Raj Sethi on Why Junior Developers Remain Essential in the Age of AI

Guys, thanks for throwing. We're here with Raj Sethi, who's senior vice president and go-to market leader for software development lifecycle for global Logic. And we're talking about, well, all this noise about how AI is gonna eliminate the need for junior developers.

It's a hot debate, but I don't know. It's a lot of things about AI these days may or may not be true. Raj, welcome to the show.

Good to be here, Mike. We're looking forward to having this chat with you, talking about a very hot topic. So Wells, your code, we here, it's hard for the kids coming outta college to get a job.

And we're also hearing a lot about how senior developers are now automating a lot of the tasks they might have assigned once to junior developers. And this is resulting in maybe not as many junior developers being hired. And certainly there's been a lot of layoffs lately, but I'm not sure those are because of AI or not.

But Raj, what's your take on what's going on here? Well, I think what is, this has to do with how the industry sort of, uh, you know, expanded. Uh, we had a moment in time where the, you know, fairly large amount of organizations hired a lot of people on staff, assuming there's gonna be business coming up.

And I think a lot of those forecasts were changed because of gen AI and the need for that. And that's one of the reasons why we see some of these layoffs that we see in the industry. However, I think that when you look at the larger trend, uh, it is an area of flux.

Businesses want to truly understand how gen AI adds to that productivity. There are lots of claims, but, uh, if you look into some of this claims, they're all added up with an additional stop where they'll say, oh, 60 and 70% of our lines of code is written by Gen ai, but, uh, we still need people to look at this code, review it, ensure that they're right, or they would say, we still need to proc the system, which is another form of programming, um, to, to get these lines of code. Uh, and you know, one may argue, well, this, it's a one, one-time job, or it's a constant tweaking.

That's a separate issue. The key part though is the systems are not there where they're gonna produce massive amounts of lines of code without any human intervention. So that being said, um, I think that the junior developers in a far better position, sometimes I'll look at, because they have an option to adapt.

Um, senior developers are more sort of set in their ways of thinking and doing things. So, you know, from my gen AI perspective, all organizations wanna move there. So I do see that the opportunity for junior developers is still there.

However, it looks very different. It's not about writing boilerplate code, it's not about, uh, you know, writing test cases. Those are low hanging fruit that gene AI is gonna do far better job at.

But I still think that there is the process of validating, reviewing, being, doing critical analysis that's still junior developers can do. Uh, it's really about building that muscle within yourself. So coming out of the school, think more around critical thinking, how systems are developed, rather than doing the traditional ways of software coding.

It's really about not competing with ai, but how would you use AI more effectively? How does a junior developer get that expertise? If we're not hiring junior developers in the first place, who will be the senior developers tomorrow and where will we get that cognitive capability?

So it's, it's interesting, you know, how things have changed back. You know, when I came into the workforce in the nineties, uh, uh, it was all about getting this first job to do things. There was, internet was just coming around, uh, you know, email was just coming around.

So obviously you needed access to an infrastructure or an enterprise to even gain experience. That has changed. I mean, if you wanna learn today, there's ton of information on YouTube.

There are so many courses that you could take online. You could literally build your own portfolio today and not have to wait for that first job to show up. Right?

Uh, that to me is a, a game changer today because you don't need to give your just a resume. Yeah. Actually show your work to people.

And I think often, um, you know, young people don't focus on that. They all the two, the index over index on that first opportunity, that first opportunity will certainly come their way if they're prepared. But the question is, what's that preparation?

It's not necessarily just a degree from a college. Um, it, you, you could demonstrate your work, uh, far more effectively than you could do before. There are some critics of the junior developers who are coming outta college who are maybe too attached to AI and not doing enough critical thinking of the code that's being generated, because some of that code is, shall we say, overly verbose and maybe even doesn't make a whole lot of sense and certainly doesn't run particularly efficiently.

So, um, are we in danger of kind of going down a path here that's gonna lead to, I don't know, massive amounts of technical debt that's unsustainable? Right? I don't think the issue is that I just need a engineer to drive ai.

I think that, like I said, you need to have critical thinking. You should know algorithms, you should be able to review code faster looking at the code. You can mean there are times when I, you could look at the code and say, well, this thing meets two loops.

I don't see the second loop. There's a problem in this code. Um, so proofreading and understanding what is written there is, is a, is is a paramount, uh, skill that you would still need.

I don't think that is gone. And to the extent where people are thinking about white coding, yeah, prototyping is fine. You could do white coding there.

I don't think white coding has been very successful in the large enterprise. And I'll give you an example. Uh, rep being one of the, uh, providers of it, they, they weren't like, they were in their, uh, upper, you know, maybe in 2020 third million dollar revenue.

And then when they started focusing on the business analysts and business community, uh, their a RR went through the roof five times. Um, and I'm saying that it's good to prototype, but these are not software tools where you can just get away with white coding. Um, I think that the junior developers that we're talking about need to have proper formal training in software need to understand how things work because although a lot of times when I see people pushing the boundary of what gen AI is supposed to, you know, there's a, there's an idea that somehow you can reason anything as opposed to reason few things.

Having that ability to figure out what's the right problem to solve, that's critical thinking. Um, that's not gonna come if you, if you haven't gone through formal training in schools or, or haven't gone through any formal learning. I mean, I'm not necessarily have to go to school.

There's ton of material, right, which can prep you off for that world. Do we need to revisit how we teach all these skills in school? I have, I feel like sometimes maybe, you know, the kids who are coming out are not as well versed in the fundamentals and, and they're maybe too dependent upon AI and that's probably not a good thing.

And maybe that's the fault of the way they were taught. I do agree with you. There's a fair bit of that has to, that has to happen.

Um, but I do think that a lot of curriculum is getting influenced with, uh, with, you know, how some of the material that's available online through several organizations that are offering this. And, uh, I would generally say that that is a, would be a good approach. But I do think that what needs to be inculcated is that learning doesn't stop at school, in my opinion.

You decline stocks once you leave school. So learning has to be a continual process. It's an investment one has to make.

Uh, if you stop at any point in time, well, it's downhill from there. But, uh, I think that junior developers should come with that mindset that they're coming at the top. It's just that they need to learn how to apply that knowledge and that could be something that they can invest during school.

Like I said, if you build your own portfolio where you want to demonstrate your work, that's, that's a seeding ground for, for you to show your capability in the enterprise when you join the workforce, how productive and how contributing you would be as an individual. I think the, the joy of being a software developer is that you are solving problems using code. And I think a lot of people who do that today kind of have attached to the actual writing of that code to their brain as part of the joy.

And, but is that gonna be the case going forward? Or is it gonna be more about proofreading and reading the code to optimizing when it's created by a machine who probably won't get it perfect, but at the end of the day, we'll do it faster? Well, I, I, I think it, it's two ways to look at it, right?

It's really about abstraction. I mean, think about Lego blocks. Uh, have they taken the joy of creating things?

I mean, look, and the people of all ages use Lego blocks to put things together. Nobody's interested in how Lego blocks are made. Um, but that abstraction of the block is the important part.

And I think problem solving to a certain extent works that way. We've developed these skills and capabilities depending on where the technology was. I mean, we've been waiting for, uh, an opportunity where we could talk to a machine to do something faster for us.

I think Gene has started unlocking that for us. So there is a significant, um, work that is involved in structuring your thinking. So when you talk about systems, it's never about a point where you say, solve this problem for me.

It is more to that today. Uh, and the software is not anywhere, uh, or the systems don't exist today where I could take a very high level problem and solve it at, like, for example, improve my, um, you know, uh, campaign management or improve the efficacy of my marketing by 1%. That's the problem.

Business would pose, what does that translate back into the systems is a problem that gene AI is not quite capable of doing that, but a human would be. Uh, you could use gene AI to make that faster, better, but that's the way you need to start thinking about that. It's really not about coding.

Uh, it's more about structured thinking and having that ability now to communicate. So I think it unlocks the opportunity to, for lots of folks, uh, to, to sort of participate in systems building rather than just the developers who knew how to code in a specific language or a set of languages. So I think there's, there's more fun coming up.

Speaking of fun, um, are we going to build more software faster or maybe we're just focused, uh, too narrowly and maybe we should be thinking more about maybe making the software we do build today higher quality and just better. I think it's all of that. I think software is going nowhere.

Uh, the interface between the machine and the human, uh, is, is will be bridged with technology one way or the other. I don't know at what point in time we build biological systems that are exactly the way we work as humans or, or, uh, organisms. But the key part is that the software is there to stay.

The question is whether there's gonna be packaged software, the, the one that we have known for the longest time, that where you buy and then you configure or customize it to suit your business needs. Or would there be a situation where you build software to your specification, you could talk to it. I, I think that that's where the trend is going to be.

Uh, or at least I predict. I I, and nobody has this crystal ball, but I think that there will be an opportunity for people to create software that meets their needs and that's secure and that's high quality. Um, and it's, it's not, or it's always end when it comes to quality and and security, right?

The key part is that do I need to deal with packaged software or do I get something the way I want it? I think that's where the market's gonna move you, you would wanna buy software the way you want to use it. Well, you talked about abstraction, so let's go down that path for a little bit.

'cause I might argue that, you know, we created Java and other programming languages to have a higher level of abstraction with the machine. But maybe as we go along here and we start talking to the machine to write code, well just write everything back in assembly because that's more efficient. Yeah, I, I do agree with you.

I think that these are intermediate languages. Um, you know, compilers were created because it was too hard to quote them in assembly. Uh, I think that gen AI gets rid of some of this stuff and says, well, I can go straight to this or create some intermediate language and that gets compiled down, but it's not relevant, whether it's Java or whether it's C the key aspect is can you define your system, um, you know, in a precise manner?

Can you define the way you would want to test it? Those are the elements that will become material. So spec is there to stay.

I think that machines will get better so that you don't even need to define spec at that, the lowest level. But spec is the key part. I mean, we still need to communicate with humans.

So even if we wanna get work done, we, we communicate with humans. And I think that's the line of communication that that will needs to be established with a machine. So gen AI is suitable.

Whether these intermediate languages are required, I don't know. I think that, uh, they probably will lose favor over a period of time. They're going nowhere in the short term, for sure.

Mm-hmm. So ultimately, what's your best advice though, to kids coming outta college right now? What would you be telling them and what might you tell their parents?

Oh, I, I, I think that, uh, they should be constantly learning, start experimenting right now. There's so much information. It's democratized today.

It's all over. Whether it's not just necessarily in developed countries, it is available to every human on this earth. And, and the question really is how driven are you?

And, and, and you know it, so it'll all depend on what drives you. If learning is something that you enjoy, this is the best time to live. Um, and, and, and I think that you will, you will bear the fruits of your industry.

Uh, if you are in that path of learning, you will get opportunities. And you know, I look at it from a very optimistic point of view that if ever in the human history, a technology got rid of the young, we would be doomed as a species. Not happened, hasn't happened.

And I don't think it will happen. I think the young will always adapt. I I think more about senior folks like myself and others, that what will happen to them?

Will they work till they're in their seventies, eighties? I don't know. That is, that's a question that I, I pondered out.

I, i, I worry less about the young. But I do would say that focus on learning, focus on learning, focus on mathematics, focus on linguistics. This is where the game is, uh, sciences, uh, and this, there's no better time to live today to learn about all these areas, cross pollinate ideas, use gen AI as, as a communicator, as a friend that you could talk to.

I'm not asking, suggesting getting social advice. I'm talking about, uh, very specifically, um, you know, science-based information, engineering based information. This are, this is a great idea where you could cross pollinate ideas, take ideas from software industry or automotive industry, apply it to, um, you know, energy and vice versa.

I mean, things that are now at your disposal. Uh, when there, I mean, you have a PhD friend sitting to you who is knowledgeable in any field you want. I don't think that has been an area we've had in the past.

So young people, great time to live. Don't worry about these short term stories about this thing. It'll come and go.

I think it's a matter of few years here or there. The industry is gonna settle down. All right folks, you heard it here.

The diploma is not an end. It's a means to a larger end. And frankly, there's more interesting things to learn outside the classroom as there are in it.

So maybe just enjoy what we do because there's always gonna be some great new innovation and some new awesome thing that nobody ever thought of. It's all gonna start with somebody who did something with code. Hey Raj, thanks for being on the show.

My, my pleasure. Thank you. All right.

And back to you guys in the studio.