Vibe coding has become one of the first examples we’ve seen of generative AI transforming an established industry. It’s received a lot of attention for how it can help companies accelerate into the future, but that’s only an accurate description when the appropriate guardrails are put in place and when it’s embraced hand-in-hand with human developers.
That part sometimes gets lost in translation, especially regarding junior developers, who find themselves at risk of being unjustifiably undervalued and marginalized in favor of AI.
EMBRACING “WHY” INSTEAD OF “HOW”
With AI’s emergence, we have lifted the burden of understanding “how” and raised the bar to understanding “why,” which is a reality to which young developers and businesses training them must adapt. It used to be that we valued IT workers because they knew how to write code in a specific framework, a specific location, cloud, data center, or edge. That has all become commoditized, because AI agents can do all that. We will always have a need for understanding “why,” though: Why do I need to write that application?
Verifying AI-generated code is necessary. It’s a task companies tend to put in the trusted hands of their senior developers, oftentimes keeping juniors out of the equation altogether. Recent data suggests junior-developer jobs are among the most at risk, as companies integrate AI as a result. For executives seriously considering going down this path, that would be an incredibly short-sighted decision.
Young developers are actually in the best position to become the new experts, compared to senior talent. It’s a matter of companies training teammates with (and on) AI—gamifying it in our case—and simply approaching the situation with common sense. Here’s how:
- Embrace “why”: Ensure onboarding focuses on training on business context and not just technical material.
- Modernize apprenticeship: Pair junior talent with seniors, who educate them on why decisions are made instead of how, while directing AI agents to execute those decisions.
- Move smart, not fast: Remember AI is a supplement for human growth, not a replacement.
Having humans verify AI code and generally oversee the software-generation process is a must-have. It’s a skillset grounded in technical and business acumen, which separates good coders from great architects. AI is an incredible asset, but it’s not perfect. You need an engine in place to continually develop human talent trained on how you specifically operate and the value you bring as a company. So, why would anyone decide to bring that “talent train” to a stop?
KEEP YOUR TALENT TRAIN MOVING
Employee turnover is very real. No company ever keeps their best developers forever. They eventually retire. Some will leave for other companies. It’s not just about replacing them, either. The best organizations will attract top talent and hire younger developers who can grow with the company.
You obviously still need to be able to trust senior technical talent, whether they are developers, architects, etc. With the right training, your juniors can eventually step into roles as seniors. If a company fails to strike the right balance between veteran leadership and young talent, it becomes less competitive over time. Even with AI, that’s true.
It doesn’t eliminate the need for humans. It just shifts their roles. There isn’t a way to artificially replicate learned real-world expertise acquired after years of experience. Consider the arguments presented in this Dead Neurons blog, namely the example of deciding when it is safe to cross the road. Basic code could be based on three variables, in this case: the color of the light, whether a car is approaching, and how far away it is.
That would work most of the time, not all. In real life, pedestrians process dozens of other variables in a split second, including the car’s speed, how attentive the driver seems to be, whether the pedestrian is carrying anything that might hinder their ability to cross faster, if necessary.
The result is a rule that’s significantly more sophisticated and has been perfected over thousands of crossings. It also cannot be properly taught or serialized in a prompt, because it would need to include an incredibly complex decision tree. That makes it only capable of being applied in real time by a human with the required years of experience.
TRUE HUMAN EXPERTISE
Veteran talent will still carry the load with a slew of responsibilities, including contextualizing decisions in business value and verifying that architecture and code generated by AI matches it. But junior developers should still have a place in any organization. They just need to be brought along the right way and regularly exposed to everyday occurrences, so they, too, can develop the same level of expertise.
Companies sometimes get blinded by how efficient AI can be. Those that use it to empower their next generation of teammates, instead of skipping over them, will find the most success. Go back to crossing the street. It’s not necessarily about getting across as fast as possible. It’s about getting across as efficiently as possible. But why?
Working for a solutions integrator, I understand better than most that our value resides in ensuring the “how” is done properly to meet that “why” for clients. In large platforms, understanding why and mapping that to what it brings to the table is still going to be a very big lift for clients. That’s where we build our expertise.
Training your junior talent on how to “cross the street” misses the point, because the goal is for them to know the end goal. We have always strived to develop teams to bring higher value to the organization as their capabilities mature. AI hasn’t changed that. It has just changed what we need to focus on: business understanding that is grounded in technology, rather than technology first. Once that mindset sets in, long-term success is likely to follow.
Juan Orlandini is chief technology officer of North America for Insight Enterprises.
