Ep63: Why Career Change is a Bad Idea
===
[00:00:00] As of the 9th of, march, 2026, the labor market data from the US Bureau of Labor Stats and Major. Industry analysis confirms that there is a massive structural shift in how the labor markets are now being built.
I have recently come to the conclusion for the first time in my entire career.
That it's not a good time to make a career change to code.
Welcome to Easier Said Than Done with me, Zubin Pratap, where I share with you the tens of thousands of dollars worth of self development that I did on my journey from 37 year old lawyer to professional software engineer. The goal of this podcast is to show you how to actually do those things that are EASIER SAID THAN DONE.
I have recently come to the conclusion for the first time in my entire career.
That it's not a good time to make a career change to code.
So much so that I've stopped taking students to my inner circle coaching program because I no longer believe it's the honest or the right thing to do.
No shade on those who want to keep selling courses and stuff like that, that's fine, but I do not believe in my personal.
Opinion that this is the right thing to do for people. Because quite simply, none of us on planet Earth having experienced something like this, and without having direct experience of something,
we are not really in a position to tell people what to do. And that's why I wanted to have this episode to talk about
one of my recent posts on LinkedIn about two weeks ago where I said that I believe this is the worst time to make a career change.
Not that it'll be like this forever, but right now.
At least for the next few months, perhaps even a couple of years, it is a bad time to make a change because there are so many big structural changes. This is not a cyclical thing. I was there in the.com [00:01:30] bust. I was there in the 2007, 2008, 2009 global financial crisis.
I've been there through all the little ones in between as well. This is different.
This is not cyclical. It is structural. And it is very important to understand why that means everything is different this time. So in the next 10 minutes, I'm gonna talk you through the key pieces of data that have informed my thinking and that made me pause on the Match Mastery in a Circle program and instead try to walk you through what's actually going on.
You can then use that information to come to your own conclusions. But I will also give you some thoughts at the end of this video on how I would approach it if I was trying to do a career change now. And just before I go down that path, I wanna be very clear with you all about something.
Nobody has a crystal ball. Alright? Anybody who tells you they know what's happening in the future is lying because nobody knows and things are moving way too fast. However, there are certain meta principles and those of you who have been following me on LinkedIn for the last few years, or on YouTube or my podcast, will understand what I'm talking about here.
There are meta principles that generally hold true. I'm no longer convinced all meta principles will hold true going forward, but generally. They will hold true.
One example of a meta principle is direction is always more important than speed, right?
Because it doesn't matter how fast you go if you're going in the wrong direction.
Now that's a meta principle. That's generally correct because okay, you may go in the wrong direction and get lucky and get someplace better. Sure.
Even a dead clock shows the right time, twice a day. It doesn't mean you want to use it a dead clock.
So you can get lucky, but generally. You want to go in the right direction because even when you go slow, you'll eventually get there, whereas going fast in the wrong direction, we'll never get you there.
Now, that is [00:03:00] one of the meta principles I mentioned in my post on LinkedIn where I said, it's really hard to figure out what the correct direction is right now because things are changing way too fast.
And the simplest example I, I can give you of this is right now, if you tried to send a rocket ship either to the moon or to Mar, or whatever it is. If you assume that the planet is staying still, you will miss it, right? Because the planet is moving constantly. So you have to perform some very complicated calculations to calculate how fast your rocket ship is traveling and which direction it should go.
Such that. By the time it arrives, it'll arrive wherever the planet is gonna be at, wherever Mars is gonna be at, right? So this is a classic problem, and this only works when you know Mars is gonna go around the sun and it takes a certain number of days to go around the sun. And in six months and three days and 18 hours, you know exactly where it's gonna be.
Great. Then you can skate to where the puck is going, right? But what if you don't know where the puck is going? What if the P is moving around in four dimensions and you have no idea where it's going? You're not going to be able to set proper direction. Okay, so with that in mind, let me walk you through some of the data that's gonna explain to you why I've come to this conclusion. Number one, entry level job postings. In the first quarter of 2025, which is a year ago, were 45% below the five year average. Okay? So we're talking pre pandemic times or just to the start of the pandemic and early career employment.
For the age group, 22 to 25 has seen a 13% relative decline. This is some, some institutional ads in Stanford analysis. Even if you wanna round down the numbers, who cares? The point is there is a very clear trend line here, right? The entry level gate is closing because
AI now handles a lot of the low complexity tasks.
A [00:04:30] lot of people have started to see this at their work. I see it at work. A lot of my colleagues here at work, we starting to see this everywhere. People are expected to do more because AI helps them do more. Traditionally, you'd hire new people into the team because you are out of capacity. When AI improves your productivity, you basically get extra capacity for almost no cost, and that's why a lot of the juniors are now being phased out because a lot of the entry level work in any career can be done by AI.
Now, what that means for career changes is that when you hop careers, you're usually moving to a new mountain, which means you have to climb down the one you're on and then climb up the other one you're on or moving to a new ladder, if you like, and the bottom rungs are no longer there, okay?
The bottom rungs are no longer there. So when you are making big career change, you can't really start again because the bottom rungs are no longer there.
Now 66% of global enterprises are reporting that they're cutting entry level hiring specifically due to the AI automation capabilities.
And recently we saw that Jack Dorsey in block cut pretty much almost half the company, right? Some 4,400 people or something. And he explicitly said, it's because we don't really need that many people because AI can do a lot.
Now, this is not to say that AI will replace everybody. That's not what it's saying. What it's saying is that AI can replace many people unless there is so much growth in the company, that everybody that AI replaces needs to then be replaced by actually having people, because that's how much the growth is.
But it's unclear that AI can't scale faster than human beings right now, the number two data point is what I call the slow leak problem. Okay. The US economy shed 92,000 jobs in February, 2026. Okay. [00:06:00] And that missed. Analyst expectations 'cause analysts were thinking a 15 thou a 50,000 job gain? No, there was a 92,000 net job loss.
Okay. The big sectors that were hiring, if you want to take it into some sort of positive territory, was I think healthcare and government. But that's not normal. That's basically the government expanding how many people. It hires and some healthcare sectors doing that. Okay, but for white collar knowledge work, there is a structural contraction, not a cyclical, a structural contraction.
Okay? Even though the health and government sectors may be adding jobs on a net basis, it's very important to remember this is structural, not cyclical, meaning this is not some sort of weird cycle we are going through currently.
It appears to be a fundamental shift in how the economy is structured, how jobs are delivered, and how job markets operate.
Will it change? Almost certainly. We just don't know when, because this has never happened before. It could be six months from now, it could be five years from now. We don't know, right? What we do know is history teaches us that this kind of disruption or structural change can last. Many years. Okay. When the industrial revolution happened, it wasn't like it just was uncomfortable for three years.
It affected an entire generation of people. Okay? I remember reading a stat somewhere about how people who operated hand looms were once among the highest paid people in a factory, sometimes earning more take home paid than even the factory owner who only has equity. But then I think within 12 or 18 years, some really short number like that, they were earning less than the minimum wage because the invention of automated looms in the industrial revolution, right?
So. Unlike cyclical layoffs, a steady month attrition or drop in in employment numbers would [00:07:30] suggest when there's no other cyclical or economic reason, would suggest that AI is displacing or at least pricing out very routine entry level tasks. Alright, now, data point number three.
The average number of applications per job ad reached 184 in 2025. Okay, that's 167% increase since 2022. Now, what does that mean? That means that AI tools and human beings are flooding the system with a lot of noise, a lot of applications. Now, if you remember, of those of you who have been following me for a while, I talked about this in in 2023 and 2024, when I said the LinkedIn easy button button.
Or the LinkedIn easy apply button is the worst button you can use. It is not helping you. Right? And I said this on my podcast as well. So the reason it doesn't help you is all it does is it floods the supply side of the market, the labor side of the market. The hiring side of the market now is a problem because they have too much.
Noise in the system. Alright? But that's why standard job boards are broken. That's why you're getting ghosted. It's because applying coal to any sector is statistically likely to result in zero visibility. Now think if you're moving to new career, right? If, if that's already happening in existing careers, it's going to get much worse in a new career because not only are you lost in the noise, you're also much more likely to be ignored, even if someone.
Notices you because you have no background. Now, that's not unique to the post AI world. That's what I've been teaching my students for years now, you are going to be much less visible in a new industry than you are in your existing industry. I'll put it another way. You're much more likely to be ignored in your new prospective industry.
Than you are in your current industry, which is why when people come to me for the inner circuit program, I used to teach them to avoid [00:09:00] job boards and avoid mass application by raising their credibility over a sustained period of time, such that people are automatically interested because they've seen that track record.
They're not just cold applying to places. Right now, back to the AI world, a high volume of AI generated resumes has forced about 87% of the companies to use AI driven filters that overlook. Most of the qualified, atypical candidates, right? Even if they're qualified, but they're atypical Now, this is understandable.
Okay. When. When we have big storms, we typically apply things like filters to the to the gutter systems, to the drainage system, et cetera, so that things don't get clogged in there. Okay? It's exactly the same situation with applications. There are only so many human beings who can process all these applications in hr.
Okay, if you start getting to 184 or whatever it is per person you are going to get to a point where you can't actually physically review that many applications. Like even if it takes you a good 10 minutes per person, that's 1,840 minutes. That's a long, long time, and that's. Not possible. Think about how busy you are at work, and now imagine if you had to look at 184 applications on top of that, okay?
Or even 90 applications on top of that. So when there's that much noise in the system, the filters get stronger and stronger and stronger. This is also why, in my opinion. Data structures and algorithms. Questions these days are much more complicated and interview loops are much longer at the big tech companies than they used to be 20 years ago because there's been so many people coming into the market who are learning how to do interview like this and do DSA and do good behavioral interviews and stuff that they have to raise the bar higher and higher as they filter, even if it has nothing to do with the job.
Okay. Macro data. Point four, [00:10:30] the Bureau of Labor Statistics in the us. Revise them March, 2025, so that's exactly a year ago, March, 2025, employment levels were revised downwards, not by a small amount, by 862,000 jobs. That's almost a million jobs downwards, meaning they were almost 1 million jobs off, or they were 862,000 jobs off What they estimated.
That's how big their margin of error was. Okay? And they admitted that the market was far weaker than they originally report. Now what does that mean? That means that these indicators, these BLS statistics, for example, that measure what happened are what's known as lag indicators, right? They lag six to 12 months behind the reality because it only got reported a few months ago about what happened in March, 2025.
So there was a nine to 12 month lag. Right now the problem is that by the time a job loss is officially recorded and you can start measuring these things, which is about 12 months later, the AI that caused that erosion is usually 500% more capable. In other words, by the time the 12 month data that's lagging gets correct.
The AI that caused that issue 12 months ago is now five times better than it was 12 months ago. Why? Because AI follows scaling laws as far as we can see right now, and it dramatically improve its performance in one year, much more than a human being can. Okay, now. How do you make a long-term career pivot based on official data when the data is probably at least 12 months old and the state of the art in the AI or the market has moved on, right?
The speed of AI adoption is months, and that's significantly faster than the Bureau or any other statisticians ability to revise the, [00:12:00] you know, the, the, the, the numbers on it. Okay? And what this means for you is you cannot skate to where the puck is because you, your idea of where the puck is was based on 12 month old data.
Right. So when you are looking at the puck, it's actually a historical snapshot of the puck. It's not where the puck is right now, and that's why I no longer hand on heart as a matter of integrity. Could think of how I could educate people on career change to code because it's completely different from anything we've ever seen.
And I, as an engineer, as a software engineer, my own way of practicing software engineering has changed so much in the last. 12 to 24 months so much. Right. I almost never hand write code. Now. One year ago I was handwriting code, like that's how fast things have changed. So, and I'm in the industry, I'm kind of in the cutting edge.
Okay. How can you tell people that their careers are gonna be fine, that they, that you know exactly what they need to do to get into their new career? When you know that you can't even measure where the puck is, okay, that all your data is lagging, it's outdated, and things are moving so fast that you're probably wildly off in your estimate.
Now, remember what I said right at the start, because direction is more important than speed. You need to get directional correctness for a career change. Okay. If you're not directionally correct, you are gonna spend years of your life demoralizing yourself, getting no results, wasting the opportunities in your existing career.
And if you don't like it, possibly wasting some earning potential. But most importantly, you'll break your own confidence over nothing. Not because you're not capable, but because you are moving in the wrong direction at speed.
So this is why I'm so adamant. That the only thing we should be doing at this point in time is pausing, which now brings me to the solutions of this.
If we know these four situations, the labor market has changed, [00:13:30] the, the data is lagging, the speed of AI is considerably more than anything we've ever seen before, and there is a structural shift in the market. What do we do about it? Well, the best I can offer you at this point in time are to try and again, step back to meta principles.
Think from first principle say, okay, what do we know? What do we think? What we think is very different from what we know? So what do we know? What do we think may be the case? And what is the data saying? Okay. And if we don't like what the data is saying, doesn't mean we attack the data. It means we need to change what we think.
Alright. So some of the solutions I can think of are
one, instead of doing a pivot to a whole new vertical or a whole new industry, start to integrate AI into your current expertise. Okay? 'cause most of my pe, most of the folks who watch my channel are people looking to change career, which means you already have expertise in your current career.
You've already developed a lot of judgment and a lot of knowhow, and a lot of experience. Combine that with AI and you will command a premium. Now, it's not gonna be a long-lived premium, okay? Maybe. A couple of years, but if you operate in that level now, let's say you have about 12 months to really learn some AI related skills, start applying it every day in your work and boost your own productivity by 30 or 50%.
Let's say you do that in the next 12 months. You are more likely to get promoted, you're more likely to get really good results. You're more likely to get other job opportunities in your industry that will then give you that promotion, that higher income. And from there, it just keeps compounding. Maybe you'll get into managerial functions.
I personally believe that. Management functions will have more of a premium than ever before, much more than the technical individual contributor functions, because that's what the AI can do. But the management judgment layer is what's really important. So I suspect that managers and like [00:15:00] myself are going to start doing more what's known as individual contributor work.
Because we can get an AI to do that individual contributor work and we will supervise human and AI teams. I, I truly believe the modern team is going to be a combination of some humans and several ais. And a manager has to apply their management skills to both. Right? So if you can get into the management track, try, how do you do that?
Capitalize in your current industry, go really hard in it and add AI to your skillset. And I don't mean coding, you don't have to just learn how to use tools. Like whether it's the Codex Tools or the Codex app or the cloud. You know, the cloud cloud app that they've now released work, it's called Workspace.
I can't remember what it's called. It's just slip my mind now. But cloud, cloud cowork, that's the one. Right? So there's so many, many of these tools just start learning and start to use them, right? And hey, don't be cheap. Yeah, this is an important investment in your career. Even if your employer doesn't support it, fine.
You have to be careful and mindful of the legal implications of using work data and these things, but pay for your own subscription and start doing some personal projects on the side just to learn how to use these tools. Are you really going to sacrifice your career over 30 bucks a month or 20 bucks a month?
It's crazy, right? Instead of going for dinner, just. Buy yourself two of these subscriptions. Play with two ais. Spending 50 bucks a month, who cares? It's a good investment. Okay? It's only 600 bucks a year. You will thank me for it later on.
So don't be cheap. Subscribe to a couple of the AI plans and, and get on top of that. Okay. Number two thing you can do is don't pivot into a sector just because it shows some signs of high productivity or stock growth, et cetera. Okay? Why do I say that? It's because there's been a decoupling between productivity and the number of hours spent per employee. So what's happening is it appears to be that the number hours spent to produce a productive outcome are slowly [00:16:30] dropping, but the productivity is going up, you know, 4, 5, 6 percentage points, right?
So what that means is people are able to do more with less, which means AI is taking over some of that work, right? So don't just go into a high. Productivity or high growth sector because it may be that a lot of that growth is coming from ai. So you wanna verify that the growth is at least partially attributable to employees.
Okay? So the output per employee should also be rising so fast, and therefore it indicates that human hiring will continue. Now, we don't know because we don't know what next industry is gonna be affected. We know coding is a logical place to start. Why? Because who built the AI software engineers?
What did they understand? Software. Therefore, they're going to build in their own domain. But already we started seeing. Marketing finance, healthcare, r and d definitely the legal industry in the last 13% of its stock, stock market. There's YC companies right now that are targeting accounting firms.
All knowledge work is gonna be affected, right? So we just need to figure out, okay,
if it's gonna be affected, what are the skills that we have that brings us to the higher level, like I said, potentially management skills and judgment skills.
The number three thing you should think about or what you can do is analyze what a pivot means for you in terms of the kind of role it's gonna land you in. Okay. So you want to move into a role where even if the automation is coming, you have other skills that'll make you be one of the people supervising that automation not being automated out.
That's very hard. A lot of people come to me and say, oh, I want to be a you know, quality assurance tester, or I want to be an IT customer services, and I'm like. Why by the time you re-skill and start getting those kinda opportunities, chances are that you're gonna be competing with AI and people are being [00:18:00] displaced by ai, right?
So keep in mind your competition is not other people like you are changing career. Your competition is nowadays AI and people displaced by ai, which means they have a lot more experience than you. Alright? That's number three. Number four is, please, for the love of God, stop mass supplying to jobs. Okay,
more than ever, you need to not be just one of the many numbers, which means you have to get out of your armchair, which means you have to try and move to a city where there are more opportunities.
Or if there's, if you're in a smaller city, and that's fine, then accept that you'll have fewer opportunities. That's just how things work. Okay. And then get out of your armchair. Start meeting people. Start going on offering your services to people. Educate yourself online. If you want to educate yourself with ai, fine.
I'm not the biggest fan of it because I know how badly wrong it can lead you. And information has never been the problem. The problem is knowing which information is useful.
Information has been free and abundant for 30 years now. Okay? But knowing what information is gonna move your needle, that's the hard part.
So if you can figure that out, great. Otherwise get a coach. And then start meeting people and establishing a credibility, not once, not twice over months. Okay? Volunteer for projects. Tell them you, you'll do it on the weekend and nights just to learn and you don't want any money. Start building your credibility.
Why? Because when somebody is looking for a human being to hire into a role. They're always gonna look for the person that has the most trust and credibility and the most familiarity with them. Okay. That's usually what employers care about a lot. Once they think that you have the basic technical skills after that, or whatever skills that is that you need, after that, all they're looking for is.
Is this person incredibly gonna solve our problem? Can we trust and rely on this person? Okay? [00:19:30] And the best way to prove both these things is by working in a high visibility environment in front of them, meeting people, putting yourself out there, getting outta your chair, and not clicking buttons. Alright, now number five, suggestion for you.
Is be careful of the terms like AI engineering and stuff. Yes, absolutely. There's a huge spike in the number of jobs mentioning AI skills. Okay. But consider what your competition is as a career changer. Your competition is other software engineers like me, who've been around a while, who are also re-skilling on AI to apply to a software engineering discipline.
Okay. You are competing. With people who aren't changing career, you are competing with people who are upgrading their career, but you are changing your career. That makes it really hard for you. So be very mindful about all the change that's happening in the industry and what it's doing to your competition.
People tend to forget to understand or forget to analyze what the competition looks like when there's changing career. Now, when I was doing the Inner Circle program. I spent months teaching people how to analyze the competition and it's never ever what they thought it was. Okay. You have to really go deep into this sort of stuff.
Okay. The sixth and the last suggestion for you is because the speed of AI adoption is really measured in months, not years. And it's significantly faster than how much, how fast we humans can learn and. Frankly, official stats can't keep up, as we saw with a six to 12 month lag. So if the AI is improving, you know, 500%, five times while your career pivot takes 12 to 24 months to execute, then you're effectively running a race where the AI is moving at a hundred miles an hour and you are moving at 20 miles, a 30 miles an hour.
Okay? You're running a race against a vehicle that is accelerating exponentially, and you are not, 'cause you're physically not capable [00:21:00] of it. The human brain cannot do that. Okay? So when you're doing that, you can't overtake. Okay, so it's best to stay in observation mode until you understand what the AI can do, and more importantly, what the market values, but the AI cannot do, and you need to try and figure out where you can try and fill that space.
Now, my current best hypothesis is people who are very strong communicators, people who have strong management skills, who are high empathy, and who are able to work with teams are going to go much further
then people, let's say, who are really smart software engineers but aren't good at communication, aren't very good at structured thinking, aren't very good at seeing around corners, aren't very good at, at working with people, what I call coalition building, driving influence, driving change, communicating a certain point of view and building that partnership cross-functionally.
Those are skills that you don't have and you can't negotiate with other teams and you can't communicate with other teams. You're probably gonna have a very quick ceiling on your, or very quick cap on your career, whether you're a software engineer or not. This is probably gonna be true because the AI can do a lot of the pure work that needs to be done, but it cannot do a lot of the orchestration work on top of that.
Okay? That's where human beings need to come in, and as long as human beings are building and driving companies, and as long as human beings are the consumers of what these companies produce. You kind of need to be able to deal with human beings because the machines can deal with each other much better than they ever could, and they always were good at that.
Now they're getting quite autonomously good at that. Okay, so those are my suggestions for you. I will say one last thing. Do not despair. Okay? Because the easiest thing in the world is to give into despair. I'm not gonna lie and tell you that I'm always cheerful. I'm not, I'm not always positive. I, I often feel despair too.
Not just about, I mean, not so much about the AI [00:22:30] stuff, but in my life, I've, I, I've felt it several times when I've been smashed in the face a lot more times than I would get to admit. And often nine times outta 10, my plans haven't worked out. Right. And the one outta 10 times that I did work was, you know, outta the park.
So it, it life compensates you in weird ways, okay? But if you despair, what ends up happening is when you are sad or feeling despondent, you stop to look for the opportunities. In other words, you know
when a door slams in your face, you're so busy looking at the door and leaning your forehead against the door and feeling sad for yourself that you don't realize that another window's open down the hall or another door open behind you, right?
And that's often how life can be. Now, it may not happen at the exact same moment, the door slammed in your face, but if you given to despair, you will stop paying attention to the world around you. And the best opportunities come from just constantly paying attention to the world around you and being patient and trying to pick a direction that you believe hand on heart is the best direction for you.
Why? Because direction's more important than speed. You could be wrong, but if you're very thoughtful about it, if you're logical about it, you'll listen to what the data is saying. You keep your eyes open for opportunities. Over time, you'll get better and better. Your gut instinct will get better and better, and then you will at least be able to have a little bit of peace.
In the chaos and turmoil of the world around you, because even though the data may be demoralizing, at least it's data and most of the people in the world will be operating ignorant of that data. So take advantage of the fact that you may have the data and you may not like what it says, but you have now practiced and cultivated the habit of being able to analyze that from first principles.
Choose the direction that you believe the data will support, and then go really hard in that. Okay? So not a great time to career change into any completely new segment. If you wanna augment your career, [00:24:00] you wanna expand your career, that's fine. If you're broadly in the same industry, you're good. But I think
Skipping tracks to a whole new industry. Pretty risky. In three years, in two years, we are probably gonna be back again to people doing that. It'll be fine because the, you know, the ground will be stable under our feet, but right now we have no idea where that puck is.
By the time we spot it, it's actually historical data and the puck's gone off somewhere else and we're not quite clear where it is. Okay. Because that's the nature of lagging indicators. Alright, hope that was helpful to you guys. Please like and subscribe if you've got some value to this. Leave comments.
I try and respond to them personally as much as I can. And I'm doing this really to try and share what I've learned about the world in the last, you know, 25 years of working. If there's anything else I can do, just drop it in the comment. All the very best and I'll see you next time. Don't forget to like and subscribe.
Bye.
Just subscribe, you know you gotta do it.
