AI agents aren't replacing software engineering but expanding it far beyond code, researchers argue

The research comes from an industry-academic collaboration with Volvo Group, meaning the findings are grounded in how a large industrial software organization actually deploys agents today, not a theoretical model. That practitioner grounding is what separates this from the usual think-piece about developer futures.

This is largely disconnected from recent activity in our archive, as we have no prior coverage to anchor it to. It belongs to a broader ongoing debate about AI's effect on knowledge work, sitting alongside research on agent autonomy and the changing definition of 'engineering' as AI handles more implementation detail. The relevant intellectual neighborhood includes studies on how developers already use copilot-style tools, where the consistent finding is that time shifts toward specification, review, and system design rather than disappearing. This paper appears to extend that line of argument to the more autonomous agent tier, which is a meaningful step up in scope.

Watch whether Chalmers or Volvo Group publish longitudinal data on how engineer headcount and role definitions actually changed over a defined period after agent adoption. A cross-sectional argument about scope expansion only becomes credible evidence against displacement if the job numbers hold up over 18 to 24 months.