The Impact of AI on Older Programmers

In my view, the value of older programmers has bottomed out and is beginning to rebound.

Our team hires fresh graduates every year. In the past, it was striking how quickly they developed. Their technical career paths overlapped heavily with those of senior employees, but they had more energy, were more receptive to new technologies and frameworks, and commanded lower salaries. They outperformed older employees across virtually every dimension.

But things have changed over the past two years with the emergence of vibe coding.

The growth of recent graduate hires has clearly fallen short of what it used to be—or, more precisely, they are developing in the wrong direction. Their growth now resembles that of product managers and software architects, while they gain very little understanding or command of technical details.

As technology evolves through increasingly rapid iteration cycles, they have even less time to explore the finer details. Consequently, their weaknesses become increasingly pronounced.

Without developing a solid grasp of technical details, it is easy to get trapped in endless loops with AI. For roughly 80% of problems, junior and senior programmers perform at similar levels. For the remaining 20%, however, the efficiency gap can be dozens of times or more.

Junior programmers often make virtually no progress when they encounter such problems. Senior programmers, by contrast, can draw on past experience during architectural design and solution discussions to avoid a large number of potential issues. While interacting with AI, they can also use its feedback to continually deepen their understanding of the areas in which AI falls short.

These solutions are also highly dependent on their specific application contexts, making it difficult to consolidate them into reusable skills. This gradually creates a deadlock.

Once this kind of circular struggle begins, progress does not merely slow down—it comes to a complete halt. When it happens repeatedly, any delay in proactively reporting the problem can affect the entire schedule. The only option is to track progress closely and frequently. Whenever someone gets stuck, you have to investigate and determine whether they have fallen into another technical pitfall.

For me, this has substantially increased the workload. As a result, our team may not recruit anyone at all this year.

Senior employees, meanwhile, perform very consistently. Ironically, their greatest development over the years has been learning how to argue, negotiate, and wrangle through issues in natural language. They can now transition seamlessly from doing that with people to doing it with AI.

Their main weaknesses are limited energy and a tendency to take shortcuts. AI compensates for these shortcomings extremely well.

The rapid development of AI is, to a large extent, benefiting from the existing pool of experienced programmers. If AI continues advancing at this pace without leaving enough time to cultivate the next generation of talent, the future shortage of experienced professionals will only become more severe.

In the past, seniority offered relatively few advantages, while age was a major disadvantage. With AI assistance, however, the advantages of seniority have become substantial, while the disadvantages of age have become relatively minor.