Loading summary
Transcript49 lines
- [00:08]
A
Welcome to the AI to Roi podcast. I am Peter Buchanan. I am the managing partner of New Plan.
- [00:15]
B
And I'm Ray Reich. I'm the founder and CEO of BenchMarket.
- [00:19]
A
And this week we're going to talk about what we call the great AI Jobs war. AI has created tremendous upheaval in the workplace. Let's get started. Ray, how should we think about the great AI Jobs work?
- [00:37]
B
Well, one of the things I've always been told, and I believe is that history is the best predictor of the future. To me, how AI is going to impact jobs is going to be an accelerated version of what we've seen over the last 100, 150 years, starting with the Industrial Revolution and the mechanization and manufacturing in both the US and the UK. So that was what, like in the mid-1800s to late 1800s?
- [01:05]
A
You bet.
- [01:06]
B
Right? And then the famous Eli Whitney and his creation of the cotton gin transformed how we actually conduct agriculture. And then one of my favorites, and as you know, I, I used to lead the manufacturing automotive industry at our division of GE when we first met. It's Frederick Taylor's time and motion studies that he did in partnership with Henry Ford and led to the assembly line concept for the Model T. And then lastly, we had the digital revolution kind of in the 1980s and 1990s, where we disrupted some enterprise workflows through automation. And as much of those impacted civilization, I think what AI is going to do is going to be faster and with greater magnitude than any of the things I just mentioned.
- [01:57]
A
All right, so let's get specific. CEOs have totally drunk the Kool Aid for AI and they're promising their boards and investors major productivity gains. They're saying they're, they're happening right now, they're around the corner. But the reality is that AI driven productivity gains are pretty uneven and they're elusive for most companies. You feel like you're getting there and you don't quite make it. It's like a game of Mother May I? So the only thing that's really certain is that AI is absolutely crushing the hopes of early career employees, which we're going to talk about in a little while in more detail. Yet. CEOs are just lethal. So, Ray, how are CEOs portraying the progress of AI and what's the actual reality?
- [02:48]
B
Well, it was really appropriate that you said they're drinking the Kool Aid because aid starts with AI, so they had to be drinking something that had AI in it. Right. So the first thing that you can tell that when something's a real trend, is. Is it being talked about on earnings calls? So, and Peter, you did some great research on this. 306 of the S&P 500 companies mentioned AI in their Q3 earning calls. By Q4, AI had become the number one topic appearing in 47% of those earnings calls. And things like agentic AI and digital labor, just those terms from Your research increased 779% year over year. But that's the hope, the vision, the reality is only about 10% of companies are reporting meaningful AI to ROI benefits, with a staggering 56% saying they're getting little to nothing out of it. At least little to nothing that they can measure and show in their income statements. But AI is being used for a lot of labor productivity impact. And by the way, I, I do this other podcast called the Metrics Brothers with Dave Kong, and we just talked about how US labor productivity is being impacted by AI. But here's some examples that have been credited or blamed on AI. Amazon, 30,000 jobs have been cut since October, partially attributed to AI. Dow Chemical, 4,500 layoffs explicitly linked to AI. Even Pinterest, here it is a social media digital first company. 15% workforce reduction tied to the efficiencies from AI and Forster. They're calling this AI washing, attributing financially motivated cuts to AI implementations that have just begun and really aren't delivering the ROI yet.
- [04:49]
A
Wow, that's. That's kind of harsh. So let's dive a little deeper here. A few companies are getting tremendous value out of AI and they produce some cold, hard metrics. Give me a couple of examples there where there have been workforce impacts that are really significant, but the results have been absolutely spectacular.
- [05:15]
B
Yeah, and I think it was in the AI newsletter that we published on Monday, February 9, that you had uncovered this amazing fact. I'd like you to add to this. And it was about Klarna. Now, I remember klarna from about 18 months ago because they came out and said, we just use AI to code a copy. Not a copy, that's a bad word. Our own CRM to displace Salesforce. And everyone said you're never going to be able to use AI to code. But here's some McLarna results. They've reduced their organization from 5,000 employees to 3,000. They've doubled their revenue in that time frame. And Peter, I think they've actually increased their revenue per employee. Can you share some of the details on how much they've increased revenue per employee?
- [06:06]
A
So they were, when they started this journey, they were $575,000 per employee, which is actually really good. They had 5,000 employees. Now they have 3,000 employees. They didn't lay them off, they just had a lot of attrition and they've hit $1 million per employee, which is sort of the magic number for an AI native company today. But the other thing is they have retooled internally their workflow processes. They now say that 96% of their employees use AI every day to do their job.
- [06:42]
B
Yeah, and what I was amazed when I dug into that number. You just shared that million dollar of revenue per employee, which, by the way, and I think we're going to talk about this later. I think it's going to be an ultimate metric to track that shows the real productivity impact of AI. But in Q3.25, which is the latest full quarter that they have released earnings on, it was 1.1 million. That 1 million is the trailing 12 months. So they're up to 1.1. So I found that incredible. But then, you know, so that's one industry. And then there was an insurance company called Hiscox and they actually deployed Microsoft 365 copilot. And we're not here today to debate whether Copilot's gonna make it or not, but they have actually have 3,000 employees across 14 countries. And they say that on average they have used AI to decrease the, the claims processing process from an average of 60 minutes to 10 minutes. And the underwriting process that used to take three days and multiple peoples now is taking three minutes. If you translate that into the number of claims processors or policy underwriters are going to be needed in the future, you're going to see real impact on those jobs.
- [07:58]
A
But you're also possibly going to say, I, I don't necessarily want to have fewer underwriters. I mean, I don't want to have a mass reduction. What I really want to do is use that free cash flow coming out of there to go enhance my sales processes and get. Now I've got a war chest to go sell more, raise my revenue more. Like I have cash that I didn't have before to deploy into my business. It's fantastic.
- [08:23]
B
Peter, that's a great strategy because growth is typically rewarded more than increasing profits. Of course, investors want both. In fact, in the software industry, a point of revenue growth is worth almost 2.5 times more than a point of earnings growth. So you're right. But at the end of the day, if I'm an investor, I want to make sure that that earning per share and my earning per share multiple is increasing. And I want you to get up from a combination of increased profitability and increased revenue.
- [08:57]
A
Right. So there are also some AI tools that are really creating tremendous productivity gains. That's right. So give me some examples there.
- [09:07]
B
Well, and I know, I mean, you're using a lot of AI tools, not necessarily from coding, but you can't turn on CNBC or open up your LinkedIn or Twitter without hearing things like Claude, code cursor, of course, GitHub, copilot, lovable Replit. So recent research says that these automated AI coding tools are increasing developer task completion and it's really compressing the time by 51 to 81%. So it's all about velocity and getting more feature function out there faster. It's hard to believe this data, but Another report says 41% of all code being written is being written with AI coding tools.
- [09:57]
A
Yeah. And that's accelerating incredibly rapidly.
- [10:01]
B
You and I were talking earlier and we were talking about a really big company and I think it was bank of America that they've used it for customer service. Do you have some numbers there for us, Peter?
- [10:11]
A
So bank of America, this is a different set of tools and coding, but they've used customer service agents. They built an agent called Erica, And Erica handled 2 billion customer service interactions last year. They resolved 98% of them within 44 seconds without going to a human. Or you look at Lyft, they have customer service things that come out of the rides that they give to people that could be labor intensive, but they've been able to reduce resolution times using AI by 87%. Gartner actually predicts that AI will autonomously resolve 80% of common customer service issues by 2029. Of course, these are big companies with a lot of money and a lot of time and budget to devote to this. But the tools will get better and better and it's going to rol downhill to customer service organizations in every size company.
- [11:17]
B
You know, there was one other one that I know we had been talking about, and that was regarding HR automation. And before I go to this one, I was amazed at the news yesterday that the CEO of Workday had just announced laying off 400 employees. And the next day it was announced that he was leaving. And Anil Bushiri, who was the co founder of PeopleSoft at Workday is taken back over. So I'm not sure we're always rewarding the reduction of employees, Peter.
- [11:44]
A
No, I don't think so, but there are a lot of HR processes that you shouldn't need a human to take care of them. And HR should be a more strategic and less administrative function. And so HR tools are a prime candidate, well, for disruption.
- [12:07]
B
Well, Bershon Associates, which is one of the leading analysts for the Human Capital Management, they're predicting a 36% reduction in HR staff by the end of this year. And I think some people are saying that 61% of human resource leaders are planning to deploy generative AI in 2026. That's up from 19% just two years ago. So I think every function is going to get impacted. Peter?
- [12:35]
A
Oh, no question. We've got all these enterprise AI assistants, ChatGPT, Claude, they're being deployed all over the place. What are the reports there on productivity, Ray, for just these products that all of us seem to use every day.
- [12:53]
B
I don't know if I have any data right at the tip of my brain here. I do know that it's really impacted the number of jobs being hired for. So is it okay if I pivot to that?
- [13:06]
A
Sure. So let's first go to the macro view here. And here's where it gets kind of complicated because the net job numbers don't look terrible. And there are a lot of studies that are very positive about the effect of AI on jobs. The World Economic Forum projects that AI will create 170 million new jobs by 2030 and only displace 92 million people. And that's 78 million person gain in another org. Right. How do they know that's really. And there's a couple of times the Yale Budget Lab says AI just about as disruptive as the dot com boom.
- [13:56]
B
Move along.
- [13:57]
A
Nothing to see here. But these studies, they hide some harsh realities. So let's dial in on one of them here. That's just in the extreme. You and I both have young adult children in their. What's up? What's up with their generation in AI?
- [14:18]
B
Well, you know, this really does hit way too close to home because my third and youngest child is graduating with a Data science and statistics degree with a 4.0 GPA. So he's been going through the job application process and entry level for data science. Now the criteria is two to three years experience. Wait a minute, that doesn't sound like it's entry level. And I'll give you an example. So entry level job positions, just job postings. They're down 35% in the last three years. Entry level hires for IT type positions used to account for about 25% of all hires just back in 2023, Peter. Last year that was down to 7%. I still remember, you know, and my wife was in investment banking. And that entry level position of an associate who did all the Excel modeling and worked 80 hours a week for three hours to become an investment banker. Those positions fell by 24%. College graduate unemployment exceeds the national average for the first time in almost 50 years. So. And then you've got people who are trying to create the future with AI, such as Dario, the CEO of Anthropic. And he's saying that AI may eliminate up to 50% of entry level white collar jobs in five years. Man, if any of this comes true, it is a brave and scary new world.
- [15:57]
A
Yeah, well, you know, young people are very creative and it might not necessarily benefit those of us who are older. So now that's bad for young people starting out, right? Obviously it's terrible. It's a structural problem. What is the structural problem here, Ray? And why are these entry level workers not needed or perceived to be not needed?
- [16:23]
B
Yeah, this reminds me, my father, he worked in a factory and he came to me in 1978 and said, computers, getting computers ready. And I said, why? He goes, because all the factory tools that I used to run and my team runs are now being controlled by computers. I just spent a day getting trade, cnc, DNC machines. But what happened was it wasn't automation that put manufacturing jobs at risk, it was actually outsourcing. So now I'm thinking about AI is almost another type of labor arbitrage. It's going to do a lot of that drunk work, digital grunt work, those tasks that junior employees are trained on, creating Excel models, writing emails for campaigns, et cetera. So as a result, you know, 35% of those entry level postings are requiring two to three years of experience, as I mentioned with my son. And then I still remember this, and I think it was over two years ago, and Jensen Huang, the CEO of Nvidia, said, don't go into computer programming, go into mechanical and physical automation jobs. And I was like, oh, that's really interesting. I guess he's thinking AI enabled robotics are going to replace workers. And here's what's happening though, to build out the vision. McKenzie says that by 2030 we're going to need 130,000 more trained electricians, 240,000, almost a quarter million new construction workers, and another 150,000 construction supervisors. So I still remember talking about, you know, all these boomers are going to be Retiring, they're going to sell their H vac and plumbing businesses. This is another reason that blue collar, I think, is the gold collar job over the next 20 years.
- [18:26]
A
Right. And so what's happened to salaries in those businesses for those workers?
- [18:32]
B
You know, you might have more of the data right off the tip, but if I'm a macroeconomic and I'm thinking about supply and demand, with all that demand, salaries have to be going up dramatically. Peter, you have some data there?
- [18:45]
A
I do. So if these positions are on AI related jobs, like constructing a data center or a complex, say power plant that supports a data center, those jobs, the salaries for those jobs have doubled. The other thing is that tells me that blue collar trade schools are going to be booming too, because we have to produce a lot of them. You have to have those certifications and skills.
- [19:18]
B
Not to say anything bad, but maybe the University of Phoenix can go from being a white collar manufacturing degree to a blue collar manufacturing degree.
- [19:29]
A
Yeah. Devry is back, baby. Yeah. So let's talk about what the AI infused workforce is going to look like in 2030. So what do you think we're going to see in terms of workforce composition, particularly on the white collar side, what the senior mid junior landscape look like?
- [19:56]
B
Well, there's a short and then a midterm impact over the short term to implement all of these agentic AI initiatives to provide some human reinforcement and oversight and I call it AI explainability. You're going to need even more senior workers, those people who have the experience, subject matter expertise, the ability to use and apply good judgment to oversee the digital AI workforce. And I think this will grow. In fact, I think McKinsey projects like 3.8 million of these higher wage jobs will be needed. And then when those people who are today in their 40s and 50s begin to retire, it's going to create that midterm knowledge crisis. Because you don't have as many early career people. They're not gaining the experience. So who's going to be that wise old overlord to monitor all these AI agents? So at the same time, I think mid level workers in the short term are going to be squeezed out. And that's not so good for those people who are in their late 30s, 40s, who are in the peak earning power of their lives. And if I'm a CFO and just looking at spreadsheets, it's like, hey, why do I need that $140,000 a year middle manager where I can just have a lot of great AI agents and a couple People who are my digital AI overseers. So I think the mid level workers are going to be squeezed out and then as I already have said, in the near term, entry level positions face the deepest cuts. And honestly, I have no clue on what's going to happen to early career jobs over the next three to five years.
- [21:47]
A
So my theory here is that the cuts will be here now, and then companies are going to start to realize that they can actually use AI to make these workers, these entry level workers, a lot better, a lot faster. So you can get mid level skills out of cheaper employees faster because you're creating, you're just basically accelerating the work they actually do. But of course, most companies haven't figured this out. They're going the other direction. So what are some other sort of workforce trends here? There's, there's in different parts of companies.
- [22:29]
B
Well, Gartner says that almost no IT work will be done by humans without AI assistance. So it's a little bit of not taking a position on how it impacts IT workers. It just says IT workers will always have that AI assistant beside them. Other people are saying 75% of white collar jobs will be augmented by AI systems. Honestly, I'd like to see those 25% of white collar jobs that aren't going to be augmented.
- [22:55]
A
That's hard to believe, right?
- [22:57]
B
And then McKinsey says that workers in that lowest wage quintile are 10 to 14x more likely to need to change occupations than.
- [23:06]
A
Wow, that's tough. Yeah. All right, so let's dial it in. Now. We have basically conflicting data. We've been through 20 plus minutes of this podcast. We have AI's employment impact that depends on variables that we just can't predict. It's adoption, speed, regulatory responses, business model, innovation, human adaptability. We have optimists, and the optimists say that technology has consistently created more jobs than it's destroyed. 60% of the workers that are here today are in occupations that didn't exist in 1940. And the pessimists say this time is different. We're automating intelligence itself, not just physical labor. Then the realists are in between and they say, oh, our best guess is that two thirds of jobs will experience partial automation. Task level change, not wholesale replacement. Ray, if I'm a CEO and I have my C suite executive sitting around me, what should I take away from our discussion of the great AI jobs war?
- [24:21]
B
Well, as you know, Peter, my biggest hypothesis of how to see what the ultimate impact of AI is on labor productivity is going to be revenue per employee. I would immediately tell C EOs and CFOs for every AI initiative pilot proof of concept. When you're looking at what's the investment to do the initial proof of concept and then to deploy it, measure it, and ask the executive who's leading the charge for that particular initiative, how much revenue is it going to increase, how much cost is it going to decrease, and how is this going to impact the number of employees that you need in your organization? I won't share any names, but I personally know that a Fortune 100 CEO who has over 200,000 employees believes he can reduce that in half over the next five years through AI. So know how these AI initiatives are going to turn into ROI at the same time when you're hiring? One of the things I would bake in every job description for a white collar employee is AI aptitude, AI curiosity and applicable AI skill skills. And by the way, I'm going to flip that. As I talk to young people, especially my youngest son and all of his friends, I'm like, every month let's have you do a project that's using a different AI tool. So he's been using Copilot, he's now doing some stuff with cloud code. He's did some stuff in clay for go to market for me. Because it's those people who actually have shown curiosity and gained some experience with AI tools that I think are going to be best positioned to get those early career jobs. Hell of a lot better position than a person who just got their $250,000 liberal arts degree and they've never done anything with AI tools. Right?
- [26:34]
A
Right. For sure.
- [26:35]
B
The other thing going back to the CEO, I do think you need to place your bet on augmentation tools to help your employees be more productive getting that revenue per employee up. And then the last thing I would say is if you think you have a talent advantage or talent is a big part of your unfair competitive advantage, invest in that talent. But unlike when I was at GE where they invested in me to have all these great applicable skills, right? And I went through the Salesforce development program there and then their executive management development program, and it made me a better manager, Tomorrow is going to be those companies who put together amazing AI training programs for each stage. From increasing your personal productivity, how do you do that as a marketer or as a designer or as a finance person to middle management, how do you leverage AgentIC AI to have your organization be more productive to a senior executive, how do you invest and leverage AI to drive higher revenue per employee, increase earnings per share, and ultimately increase enterprise valuation for your company.
- [27:57]
A
Wow, it's been a content rich episode here. Ray, any final words before we go?
- [28:03]
B
No, I just. You know, Peter, you and I have lived through a lot of the digitization era, right? For sure, every white collar job was assisted with automation, whether that was PC, SaaS, mobile technology. But do you know what's amazing, Peter? US labor productivity in the last 20 years with trillions of dollars of investment in SAS, cloud and mobile twos, kind of was only about 2.1% a year, the same as it was 50 years ago. I think today we as executives with all the potential of AI, we owe it to our shareholders and to our employees to try to maximize the highest return on this AI technology. And measuring every step along the way.
- [28:57]
A
Yeah, I think you're absolutely right. All right, I think it's a wrap for AI to ROI for today. Thanks for listening and we'll be back soon.
- [29:05]
B
Okay, thank you, Peter. This was a fun one.
- [29:08]
A
Yeah, you bet.