I’ll never forget my first real encounter with modern AI. It was during a quiet holiday break when I finally had time to properly explore tools like Claude and ChatGPT. Like many engineering leaders, I had my “oh shit” moment – that stunning realization that everything about how we build software is about to change. If you haven’t had this moment yet, you will.

This transformation reminds me of another pivotal moment in computing history: the introduction of high-level programming languages. Before FORTRAN, programming meant writing machine code. The idea that a computer could translate human-readable code into machine instructions was revolutionary – and controversial. Many programmers at the time argued that hand-optimized assembly would always be superior. They were looking at the future through the lens of their present.

But what’s easily overlooked is how this shift didn’t just change how we programmed – it fundamentally redefined what programming meant. Assembly programmers weren’t just asked to learn a new tool; they had to reconceptualize their entire profession. The role of “programmer” transformed into something their predecessors wouldn’t have recognized. We’re standing at a similar inflection point today, but the pace and scale of change promise to be far more dramatic.

The Case for Junior Talent

Many of my peers in engineering leadership are responding to this AI revolution by doubling down on senior talent. Their reasoning seems sound at first glance: current AI tools operate at a junior to mid level, so we need senior engineers to maintain quality and guide development while more junior roles are made obsolete. I’ve been wrestling with this logic, and I’m increasingly convinced it’s optimizing for a world that won’t exist in 18-24 months.

Think about what happened after FORTRAN. The skills that made someone an excellent assembly programmer didn’t disappear overnight, but they became progressively less relevant. New developers who learned to think in higher-level abstractions had advantages over those who had to unlearn old patterns or were resistant to the industry shift. We’re facing a similar transition now, but likely at an even faster pace.

The suggestion that junior talent might have advantages in adapting to AI-augmented development seems counterintuitive at first glance. But this isn’t just about learning to use new tools – it’s about preparing for a future where the very concept of software engineering as we know it may become obsolete. As AI shows increasing capability to generate, optimize, and even architect code, we’re approaching another fundamental redefinition of our field.

What happens when AI can handle not just the coding, but the higher-order decision-making that we currently consider the domain of senior talent? Once AI can design architectures, optimize trade-offs, and generate complete software systems, what will the role of a human engineer become?

I don’t think anyone knows yet. But our chosen historical parallel might offer some insight. During the transition from assembly to high-level languages, many experienced programmers initially struggled to adapt their thinking to new levels of abstraction. The mental models that made them excellent assembly programmers sometimes worked against them when dealing with compilers and higher-level concepts. Meanwhile, newcomers to the field, unburdened by years of optimization techniques and platform-specific knowledge, often adapted more readily to thinking in these new abstractions.

The instinct to prioritize deep technical expertise and architectural knowledge in hiring – skills that AI is rapidly mastering – may be a fundamental misreading of where our industry is headed. The truth is, we’re no longer hiring for the skills we know we need today. We’re hiring for the ability to adapt to a landscape we can barely imagine.

What we’re discovering is that agency – the capacity to act independently, make decisions, and drive outcomes – is becoming far more valuable than raw intelligence or years of experience. In a world where AI can provide vast knowledge and technical skill on demand, the engineers who thrive won’t necessarily be the smartest or most experienced, but those who can effectively exercise agency in directing these powerful tools toward meaningful outcomes.

This shift from valuing intelligence to valuing agency represents a profound change in how we should evaluate talent. For decades, we’ve selected engineers based on their ability to solve complex technical problems independently – essentially, for their intelligence and knowledge. But in an AI-augmented world, the ability to identify which problems are worth solving, to frame those problems effectively for AI assistance, and to synthesize AI outputs into coherent solutions becomes paramount.

The most valuable traits in this new era might be ones we’ve traditionally undervalued: intellectual flexibility, comfort with ambiguity, and the ability to rapidly synthesize new ways of working. These aren’t necessarily skills that come from years of coding experience. In fact, deep expertise in current methodologies might even be a hindrance if it creates resistance to radical changes in how we approach problems.

This suggests that the skills we should be hiring for aren’t the ones we’ve traditionally associated with senior engineers – deep technical knowledge, architectural design abilities, or algorithm expertise. Those aspects of our work are precisely what AI is poised to handle. Instead, we need people who can:

  • Exercise strong agency by identifying opportunities, setting direction, and driving outcomes
  • Think flexibly about problem-solving beyond current programming paradigms
  • Adapt quickly to new tools and methodologies without being constrained by “how things have always been done”
  • Collaborate effectively with AI systems, understanding their capabilities and limitations
  • Navigate uncertainty and ambiguity while maintaining a clear vision of desired outcomes

These capabilities don’t necessarily correlate with years of experience. In fact, those with less ingrained habits and assumptions about software development might be better positioned to embrace these new ways of working. When comparing a brilliant, experienced engineer with weaker agency against someone with moderate technical skills but exceptional agency, the latter may ultimately deliver more value in an AI-augmented environment.

Looking Forward

We’re moving beyond just learning new tools or frameworks. We’re entering an era where the fundamental nature of software engineering – what it means to be an engineer – is being redefined. AI is pushing us into entirely new territory, where human agency and adaptation may matter more than traditional technical expertise or raw intelligence.

For engineering leaders, this suggests a counterintuitive strategy: instead of doubling down on senior talent, we should be investing in those who demonstrate strong agency and can help us navigate toward an unknown future. This doesn’t mean abandoning experience altogether, but rather recognizing that there may be more valuable skillsets we should be selecting for when hiring. Skills that experience can give, but not necessarily in the ways we’re used to measuring.