfy what the human needs are?
Tsipursky: Yeah, let’s address those two big issues, the identity threat and the economic anxiety, because those are the fundamental problems that we’re seeing. To address the identity threat, when I work with companies, what I advise them to do is not focus on the things that are most profitable to automate. That’s very counterintuitive, because the most profitable things to automate are the marketing brief and the creation of images. However, those are the ones that create the biggest identity threat to people, because they’re more core to their jobs. So what we do instead is run a survey. One area of the survey asks: What are the aspects of your job that you hate? What would you like to get off your plate? Another asks: What are the blockers to the things that you love? Where do you have to gather data from a variety of sources to get what you actually need to do your tax prep, or your image, or your PowerPoint, or whatever you love to do? We then build skeleton tools that will help them automate these, and the people who are actually going to use AI customize them to their own needs, because AI is very flexible and people can use natural language programming, just talking to it, to customize the tools. So we’re giving two types of motivation. One is: Don’t you want to get these blockers off your plate? Don’t you want to get the things that you hate off your plate? And once people build these tools themselves, based on the skeleton examples that we give them, they’re emotionally invested in them. So you have that emotional investment. That’s one half of the picture, the identity threat, and it’s very important. The other half is the economic anxiety. What you need to understand is that right now, people are worried that if the company gains effectiveness and efficiency from AI, it’ll fire people, and they’re worried that by using AI themselves, they’re training their own replacement. You need to address both of those elements. First, as a company, you need to make a commitment: If we get an economic efficiency from AI, we will not be cutting staff. We will focus on growth. We will focus on seizing market share away from other companies that are using AI less effectively, and there is definitely evidence that companies are doing that. There was a recent study out of Stanford showing that companies that are more effective at AI adoption are growing their top-line revenue 9% faster, but they’re also growing their headcount 6% faster. So they are being more productive. They are using fewer people to do more work, but they are still hiring people because they are seizing market share away from companies that are less effective. That’s one part. The other part of addressing economic anxiety is to tell people: You’re worried that you’re training your own replacement, but the reality is, we’ll keep around people who are productive, efficient and effective. You can work harder, or you can use AI tools. Whatever way you want to be more productive and efficient, you can choose how to do that. You can work harder, or you can work smarter, and most people will choose to work smarter, not harder, by using AI tools. So you need to flip the script for them. You need to show them that they’ll be more valuable by using AI tools, not that they’re training their own replacement. And even if something happens to them inside the company, say they have a falling out with their boss, they will be more valuable outside the company, because right now the job market is very favorable toward those with domain expertise and AI expertise. So let’s say you’re in claims or underwriting, or in sales, marketing or any other area. If you have domain expertise plus AI expertise, you’re much more valuable than if you just have domain expertise.
Dunne: People learn new skills. As a behavioral scientist, is part of the calculus to look back at history and say, “Look what happened in these industries. People adapted and evolved”? Is that part of this?
Tsipursky: That message resonates with people who are in a position to learn new things and adapt to new industries. If you’ve been working in the same field for 20 years and it’s difficult for you to learn new things, that’s the challenge. Look at, say, factory workers in the ’80s. You could say that we have more jobs now than there were in the ’80s, and the factory workers could retrain for those jobs. But plenty of factory workers don’t have very good lifelong learning skills to retrain for new jobs. There was a letter signed by 16 Nobel laureates in the past few weeks saying that in the next several years, we’ll have massive job displacement. AI is being adopted at a much faster rate than previous technologies. So not only did previous technologies leave many sectors and many people behind, but they didn’t integrate nearly as fast as AI is integrating. It may be the case that eventually we’ll have enough jobs to replace all the current jobs that are lost. It may not be. We have never had a technology that replaces human intelligence, which is what AI does. So maybe what we’ll have is a lot more physical jobs in the real world, if AI isn’t as good at robotics. It might be, or it might not be. We don’t know yet.
Dunne: Are there characteristics of employees who are going to be better at innovating along with AI, at adopting AI and seeing it not as a threat but, as you pointed out, as something that can help them work smarter?
Tsipursky: I think the biggest test is going to be: Are people actually using it effectively? We talked about how AI is different from, let’s say, adopting new customer relationship management software or other tools like that. The key with AI is that anyone can build AI tools. It’s just a matter of talking to it, of expressing what you want to do and what you want to get, and then AI will adapt to your needs. It’s great. It’s wonderful. It’s super, super helpful and super, super beneficial. But you need to see whether people will use it after you address the psychological problems that really block AI usage. There’s a recent Pew Research survey showing that 52% of respondents are more anxious than excited about AI. Only 9% are more excited than anxious, and the rest are equally excited and anxious. If you don’t address that anxiety, which is a combination of economic anxiety and identity threat, then you won’t know who’s actually good at using AI tools. Some people might be great at using AI tools once they get over the emotional blocker. They might be better than your current AI users, who are in that small minority of 9%. The key is getting people to that stage. You can lead a horse to water, but you can’t make it drink. You need to motivate it to drink. You need to make it want to drink. And that’s what we’re talking about here.
Dunne: Are there a lot of misconceptions about AI that have bubbled up through our society and the general public? As someone who’s an expert in behavioral science, are there things that we as a society, as business leaders, as political leaders, as members of the media, should be doing better to explain AI, perhaps to help people get over that fear of economic replacement and the other fears you talked about?
Tsipursky: Yeah, this is a great example. Right now I’m sitting in Niagara Falls, on the Canadian side. This morning I did a presentation to a group of executives on AI adoption, and as I was talking about this, one of them said, “Oh, you know, I get it now. We’re a bookkeeping company. We were presenting our bookkeepers with very cool AI tools that automate a lot of the bookkeeping process: accounts receivable, accounts payable, all of this stuff. And they keep not adopting them, and we didn’t understand why. Well, now I understand why. By showing them how cool the AI tools are and how much they can do of what these accountants are currently doing, they’re feeling, ‘Oh my gosh, this is what I do. This is my identity. This is crucial to who I am, what I feel proud about.’” That’s the identity threat, and of course there’s also the economic anxiety. Some of them don’t feel that much economic anxiety because they’re senior and well established, but they feel an identity threat. Some of the more junior ones especially are feeling more of the economic anxiety, because AI really can replace a lot of the tasks that junior accountants and bookkeepers are doing. So that is the problem. When leaders roll out all the cool tools, all the cool AI uses, it’s because they’re in the 9% excited camp, not the 52% anxious camp. They’re thinking about where they are. They’re not being empathetic to where their audience, their employees, are, and so they’re not understanding the economic anxiety and the identity threat.
Dunne: What if that fear, wherever it comes from, is rightly placed? What I mean by that is, are there times when employees would be right to say, “We don’t want AI because we think it’s actually going to hurt both us and the company as well”? Are there times when AI is not the solution, from what you’ve been able to glean?
Tsipursky: I haven’t seen situations where AI would actually hurt the company, just being frank. AI can do pretty much anything that you can do with a computer. Now, you can certainly use AI badly, and I’ve seen cases of AI sabotage, where people engage in malicious compliance. They say, “Well, you’re forcing us to use an AI tool. Fine, I’ll use AI.” Then they don’t check the output, the output is bad, and they send it downstream. But that’s a human problem. That’s called AI workslop. It’s not because the AI is bad; it’s because the human is bad. That definitely happens, but not because the AI is bad. So the problem is the human. Realistically, what we need to understand is that the AI is not going to be the problem. AI is just a tool, and it can do pretty much anything that you can do with a computer. So for the company, use AI. It’s always beneficial when you use it correctly, and when you use it correctly, it’s going to be faster and just as good. Pretty much any process that you can automate, if you give it the right feedback, it can do in a way that’s very high quality, and it can automate something like 90% of what you do on a computer. So that’s about the use of AI. Now, in terms of the impact on human beings, of course, if you’re in customer support, realistically, it’s going to require fewer people, because there’s only going to be a limited amount of that work. Even if you’re growing the company, AI is going to take over enough functions of customer support that you’ll need fewer people. The same goes for marketing or translation. Lots of these areas will need fewer people. So the key is: Are you going to be one of the people I talked about earlier, who uses AI effectively and successfully, and are you going to stay in that function? If you want to keep doing marketing, you need to be better than other marketers at using AI, building AI agents and managing them effectively, or you need to switch to a different field. And again, using AI is going to make it much easier for you to switch to a different field. Those are the kinds of options that people are facing in the current reality.
Dunne: OK, Doctor, my last question for you, and it sort of plays off your book title, the “From Resistance to Results” part. However you’d like to tell us about this, and you don’t need to name specific companies, have you personally seen and helped a company transition from resistance to AI to getting great results?
Tsipursky: Sure. I’ll tell you about a company in an industry that’s not typically associated with AI: roofing. With the roofers, we worked with them to create a sales assistant. One of the reasons roofers don’t get sales is this: When there’s damage, let’s say from a hurricane or hail, they go look at the roof and give the customer a quote. Then the customer has to deal with their insurance company, and of course, the customer doesn’t have experience dealing with an insurance company. So they might get a complete denial of coverage, or they might only get a few shingles replaced for $2,000 or something, when the roofer wants the whole roof replaced for $20,000. So, working with the roofers on the ground and with independent adjusters, we created a tool that essentially holds the homeowner’s hand as they go through the process of applying for coverage, to get as much of the roof replaced as possible, ideally the whole roof. It creates a complete information packet that tells the homeowner exactly what they will need to do and how the roofer will collaborate with them and support them as they go through the whole process of applying to the insurance company. The roofer just puts in the homeowner’s insurance company information, plus some pictures of the roof, and creates that packet. That makes the roofer much more likely to win the job from the customer, and much more likely to get a much bigger sale. As a result, within six months of implementing the tool, the company has 30% higher sales, and each sale is about 25% higher. So that’s an example of a very successful rollout. Previously, people were resistant to employing AI. They didn’t see how it was going to benefit them. Now it’s benefiting them, and they’re getting great results from it.
Dunne: He is Dr. Gleb Tsipursky. He is the CEO of Disaster Avoidance Experts, and he is also the author of the forthcoming book “The Psychology of AI Adoption at Work: From Resistance to Results.” Doctor, thank you so much for taking time to talk with us.
Tsipursky: Oh, thank you so much for inviting me, Michael. It was a pleasure.
Dunne: That’s the show for today. All episodes of Oregon on the Record are available as a podcast at klcc.org. Tomorrow on the show, we’ll learn more about the backbone of AI, data centers, and hear some truly surprising statistics about data centers in Oregon. I’m Michael Dunne, host of Oregon on the Record. Thanks for listening.
