The Decibel - The first wave of AI layoffs

Episode Date: September 3, 2025

Artificial intelligence in the workplace is here to stay. But what does that mean for the job market? While many Canadian companies are embracing the technology, in a few notable cases, this has led t...o mass layoffs. In other workplaces, it has meant a revolution in workflow, company culture and hiring practices.The Globe’s Joe Castaldo, who covers AI, and business reporter Sean Silcoff, spoke to executives who are welcoming AI in workplaces. They join us to talk about what this all means for businesses, jobs, and the future of work.Questions? Comments? Ideas? Email us at thedecibel@globeandmail.com

Transcript
Discussion (0)
Starting point is 00:00:00 Earlier this year, the Vancouver Software Company, Clue Labs, Inc., was looking to fill a job after an employee left. Then, their CEO, Jason Smith, had a different idea. So the company was, you know, writing up the job description about the post-the-job, when Jason, you know, had this realization that maybe they didn't have to hire somebody at all. Maybe they could build like an AI workflow to do this job. That's Joe Costaldo. He covers AI for the globe. Sometime later at an all-hands meeting, Jason told the company that up to half of them could be laid off because he needed to reinvent the company for AI. In the end, about 85 people, which is 40% of the company left. Some of them took buyouts, and some of them were laid off.
Starting point is 00:01:00 We've been hearing warnings about how artificial intelligence will disrupt the job market for years. And while mass layoffs are rare, more companies are starting to think about how AI can change the way they work. Today, Joe and the Globe's technology reporter Sean Silcoff are here. They'll tell us why executives are embracing AI, what it means for employees and how likely we are to see mass layoffs because of it.
Starting point is 00:01:32 I'm Allie Graham, guest hosting the decibel from the Globe and Mail. Joe, Sean, thanks so much for joining us. Thanks for having us. Thank you. So, Sean, I'm going to start with you. So these conversations about how AI is affecting the job market have been happening for years now. And before we get into mass layoffs that we're beginning to see, Sean, could you walk us through what led up to this moment? Like, what kinds of discussions have companies been having about AI and how have they evolved?
Starting point is 00:02:05 Well, you're right. AI has been a topic of discussion in the corporate world. I would say for the better part of the past decade, but things really started to change after ChatGPT came on the scene in late 2022. to we start to see some companies tinkering with AI, seeing how they could deploy it within their companies to, you know, save money. But these were really, I would say proof of concepts or experiments. Joe and I really started to notice quite a shift this year, though, in our conversations with CEOs of companies. You're now to the point where I would say there's the carrot and the stick approach. The carrot approach is at all hands meetings or, you know,
Starting point is 00:02:47 corporate getaways. People are being brought up on stage and featured, hey, here's what people in the company are doing with AI to save money, use less resources. You should try doing that yourself. And then there's more of the stick approach, which is the memos that have been going out. We've seen things like this from the CEOs of Shopify and some of the companies we spoke to, which is, hey, we're now an AI first company in terms of thinking about how we can cut costs. You need to be AI proficient. You need to be using AI. and if you want more resources, you want to hire people, you need to prove to us that you can't get AI to do the job first. And if you're not proficient in AI, look out because we'll hire someone who is to replace you.
Starting point is 00:03:32 And Joe, do we know why some of these tech companies, like Clue, are doing layoffs now? Like, do we have a sense of what's factored into the decision there? Yeah, with Clue specifically, you know, there are a few things at play. One, you know, Jason Smith, the CEO there, like he'd just been seeing how some AI tools, you know, around coding and some other things were getting better. But also he was really worried about the competition. Like, as Sean mentioned, like there are startups that can use some AI tools grow really fast without having to hire as many people, without having to raise as much money. And that's a big competitive threat to companies like Clue that were founded before Generative AI really took off. and have some legacy structures in place.
Starting point is 00:04:19 So he felt really threatened from a competitive point of view and decided, you know, I'd rather do something drastic now than not have a choice later on. So he talks about this as like refounding the company, starting again AI first, doing as much with AI as possible and then hiring when necessary. And there was another component to it as well where some companies like Amazon, the CEO said earlier this year
Starting point is 00:04:45 that, you know, corporate head count could shrink because of generative AI. And that could mean periodic layoffs every quarter. And Jason at Klu didn't want to go to that approach because that would kind of destroy morale, right, every quarter having layoffs. So he thought it was better to just rip the Band-Aid off and do it all at once. And so at Klu is the idea that AI tools or agents or what have you, they're going to be doing work as well as humans or, as well as their employees, or what's the sense there? Well, they're still working through it, right? Like, employees at the company are having to, in some cases, take on more work because
Starting point is 00:05:25 there's fewer people while building out sort of these AI workflows. So in the example of the job that they didn't fill, the person writing content, the CEO is quite clear that it's not going to be the same level of quality, but it doesn't necessarily have to be. The way he phrased it is AI today, its capabilities are the same as like an average or below average worker. And like by definition, there's a lot of average people in the world. So not every company is announcing mass layoffs, but it seems like AI is generally impacting the workforce. Joe, you spoke to a number of executives about this.
Starting point is 00:06:05 Are they planning to replace some of their employees with AI? How are they thinking about this? Yeah, Clue is not the norm today. Though I did find it really interesting that after Clue did this, Jason, the CEO there, said that a bunch of CEOs got in touch with him to like ask about how and why he did this. And like we've been considering something similar. And they really wanted to know like how he decided how deep to cut and like the timing of it. So we'll see.
Starting point is 00:06:35 But generally CEOs are more thinking along the lines of, well, how many people do I need? need to hire. Do I need as many people in these certain roles anymore? And so it's more about hiring, potentially slowing down or like, you know, just stopping for certain areas. But also considering like what other roles do I need? What kinds of people do I need? So it's not, you know, mass layoffs. It's a big change potentially in terms of how they hire, who they hire, and who they need. Sean, do we know what types of jobs or positions are? getting cut in favor of using these AI tools or agents, like are any sticking out in particular for replacement? Certainly, AI is being used across a number of departments. It doesn't always
Starting point is 00:07:22 result in job losses yet, but making some people more efficient or enabling companies to not hire as quickly. It's also, of course, leading to job losses. The areas we're seeing it most commonly are in programming, code writing. You know, instead of having three people or five people, you can do the job with one working with tools like GitHub copilot, customer service for sure with the deployment of chatbots, marketing. You can now launch a marketing campaign with far fewer people and using AI tools. And certainly we're seeing legal and HR departments affected as well.
Starting point is 00:08:01 You know, for example, in the legal world, I think AI is getting better at reading volumes of information and summarizing things. A lot of law firms will tell you that's a good thing because that work is drudgery and it's given to sort of people at the lower end of the organization to do. So, you know, machines are proving to be very good at answering basic questions, summarizing information. So I think a lot of those types of rules are going away. It seems like there's a premium on judgment. It's certainly been a feeling around AI for several years. In fact, there's a few University of Toronto Prophs who've written about this,
Starting point is 00:08:43 that the jobs that are going to continue to be done by humans and will be least affected are those that require good judgment. What that means is probably widely open to interpretation, but it means that the humans who are still going to have jobs are going to have to be smarter, more intuitive, and know what to do with a nicely summarized information that AI can provide for them. Could you give us a sense more generally, though, of what this shift has looked like for employees? Employees at some of these companies are learning that they really have no choice but to become
Starting point is 00:09:16 proficient at using the AI tools at their disposal. I'll give you one example, which may sound extreme, but we'll probably become the norm for many tech companies in particular. So Geotab is one of the largest private technology companies from Canada. It provides technology that helps fleet operators track the performance of their vehicles. It's based in Oakville. So we spoke to the CEO, Neil Koss, and he said he's starting to ratchet up the temperature when it comes to AI. So when people come up to him and say, you know, geez, are we going to be replaced by AI?
Starting point is 00:09:49 He says, nope, no, but you will be replaced by people who do use AI. What that means at Geotab is that every employee is getting an AI. score. I think we're going to see a lot more of this. Not everywhere, not every company across the board, but it's going to become more common. I'll give you an example of some of the things Geotab is using AI for, you know, things like writing customer proposals or coding for hiring processes, reviewing legal documents. It's even exploring AI agents to handle video creation and repetitive tasks in HR. So this is the norm of Geotab. It's a company with 2400 people. Joe, what is this all meant for junior-level employees?
Starting point is 00:10:29 Yeah, there's a lot of discussion right now about how AI is taking jobs from entry-level workers. And, you know, in Canada, the evidence for that right now is pretty slim to none, to be clear. But it is a legitimate concern. We did speak to one tech company that said, you know, we're not hiring entry-level coders anymore. Like, we just don't need to. There's no demand for it because our existing employees are just getting a lot more done. Like there was a startup, too, that said something similar. Like, you know, maybe in the past the startup would, you know, hire some interns, right?
Starting point is 00:11:06 Some cheap interns to get stuff done. But the CEO was saying, no, we're going to hire somebody experienced because they can get more done with AI and it'll just be better quality. So, you know, anecdotally, yeah, there's some impact. And there's this larger question about, like, generally, I think companies and society still value entry-level workers and recognize there is a need to train the next generation, right? So the question is, well, like, how does AI change that? How are younger employees going to learn? What are they going to do? So in law, there's a lot of grunt work that gets passed to articling students or junior lawyers.
Starting point is 00:11:44 But that's part of the process of how they learn, right, and can later do the hire. value intellectual work. So if a senior lawyer can do the grunt work, then what is a junior lawyer doing? How do they learn? Like they can't immediately start arguing a landmark case in court, but like someday we're going to need them to. So how does the training process change? And there's not a clear answer to that yet, but it is definitely something, you know, the law firms and other companies are thinking about. We'll be right back. So it sounds like it's too early to know how these restructuring efforts are working out at the companies we're seeing some layoffs or AI integration at. But there have been some recent reports that cast out on this idea more generally that AI improves productivity.
Starting point is 00:12:41 And the most prominent reports have been by MIT and consulting giant McKinsey. Joe, can you tell us what they found? Yeah, they generally found that despite all of the money companies are spending on AI, they're not seeing a financial return. Okay. And that's honestly not surprising. Other studies have found the same thing. And I think there's a lot of nuance to these studies as well. And I guess to summarize why companies are not seeing a return, the answer seems to be, well, they're just doing it wrong.
Starting point is 00:13:16 which is kind of funny coming from, say, McKinsey, which is a consultant that you can hire to tell you how to do AI. Just ask us. Yeah. Yeah. But some of the reasons are, you know, like in the MIT study, for example, companies are using AI for the wrong things. Like they're not looking at AI for boring back office administrative automation where you can get a return. Like if you're looking at AI for sales and marketing, it's hard. harder to draw a bottom line impact from that.
Starting point is 00:13:50 They also found that, like, employees are using things like chat GPT all the time on the down low to help with their jobs and finding a benefit that way. But when it comes to a company trying to be like, hey, this is our AI solution, we're going to integrate it with all our other software and tools and data. It doesn't work well. Not that the technology doesn't work well, the integration part of it is hard. it's more of a learning issue than there being something fundamentally wrong with the technology. And I, you know, I spoke to somebody who's like fairly critical of generative AI about these studies.
Starting point is 00:14:27 And, you know, he was saying, well, yeah, it's really early. So to expect, you know, a return at this point, that was always optimistic. Okay. Right? Because some of the claims were so hyperbolic about what generative AI can do. I think companies and the AI developers set expectations way too high. So, you know, one of the notable findings in the MIT survey was about 95% of AI pilots failing, which is a huge chunk. At the same time, maybe it's not all that surprising because, again, this technology is new and there's going to be a lot of experimentation with it to find out what works.
Starting point is 00:15:05 But also at a time when there's just so much AI hype, valuations for companies are so high because of AI, you know, this finding is worrying at one level. Is there any concern among CEOs or executives that you've spoken to? I mean, is there any feeling that this AI boom could be a bubble? Yeah, I mean, there's been AI bubble talk pretty consistently since ChatGPT came out, right? And every so often there's some new studies or some bits of news that, you know, reignite the bubble talk. And certainly there could be a valuation bubble, right? Tech stocks are riding high on expectations that AI is going to deliver private company
Starting point is 00:15:51 valuations are huge, like, you know, open AI is reportedly worth hundreds of billions of dollars with basically no profit to back that up, right? That's, and we're all waiting for, you know, AI to deliver these massive productivity gains that we've been promised. So there definitely could be a valuation bubble. That's a separate question from whether or not the tech. technology is useful, like generative AI is proving to be useful, but, you know, is it going to be useful enough to justify those valuations?
Starting point is 00:16:24 And how soon does that come before investors really get spooked and, you know, there are financial consequences, not just in the stock market, but also in the real economy as well, because like there's a huge data center boom right now, lots of data centers are being constructed around the world for AI. And that has like real economy impacts as well. Yeah, if I could just jump in with something as well. I mean, railways, when they came along, there was overbuilding, overinvestment. A lot of people lost money on railways. We still use railways. You go back 25, 30 years ago, there was a massive boom in spending on the internet. Same thing happened.
Starting point is 00:17:02 The promise of a new technology, a new way of doing things, typically brings in a lot of investment, a lot of hype, a lot of people lose money. But longer term, I mean, the world really, runs on the internet now. And people could see that would happen 25, 30 years ago. Just the path to get there was a little rocky from an investor standpoint. And so I think we're still in the very early stages of that with AI. We can see where things are going. But the path there is not necessarily paved with gold for a lot of people in the near term. Sean, in thinking about these companies who have moved maybe fast on layoffs or restructuring, is there any possibility that these companies are perhaps using AI and this hype around AI to maybe put a positive spin on
Starting point is 00:17:53 what are just actually necessary cost-cutting measures? You know, I'm sure that's the case for some. But the reality is that we're seeing early indications of the cost curve within companies shifting. And software companies in particular are very closely watched for how fast their revenues are growing and how profitable they are gross margin profits as well as their operating profits. And if the new norm is that companies can grow faster for much lower costs, you know, it's like an arms race. You have to be competitive with the new norm. And growth in particular has been more challenging for a lot of companies since interest rates started going up, you know, two, three years ago. So, you know, if you're not growing
Starting point is 00:18:44 and your costs are still high, that's a problem. If you've got these nimble young companies that are coming up and they're able to do it at a much lower cost, grow fast, they have more resources that they can use using AI to do more things. And so, again, it's that arms race element of it. If you're not using AI like your competitors or your competitors of the future are, you're get left behind. And also, all of these companies have very demanding investors, venture capitalist, public markets investors, and they are not going to be very patient if they see that your, you know, profit and loss statement looks like the one of old versus the one of new that's coming up. I imagine a lot of people listening are maybe interested in how companies are restructuring,
Starting point is 00:19:30 but also thinking, you know, about their own jobs and wondering, is my job safe? Am I next? Could this happen at my company, how likely are we to see more of these kinds of layoffs that involve AI? Oh, I think it's going to happen. I think you're seeing a lot of companies experimenting with this, a few that we've reported on. I would think anywhere in the next one to four quarters, we're going to start to see a lot more of these types of either hiring freezes or big layoffs. Okay. And Joe, what do you think? Maybe I'm naively optimistic, but in terms of mass layoffs. I really don't know. Like, there have been predictions of AI leading to mass layoffs, you know, in the next few years. There's nothing really backing up those predictions
Starting point is 00:20:18 more than, I don't know, like just gut feelings. Jobs can be more complicated than we assume even those that involve repetitive tasks involve some level of human judgment and creativity. And if you eliminate that, you might be losing something. But I do think a lot of of jobs are going to change. I do think it's likely that companies don't hire as much for certain roles. They try to get more done with the same amount of people or more done with fewer people. So I don't know how it's going to shake out, to be honest, other than to say that when you listen to some of these CEOs talk, it's clear there's going to be some turbulence for the labor market. And I would say, like, I think those layoffs or
Starting point is 00:21:07 hiring freezes, a lot of them will probably be revisited, like two or three years from now, if I'm right, and we see across the board big cuts, I think a few years from now, we'll see some revisiting of that and CEOs acknowledging, yeah, we cut a little too deep. We're having to hire people back, et cetera, because of the sorts of things Joe is talking about. Okay. So I guess I wonder, as AI continues to become more and more prominent, because that seems like the direction we're heading. Does it seem that the only possible result of that is that less people will be working? I guess I wonder if there's a sense that this could look different. Like, could it look like AI integration and people still having their jobs? Or what sense are you guys getting
Starting point is 00:21:51 about that? There's an element of how companies decide to use AI. It is a tool and you can use tools for all kinds of things. I can take a match and start a campfire to roast some marshmallows, or I can commit arson. Two choices. Only two choices. With AI, it's not dissimilar. Companies can use AI to help employees, you know, be more productive as a collaborator, generate new business lines or opportunities, and hire people to support that. Or they can use it to cut costs and cut headcount and boost profit margins and stock prices and bonuses and things like that. So I think it's likely we'll see both. Okay, Sean, Joe, thank you so much for joining us today. Thank you. Thank you. That was Joe Costaldo and Sean Silkoff, the
Starting point is 00:22:48 Globe's AI and Tech reporters. That's it for today. I'm Allie Graham. Kevin Sexton produced this episode. I produced the show with Madeline White and Bihal Stein. David Crosby edits the show. Adrian Chung is our senior producer and Angela Pichenza is our executive editor. Thank you for listening.

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.