AI Coding Tools May Turn Software Engineering Into a Short-Career Profession

Sean Goedecke argues that even if AI assistants erode engineers’ long-term technical skills, developers will still be forced to adopt them — because employers will pay for short-term productivity, not craft preservation. He draws a parallel to construction workers who lift heavy objects despite the cumulative damage to their bodies, or carpenters who can refuse power tools but won’t find paying work. Engineers who insist on writing everything by hand risk being outcompeted by those willing to trade cognitive depth for near-term output.

The deeper claim is that pre-2024, getting paid to code was also the best way to get better at coding — a fortunate coincidence rather than an inherent property of the field. AI breaks that link. If using AI degrades the ability to understand codebases over time, and the pace of model improvement keeps accelerating, the implication is that software engineering may become a time-limited career like professional athletics, with a roughly fifteen-year window before practitioners are pushed out.

Goedecke is skeptical of the common counter that engineers will simply shift up the stack and supervise AI output, arguing that comprehension itself atrophies once hand-coding stops. He’s also pessimistic that labor organizing can slow the shift, given tech’s high salaries and remote, globally-distributed workforce. The recommendation is sobering: plan for a shorter runway rather than assume the current arrangement persists.