I’ve asked the same interview question for the past 20 years. The candidate is presented with two very short, very similar-looking lines of code and asked to explain the difference.

The failure rate has been consistently above 95 percent. It’s a simple problem, but one that requires a candidate to have been willing and motivated to develop an understanding of how the operating system works to know the solution.

In my view, this kind of understanding is quickly becoming the single most important differentiator in a world where AI tools can seemingly turn anyone into a software developer overnight. As these tools proliferate, they may create a generation of architects who can assemble a building but struggle to lay a foundation.

In the traditional process of software development, an architect oversees a team of coders and developers, who generate, test, and iterate code to refine a product or software solution over the span of weeks or months. Today,  with the advanced AI tools available, that process can occur in a fraction of the time, with fewer people involved, and less reliance on highly experienced personnel. As a result, it may be that some less experienced developers may have fewer opportunities to naturally grow into the architect role over time.

The role of a software architect has similarities with that of a building architect. Both must consider the structural integrity, form, and function of their creation. AI, as good as it currently is, cannot provide this level of holistic oversight. Yet. So, we must be careful, while enjoying the productivity benefits of AI tools, to maintain a focus on what the experienced human can bring to a project.

This dynamic amplifies the so-called 10X engineer paradigm, first identified in productivity studies from the 1960s, which suggested top engineering talent could be many times more effective than the average. One school of thought is that as AI automates the more routine coding tasks, the value of average skills will diminish. Conversely, the value of those with deep, architectural insight will skyrocket.

Given this changing landscape, developers starting out on their career journey may have to rely more on their own curiosity and passion for coding to develop these qualities of knowledge and experience over time, rather than assimilating them from the traditional development process. I believe those graduates that have developed the natural curiosity to explore beyond the standard coding curriculum can be especially impactful given the power and ubiquity of the new tools available.

This brings me to the other important question I ask in interviews. After the technical grilling, I ask candidates to describe the project they are most proud of. I have them go into technical depth, to test their passion rather than knowledge.

Ultimately, I see passion and curiosity as the engines of deep understanding. They are what may drive a person to seek out why those two lines of code behave differently. As we enter a new future for software development, that may matter more than ever.