261. AI won’t kill jobs — but short-term thinking will
Jul 09, 2025
Some CEOs are already replacing engineers and junior staff with large language models.
But what if the real risk isn’t that AI replaces your team — it’s that it doesn’t, and you’re left without a talent pipeline?
In this episode, Sophia Matveeva breaks down:
-
Why cutting junior hiring today could sabotage your company in 3–5 years
-
What IBM’s CEO gets right that others get very wrong
-
How the hype around “AI-first” companies is creating poor decision-making
-
What the S-curve of innovation tells us about the real pace of AI progress
-
And why thinking long-term — not just cutting costs this quarter — is a strategic advantage
Whether you’re a government leader, innovation executive, or non-technical founder, this episode will help you make smarter, more resilient decisions in the age of AI.
Chapters
00:00 — The Real Risk Isn’t AI — It’s What Happens When It Doesn’t Work
02:45 — What CEOs Are Saying: Duolingo, Klarna… and IBM’s Contrarian View
06:30 — The Junior Job Crunch: Accountancy’s Mistake Repeats in Tech
09:15 — The S-Curve Explained: Why AI Progress Is Slowing
14:10 — Your Strategic Takeaways: What Smart Leaders Are Doing Now
Growth Through Innovation
If your organisation wants to drive revenue through innovation, book a call with us here.
Our workshops and innovation strategies have helped Constellation Brands, the Royal Bank of Canada and Oxford University.
Listen to Tech for Non-Techies on:
Transcript
Sophia Matveeva (00:00.438)
What if cutting junior staff to go AI first is actually going to leave your company worse off in three years time? Well, some CEOs are making exactly that mistake. And today I'm going to show you why the real risk is not about AI replacing jobs. It's what happens when it doesn't. Welcome to the Tech for Antecha's podcast. I'm your host, Tech of
executive coach at Chicago Booth MBA, Sophia Matheo. My aim here is to help you have a great career in the digital age. In a time when even your coffee shop has an app, you simply have to speak tech. On this podcast, I share core technology concepts, help you relate them to business outcomes, and most importantly, share practical advice on what you can do to become a digital leader today.
If you want to have a great career in the digital age, this podcast is for you. Hello smart people. How are you today? It is another glorious sunny day in London. If we've never met before, I'm Sophia Matveva. I'm a non-technical founder who has built tech businesses from scratch and helped governments and Fortune 500s upscale for the digital economy. And this show is where non-technical founders and business leaders learn how to build tech ventures.
without drowning in tech jargon. And so if that's your goal, then hit subscribe. So my aim in this show is really to help you make discerning decisions, basically be discerning, which I think is becoming increasingly difficult in this age of AI hype. So this is what this lesson, today's lesson is going to be about. And today's lesson is also about thinking strategically for the long term and not about quick hits of cost cutting for the next quarter.
You know, we often admire Amazon and Jeff Bezos and, you know, Jeff Bezos thinks really, really for the long term. And this is what I'm going to be encouraging you to do in today's episode. So you, unless you've been living under a rock, you have seen that some CEOs are making very bold claims about replacing staff and especially engineers with AI. And that makes sense because engineers are expensive and my dear engineering audience.
Sophia Matveeva (02:22.222)
those of you who are listening, you are quite hard to manage. Love you, but it's Anyway, so I'm going to show you what this move to replace engineers and junior staff with AI is actually misguided and what you should do instead. So we're going to have some technical definitions. You will have some examples of different companies and what they're doing. So basically, by the end of this 20 or 30 minutes, you are going to be way more informed. And if you find
this information useful, then I would love it if you could share it just with one smart person who also wants to be successful in the digital age. And obviously, if you haven't yet, then please leave the show a rating and review. Or if you're watching on YouTube, just press like. That really genuinely makes me happy. OK, so what are some CEOs doing and saying? If you use Duolingo, you might have heard that the Duolingo CEO, Louis von Ahn,
said that the company is going AI first. And basically that led to a massive backlash because he said, we're going to cut contractors where AI can handle the work. And there was then a massive backlash online from Duolingo users and so on. So here's another example, Klarna. You might've heard of them. It's a buy now, pay later company. So their CEO, Sebastian Sumatowski,
he said that AI has replaced 700 customer service roles. So that he basically cut 700 jobs in customer service and said that AI can do a lot of the jobs that these people do. And then what happened, customer service basically got really bad. And so he had to rehire them. So the real issue is that these sound bites, like AI can already do all of the jobs that humans consume. All of these sound bites, they...
imply that AI will replace entire functions. And this creates hype-based decision-making, which is not a good idea. That's not how you want to run a company. And also it just means that people are backtracking really quickly and really publicly. And basically, I don't want you to do that. But not every CEO is falling for the hype. So listen to this. So IBM, obviously massive tech company,
Sophia Matveeva (04:41.837)
Their CEO, Arvind Krishna, said something completely different in Fortune. So he said, the first thing you can automate is a repetitive white collar job. while AI could take over 10 % to 20 % of lower level tasks, it wouldn't take a person's job altogether because no one's job is composed entirely of these sorts of tasks.
I expect my programmers to get 30 % more productive thanks to the technology. And I don't intend to get rid of a single one. In fact, I'll get more. So interesting. There are some CEOs who saying, okay, we're going to get rid of junior and kind of low level jobs. And another CEO is saying, actually, I'm going to get more people and potentially, you know, more junior engineers as a result of the technology.
So IBM sees AI as a tool to boost productivity, not to replace people. And this is the crucial mindset difference, especially for engineering and product teams. So what's already happening? we've already seen, so I'm gonna give you two examples, what's happening on the business side and what's happening on the developer side. So let's talk about the business side first. So the big four accountancy firms have slashed graduate hiring in the UK and the US.
and AI is handling a lot of the admin. So basically there are fewer entry level roles. And you know, this sort of makes sense because, know, at TechPlan Techies, we are using AI to make a lot of our admin jobs easier, but it doesn't mean that we don't do any admin at all. Sadly, I wish that was the case. That's just not true. So, but the issue with companies like, you know, Deloitte cutting their graduate intakes,
is that means that they don't have a pipeline of junior talent. So if they don't have this pipeline of junior people that they're training now, what's going to happen in say three to five years time? A little while ago, we actually had this problem in the accountancy world because the big four stopped hiring juniors, I think about 10 years ago, and they ended up with a shortage of qualified accountants.
Sophia Matveeva (06:57.706)
And so now there are 340,000 fewer accountants in the US since 2019 and CFOs are now scrambling to fill junior roles. So the lesson here is that cutting junior jobs creates long-term capability issues. So this is why amid this AI hype, you don't just want to think about how can I cut costs today? You also want to think about, okay,
How can I increase productivity in the future? And by cutting costs today, you might be decreasing your productivity in the future. This is why I want you to be smart and discerning. Okay, so let's look at what's happening on the software and tech side of the job market. So entry-level developers are struggling to get hired. There are junior job listings that are down 30 to 50 % at many firms.
Reddit, which is loved by developers, Reddit forums are full of people saying that nobody's hiring junior developers anymore. And UK and US computer science grads face record high unemployment, which is kind of crazy because only recently, know, only when these people were applying for computer science degrees, it was like, okay, if you do that, basically you're set for life. You're going to be rich no matter what. essentially from where I'm sitting,
the pendulum has swung too far. The pendulum has swung from, okay, do a computer science degree and everything will be fine forever to, okay, nobody needs you at all. There is a middle ground. This is what I'm talking to you about. Okay, so this AI narrative that essentially AI will take over all of these junior tasks and you don't need junior people anymore, that narrative is flawed. So here are some technical reasons for that. So basically, what's he doing?
is that CEOs are assuming that AI will improve fast enough to replace mid and senior engineers. So this is the assumption. The assumption is that, okay, AI is good enough to do a lot of junior jobs. Okay, fine. We don't need the junior engineers. But when I say to them, yeah, but what are you going to do later? Because those junior engineers will eventually become senior engineers. So what are you going to do when you don't have senior engineers? CEOs, some CEOs are assuming that
Sophia Matveeva (09:19.391)
by the time they have that problem, AI is going to improve so much that actually AI is going to do the jobs of senior engineers. Now, this is a really, really dangerous assumption because it's basically assuming that something will exist that does not yet exist that will solve a really, really big problem for you. That is not the way to have a business strategy.
You can't rely on continued exponential growth of AI because that's basically not what's happening now. Now here's the technical bit I promised I'd tell you about. I want to tell you about the S curve. So if you've never heard of what the S curve of technology adoption is, you're about to find out. So just picture an S. You know what that looks like. So basically first there is very, very slow development and then suddenly kind of goes up because at first new technologies, know, first they're kind of just in the lab. So nothing really
visible is happening and then they grow really really fast and this is exponential growth. So that's like when chat gpt went from zero to 100 million users in two months. Remember when those news stories came out and everybody was going completely crazy about it. So what then happens is that over time progress slows down but that's not because people stop working on it but because
essentially the easy wins in creating this new technology are over. So it grows really fast. There are lots of easy wins. Once the easy wins are over, progress slows down. And that's when growth becomes logarithmic. So it's still growing, but it is growing at a much slower and flatter rate. So basically it goes from exponential growth to logarithmic growth. If that doesn't, if these words don't make sense to you, basically it just means that growth is still happening.
but not at the super, super high rate. And this pattern is called an S curve. So if there's a long ramp up, slow progress, then there's a sharp acceleration, which is AI hype, which is kind of the phase that we're in now. Then there is a plateau when improvements become harder and more expensive. And then this is when you start seeing things like AI systems, like large language models are hitting this plateau that's kind of on the verge is happening. And
Sophia Matveeva (11:37.564)
At that point, the models are not getting smarter at the same rate. So this is happening to some extent already because for example, one of the issues is that the models need data. so, you know, models like Chai GPT and Claude, they are built, they're trained on the data that there is available in the world, like on the internet, you know, data that they can buy and so on. They're literally running out of data. And we as humans, we're not creating
enough new data for them to train. So it's harder for them to get smarter if they don't have new information. So there are some companies that are creating synthetic data. That's like fake data. And you can imagine that there can be all sorts of problems with that. There are also large language model companies that are literally hiring scientists, like hiring PhDs to solve problems. then, so like they hire really smart people to solve very difficult problems.
So then the AI can learn on that. Now that is really expensive and definitely not scalable. So this is why there is this difference in growth improvement. So progress is becoming harder and more expensive. That's basically, just need to know progress is becoming harder and more expensive when it comes to LLMs. And when that happens and the high bubble burst, then funding is going to slow and then unrealistic expectations are going to crash. Now,
Let's move away from the S curve and actually talk about what that means for people and what that means for who you should be hiring and who you should not be hiring. So the real risk is actually not having a talent pipeline in the medium term. So let's just, you know, just kind of humor me. Imagine that AI does not improve to the level that it can replace
mid-level and senior engineers, or just mid-level and senior employees. And so imagine that AI doesn't get to the super clever level. But at this point, you need mid and senior engineers, but you don't have any because you stopped hiring juniors. So the only thing that you can do is try to pinch mid-level engineers from your competitors who have been hiring juniors. Now, do you think that is going to be easy? Do you think it's going to be expensive? Yes.
Sophia Matveeva (14:01.285)
Exactly. You see my point. So I don't want you and your companies like where you're working to face a skills drought in the medium term. This risk is especially high in regulators industries where it takes ages to, you know, to basically understand what the industry allows you to do and what it doesn't. And so engineers who have worked in say industries that require high security or industries that focus on human safety, there is a lot of regulation there.
and a junior engineer is just not going to understand all of those regulation. So the risk is high in regulated industries and AI assurance and product strategy and cross-functional collaboration because when you're a junior, you don't know how to collaborate with anybody. You don't even know what the other functions are. You can only do cross-functional collaboration when you rise up the ranks a bit. So the main point is that senior engineers do not grow overnight and AI
right now cannot replace their judgment and it is very, very risky to assume that this thing, hasn't yet been invented, is going to be invented and then everything will be fine. So what can you actually do? Well, one of things you can do is you can leave this podcast a rating and a review and press the like button if you're watching on YouTube. But seriously, what can you do? So stay informed on what AI can actually do today.
And this is where I keep on telling you, be discerning, keep on using your brain and not what very well paid PR consultants from AI companies will tell you. So stay updated on what AI can genuinely do. Also, train your teams to experiment with it safely. Look at the IBM CEO's example. He saying we are going to become more productive by getting our people to use it.
And that means that because we're more productive, we'll basically make more money, we'll make more stuff, our customers will be happier. So that means that we can hire more people. I'll actually paste the article from Fortune where he talks about that so you can be inspired by his point of view. Now, here's a critical thing. If you're in a big company, make sure that you can keep hiring smart juniors because if you don't, you will regret it later because there are some companies that are not hiring junior engineers.
Sophia Matveeva (16:28.22)
There are some companies that are not hiring junior staff at all. But I said some companies, not some industries. So there will be competitors who are hiring junior people who are training them in that particular company. And so in three years time, they are going to be in a really, really good competitive position that the people who were just chasing basically short term cost cutting are not going to be in. OK, so the bottom line is keep your people, but train them.
And that's how you get the benefit of AI. And you know, learn from accountancy because they already have these issues, right? Okay. So remember this lesson was all about thinking for the long term and making strategic choices and not just having a boost in the next couple of quarters, which I understand can be really difficult, especially if you're dealing with public shareholders or you're just dealing with financial pressure, but
Thinking long term has literally never hurt anybody. And that's what I'm inviting you to do in this episode. Okay, my dear smart person, this is the end of today's lesson. If this has helped shift your thinking or given you some interesting insights, then I would love to hear from you. So please write a comment below or write or do a rating and a review. And if you're not yet a subscriber, what are you doing? This is very useful content. Anyway.
Thank you very much for listening and I shall be back with you next week. Ciao.
Sign up to our mailing list!
Be the first to hear about offers, classes and events
