The Craftsmen Are Tired: How AI Is Forcing Software Engineers To Rethink Who They Are

Deedy Das dropped a blunt assessment on X over the weekend. “Most software engineers are facing an identity crisis bordering on depression,” the Menlo Ventures partner wrote. “The craft they loved is dead.”

His post cut through the usual venture chatter. It captured what many in the field have felt but hesitated to say out loud. Companies push AI coding tools hard in search of speed. Engineers end up reviewing, debugging and fixing what the machines spit out. The result feels less like creation and more like maintenance. Some call it botsitting. Others just call it exhausting.

Das described a split inside engineering teams. “Lazy” developers lean on AI for nearly everything. They ask it questions, generate updates and move on with minimal scrutiny. The craftsmen review everything. They carry the burden of understanding the system, spotting flaws and keeping quality intact. “The craftsmen are tired. Very tired,” he added. “The entire burden of review falls on the craftsman. The burden of understanding.”

This tension surfaces most in older, larger organizations with wide variance in talent. Yet it echoes across the industry. Business Insider detailed the pattern in a piece published today. Engineers once took pride in elegant, handcrafted solutions. Now many watch AI generate code at volume while they chase down hallucinations and subtle errors. The joy of building fades. Frustration builds.

And the data backs up the unease. Annie Vella, a distinguished engineer who researched the shift for her master’s work, found striking numbers. Seventy-seven percent of developers spend less time writing code than before. Nearly half believe their core coding skill will soon take a back seat to prompt engineering. Her analysis points to real changes already visible at scale.

Google reports AI generates more than 25 percent of its new code. Y Combinator CEO Garry Tan noted that about a quarter of the accelerator’s startups now see 95 percent of their code written by AI. The numbers sound like progress. For many engineers they feel like displacement.

The psychological toll runs deeper than lost lines of code. Developers train for deterministic thinking. Write the function. Test it. Expect the same input to produce the same output every time. AI works differently. It is probabilistic. Ask the same question twice and answers can vary. That unpredictability clashes with years of professional habit.

Stack Overflow’s 2025 developer survey captured the gap. Eighty-four percent of respondents use or plan to use AI tools. Trust in those tools fell to 29 percent, down 11 points from the year before. Hallucinations create extra verification work. Job security fears add another layer. The result is a quiet anxiety that hovers over daily tasks.

Code quality concerns compound the problem. A GitClear study of 211 million lines of code showed troubling trends after AI adoption spread. Copy-pasted code rose 17.1 percent, the first time it outpaced refactored code. Duplicated blocks jumped eightfold. Code churn increased 26 percent, with more changes revised or deleted within two weeks. Maintenance burdens grow even as initial output accelerates.

Teams already show signs of restructuring. Some organizations move toward smaller groups built around a product manager, a designer and a senior architect who directs AI prototypes. Junior roles focused purely on implementation shrink. The shift leaves mid-career engineers wondering what their next decade looks like. Senior voices must verify machine output while mentoring others who may never write core logic from scratch.

But not every signal points to decline. Citadel Securities noted in its February 2026 analysis that software engineer job postings rose 11 percent year over year despite widespread AI investment and displacement talk. Unemployment sits near historic lows. Demand for certain skills persists even as tools change daily work.

Engineers who adapt appear to gain leverage. They focus on systems thinking, intent definition and evaluation of AI suggestions. The role moves from pure implementation toward orchestration and judgment. Some compare it to the transition from hand-weaving to machine oversight during earlier industrial changes. The craft evolves. It does not vanish.

Vella argues the pendulum swings between hands-on building and higher-level management. Engineers have seen similar swings before. Those who master both ends gain an edge. They keep enough technical depth to catch AI mistakes while developing the communication and architectural skills that machines struggle to replicate.

Recent discussions on X reinforce the divide. Posts from May and June 2026 describe developers hitting an “identity crisis” after watching AI fix in seconds what once took hours. Others note AI interviews still test LeetCode-style puzzles even as the day-to-day job emphasizes agent orchestration. Talent flows show top researchers moving between well-funded labs while some startups face retention issues tied to unclear direction in the new environment.

The identity questions extend beyond individual careers. Companies must decide how to measure engineering contribution when AI handles much of the typing. Promotion paths built on code volume or bug fixes lose meaning. Mentorship becomes harder when juniors spend less time wrestling with fundamentals.

Yet opportunity sits inside the uncertainty. Engineers who treat AI as a collaborator rather than a threat report higher output on complex problems. They prototype faster, explore more architectures and spend time on novel challenges instead of boilerplate. The human contribution shrinks to roughly 30 percent of the final result in some AI-assisted flows, according to observations from Google’s Addy Osmani. That 30 percent centers on taste, context and accountability.

Das’s warning carries weight because it comes from someone who invests at the intersection of AI and enterprise software. He sees the frustration daily in portfolio companies. The craftsmen grow tired of carrying the load. If organizations fail to redistribute work or restore a sense of ownership, burnout will spread. Some engineers may leave the field entirely, taking hard-won intuition with them.

The industry has talked about AI transforming software development for years. The conversation has now reached the personal level. It is no longer abstract speculation about future job counts. It is about daily experience, professional pride and the quiet realization that the old definition of a software engineer no longer fits.

So the question hangs in engineering channels and team meetings. If the craft changes this much, what remains of the identity? Some answer by doubling down on fundamentals. Others lean into AI mastery and system design. A few step back and ask whether the current path still aligns with what drew them to code in the first place.

The craftsmen are tired. That much is clear. Whether they find renewed energy in the new shape of the work will decide the next chapter for software engineering.