Study Shows: AI Agents Won't Replace Software Engineers, Instead Expand Their Work Boundaries

A research team from Chalmers University of Technology and Volvo Group recently published a paper stating that AI agents will not cause software engineers to lose their jobs; instead, they can expand their work scope through the 'semi-trustworthy execution stack' model.

The model includes six links: classic code, prompts and natural language specifications, agent workflow orchestration, control systems, operational organizational logic, and social and institutional adaptation. Currently, engineers mainly focus on the first two links, while the latter four (like agent workflow) are becoming priority engineering objects.

Researchers believe that AI does not need to reach the level of top experts; large-scale deployment has higher value. At the same time, AI hallucination issues highlight the need for testing and monitoring, so humans still have irreplaceable value in the future.