Experis VP Bekir Atahan on Why AI Won’t Eliminate the Need for Entry-Level Employees

Hey guys. Thanks with Throw. We're here with pe Aha, who is Vice President of Data Center and AI Excellence for Experience.

And we're talking about the need for, well, the next generation of IT operations folks. 'cause some folks are saying in ai, we won't need all these young folks. Hir, welcome to the show.

Thank you very much. Thank you for having. So, on the one hand we have all heard it, right, AI is gonna replace the need for a lot of entry level folks, but maybe possibly are we about to cut our nose off to spite our face?

What do you think? I don't think so, right? I mean, I think that we keep the first rung, but we redesign it.

I think that entry level jobs are on are the courtesy, they're really the apprenticeship of engine of every profession. If you think about it, um, it's very, you learn the ta, the task stuff, the unwritten rules, the prof, professional intuition have to read the room. Um, a data set really cannot teach you that.

It also, where you learn the spot edge cases way of, you know, way, you know, way the trade offs and, and communicate judgment to a client or, or, or coworker. Um, yeah, I absolutely accelerate that learning, but it doesn't replace it. And when we deploy, I Carol, see there is two things, right?

I mean, the way that I see it is that the first is the skill atrophy, right? The classic autopilot failure, for example. Um, I mean the people will sit back onto the crisis, then the system has control to the, the least experienced person at the worst moment.

Second is the automation bias, right? Over trusting a fluent answer. Basically, models can be confidently wrong and crisp response can filter through.

AI is extremely confident when it's wrong. So that's why we designed for collaboration, right? I mean, tools that show their work surface, calibrate and uncertainty.

Keep humans on controls. Not, I mean, not premature automation that helps for perfection. Uh, thing has up display, not autopilot.

Basically, you know, HUD is basically is, is is the way to do things. And so that, um, uh, the entry-level jobs are this, is, this is exactly for that, right? We will always need it.

There's also a pipeline argument if you shrink or eliminate entry-level roles because the AI drafts it. You know, we create brutal teams with no succession, no tacit knowledge capture, decrease innovation, global morale, inability to adapt. Juniors do the UNC glamorous but wide work actually in there, right?

Um, they do turn the messy knowhow into checklists, prom cards and test cases. The very artifacts that make AI safer and better next quarter. Yes, AI growths, but entry level matters, not more, not less.

The move is to pair the junior levels, right? Give the first year employees AI powered leverage and give the organization the guard that keep human judgment, creativity, and critical thinking. Warm.

To your point, are we also assuming that the older folks are gonna stay in those jobs forever, when in reality they may retire and then we have nobody to hand off those functions to, right? That is true, right? I mean, this is why, I mean, if you, um, so if you think about that exactly, we do need to have the human in, in, in, in touch at, at, at, at any given point.

And then the other thing that I was wondering about all this is, well, maybe just the definition of one, an entry level job is gonna change and maybe we just need to rethink how we maybe even structure the IT organization. Is that a possibility? I think it is, right?

Um, the question that we always talk to our teams and, and our employees as well. When, when an end to employee comes in, how are they really going to, you know, implement AI proficiency? Um, is this going to be like a workflow of an automate or a collaborate?

So this is the distinction that we have to make. We choose automate. When the cost of an error is low and performance is consistently reliable instrument and monitor it, then, then ous collaboration, right?

When a judgment context or edge cases matter, which is the most learning work juniors should do? So I kind of look at it as a four practices, right? Heads up display experience, as I was mentioning before, right?

Like a pilot, our people should not just see the final answer. The tool, you know, must show its reasoning, expose calibrated uncertainty, and site source sources and tests. I mean, we need to avoid the, you know, black box answers basically.

Mm-hmm. Because this really pushes the, the juniors, the, the thinking rather than the rubber stamping it. The other thing I've been trying to figure out is, well, will each member of the IT team have their own portfolio of agents that they bring with them that perform a task?

Or will there be kind of like an AI agent that is trained to, uh, automate specific task on behalf of the entire team and and in effect, they are a member of the team That is Right. So everybody brings in and, and that's why that that really, that's a very good point that you're bringing, right? I think AI framework is in en in real and everybody brings their energies as long as the environment allows it to be coherent, to be able to encomp encompass every agent so that the s are as long as s are verified.

I think that is, that is the way the future is looking. The other thing that comes to mind is hopefully, rather than just kind of thinking about how do we do the same thing with fewer people that we're actually maybe thinking about maybe running more workloads than ever and doing more challenging things that we would not have attempted in the past. So, um, will the the overall footprint of it just become larger because we'll be running that many more workloads?

What do you think? Um, you are right? I mean, the creativity will increase for sure, right?

Um, jobs are a bundle of tasks, not a single thing if you think about it, right? Um, AI is brilliant at shifting the task mix, drafting texts, scanning logs. If you think about it, summarizing meetings, the integration work is really the important one, right?

Handling exceptions, weighing trade-offs, ethics, context, relationships remains deeply human on on that matter, right? Um, so, and then also the accountability, but again, the, I think that the velocity is going improve, um, uh, very fast, right? With, with those correlation of agents as well.

Mm-hmm. And aren't these AI agents, for all intents and purposes, they will each have their own personality and their own agenda, and at some point, will they not like bicker, like the my kids in the back of the car and somebody's gonna have to sort 'em out. Yeah, there, that is true.

So as long as Yeah, exactly. The governance is, is, I mean, so the way that it, you, we do need to do that. The governance is a very key point, right?

If they are, they, if they are bickering out, right? Uh, human intervention is required. Just like you, they're bickering on look, hey, hey guys, can you be quiet?

Kind of a thing, right? So it is very important to have clear rules and tooling to escalate to humans. It is going to be.

And because the models also needs to decline if the output is not available and, and they're uncertain, right? Um, we also need to be able to, observability has to be there. Logging prompts, versions, and data lineage is very important because at the of today, right?

We wanna make sure that what we are getting back is accountable. It is true. And it's not really something that is AI make or the agent is making up.

So from that work, you're absolutely right. So we do have to also quality metrics in place as well. Do we have enough appreciation for the, the level of risk that might be occurring here?

Because as one wag once, put it to me, it's one thing to be wrong. It's quite another thing to be wrong at scale. And with ai, boy, we can be really wrong.

It is true, right? Um, flu isn't the truth when ask your son, right? And if you're not in the, in the loop, right?

How can we catch the confident errors before they even become expensive? Is, is always the question. So the governance plays a huge, huge role here.

And it used to be in, in the old ways we kind of do audits, right? The right after the project completes or whatever the processes or the, now that governance, that auditing process has to be in, in inflow. Are you at all worried that, I mean, we already kind of have a shortage of IT people, and theoretically AI will make up for some of that, but it might exacerbate things 'cause more people will just decide that they don't want to have a career in a, in it in the first place.

Um, I don't foresee that. Um, because the new work emerges, right? Prompt design, knowledge, curation, qa, this is not a revolution if you, uh, if I may say, right?

I think that it's an evolution. So, uh, from that perspective, I think that it's going to really, really help us, right? Um, to, to get new jobs and, and, and, and evolve.

Mm-hmm. So what will be the new jobs in it? 'cause I'm assuming that there's gonna be some sort of evolution and I don't know, you know, today we have lots of different silos, but will the silos converge and how do you view the IT team will be structured?

Yeah, I think there's gonna be like a prompt engineers knowledge based curation and create, I mean, like you're gonna have QA for create is gonna be now this defined only for the, you know, the AI outputs and then you're going to end up having workflow of engineers, right? Though these are some of these, the traditional DevOps stuff is gonna really shift in into these. I guess here's the question.

Yeah. I don't know if you have children or not, but if you do, would you tell 'em to go into IT as a career? I mean, is this a place for them to go or should they all be studying, I don't know, bioscience or something?

No, I, so it depends on what part of the IT they're gonna, they, they want, I did have a son and IS is 16, right? So, and, and, um, that's the question that we always do, right? Um, and that's a very, very good question that, and that's a huge argument that I'm not arguing, but a discussion that we always having inside the house.

Engineering is something that he, he likes to go in and I would, I would, it is very general term, but I mean, if you're talking about kind of a coding and, and et cetera, right? I would definitely see the benefit in it with the, with the, with the math and the, and the physics and on, on, on backing it as well. So I think that the things are shifting in a, in a, in a very nice way from the AI perspective, but the creativity and being problem solving, focusing on those is gonna be really important.

So I think that there's a lot of work to be done in the, it, again, human oversight through the, IT is going to remain quite a bit. Mm-hmm. It's important to have fun at work, otherwise it's well too much like work.

So what we have fun in it, continue to like do something here that we enjoy. Um, and will that joy come in the form of, I don't know, trying to manage a small army of AI agents? I mean, what's gonna make this compelling enough for people to keep coming back for more?

Well, if you think about it, right, this a this's a good question. So if you think about it, um, this is, you can't really, if you go to the canyon and you try to, to jump right and that to the other side of it, and, um, right now it may sound like AI is gonna do all that, but it's not. Right?

So we will, we will still need to build bridges. We will still need to build, you know, hiking paths we need. We may need to walk around the, around the perimeters, right?

Or, or travel around it. But the thing is, again, building a bridge is gonna be the key and that's gonna continue to happen for, for many decades. I think that's was, that's how we need to use the AI and, and how we should approach it, even from the IT perspective is that, let's use it as a partner, let's use it to, to, to build and, and, and, um, get better actually.

So what is your best advice to your fellow IT leaders? Is there something I should be doing? I think a lot of folks are just kind of telling everybody to experiment with ai, but should I have them in a room and should we be discussing our future career paths together?

I mean, how should that conversation go? So the way that, so this is a good conversation and then the way that I see how the organizations are reshaping, right, uh, is that roles are becoming skills projects are becoming products and fusion teams that own the workflow end-to-end. That's something that we, we kind of do that in my job today.

Uh, we are valuing skill portfolios, right? I mean prompting, verification data, hygiene judgment over the fixed titles. For example, we, we should treat the AI work as a product, not at one of project.

Because once you, once you approach approach it that way, right now, you can take your product and apply it to every aspects of the business that you're in. Uh, um, prompt, you know, prompt libraries, which we will corporate with the owners version and backlogs, right? It really helps through the process, small cross functional teams.

Is there something that is, is that industry that I think organizations are changing too. Domain expert, analyst engineer, right? On the business workflow from data to decision.

They, we, you know, they manage inputs, logic, s and outputs. No more throwing work over the wall basically. So that's, that, that is kind of the future of the, the IT and, and the development teams is going to be, in my opinion, uh, again, the governance is the very important, right?

Abstain po uh, and abstain policy, deferral routes, observability, quality metrics, all these needs to be built with the help of ai. But, but we need to be in the, in the, in the center of it. Will it people maybe find themselves working more closely with business users and the fo you know, and the, and the other rank and file.

And I asked this question 'cause you know, I once asked an old friend of mine, I said, you know, why did you get into it? And he said, I like machines, cats, and people in that order, but it seems like we're now gonna have to maybe work a little more hand in glove. And does that require a new way of thinking?

It is. I mean the small cross again, coming back to my small cross-functional teams, right? That that is the part of that.

So we will need to understand the business needs, right? We will need to be, everybody needs to be speaking in business as well. So it's really a good, um, kind of kind of going into the segue.

You can be just, just like you said, you know, uh, you said the ca and machines I think, right? So I think that human touch is is even more so important now than, than than ever, especially from between the IT and business because once you get the business requirements, you can also almost very quickly continue to build with the help of AI to the outcome that they're trying to get to. So the speed velocity increases and then obviously at the end, right, our productivity inclu increases as well.

If this is used, I mean, again, using it as a collaboration tool and, and then use the automate tasks, I think that we are gonna, we are gonna get, we will be very successful. So coming back full circle, um, do governments have a role in this conversation? Because they do a lot of funding of training and for different programs and they have a vested interest in shoring up the tax base, but um, are they doing enough?

Yeah, um, I think they are. I think that they're trying to get, but again, just like any, any of the governments, right, the private sector is a lot less restrictions and also quicker to the pace government is gonna, they will catch up and, and they will need to, um, because this is the piece with the guardrails is, is is gonna be in place. If everything is that we are describing about, about on the private sector is uh, it can apply to the government and it can increase the efficiencies as well.

Well folks, you heard it here. Hey, AI is clearly gonna change everything there is about it. With one exception, we're still gonna need people and a lot of younger ones as well.

Hir, thanks for being on the show. Thank you very much. Thanks for having And back to you guys in the studio.