The Quiet Grief of AI-Assisted Coding
There’s a feeling that experienced programmers aren’t talking about enough. It’s not burnout. It’s not impostor syndrome. It’s grief.
The Core Insight
Gergely Orosz from The Pragmatic Engineer recently shared something that resonated deeply with the developer community: the dawning realization that AI will write most of our production code going forward. And it’s not just about efficiency—it’s about loss.
“It took a lot of effort to get good at coding,” Orosz writes. “Learning how to write code that works, to read and understand complex code, and to debug and fix when code doesn’t work as it should.” Years of practice, learning from mentors, struggling through university courses, finally reaching a level of mastery—only to watch machines catch up in months.
Why This Matters
The grief is real because the craft was real. Those moments of being “in the zone,” balancing several ideas while typing, hitting compile, and seeing it work—those weren’t just productive sessions. They were peak experiences. They’re what made many of us fall in love with programming in the first place.
Now AI handles the implementation while we… direct? Supervise? The dopamine hit of solving a hard problem yourself gets replaced by the cognitive load of reviewing AI-generated code and asking “did it get this right?”
This isn’t a complaint about AI being bad. It’s an acknowledgment that something valuable is being taken away—the joy of struggle, the satisfaction of mastery, the identity we built around being people who can make computers do complex things through sheer mental effort.
Key Takeaways
The loss is legitimate: You’re not being a luddite if you miss writing code. The craft had intrinsic value beyond just producing working software.
Identity shift is hard: For many developers, “I can code” wasn’t just a skill—it was core to who they were. AI forces a renegotiation of that identity.
The “zone” may evolve: Orosz speculates that being “in the zone” might shift to higher-level problem thinking while instructing complex code to be written. Whether this provides the same satisfaction remains to be seen.
Love-hate was always real: Even before AI, the relationship with coding was complicated—the focus required, the time pressure, the estimates that never worked. AI adds a new dimension to existing tensions.
Looking Ahead
The question isn’t whether AI will write most of our code—that ship has sailed. The question is what replaces the satisfaction we used to get from the craft.
Maybe we become architects and reviewers. Maybe we focus on the parts AI still struggles with—system design, understanding user needs, navigating ambiguity. Or maybe we collectively need to process this grief before we can figure out what comes next.
One thing is clear: pretending there’s no loss doesn’t help anyone. The first step to adapting is acknowledging what we’re adapting from.
Based on: “The grief when AI writes most of the code” by Gergely Orosz