Educators and Employers Diverge on How GenAI Should Shape Entry-Level Engineer Skills

An academic survey comparing four years of generative AI adoption shows a striking gap between how universities and employers are responding to the technology's role in software engineering.

Researchers surveyed 56 educators and 24 hiring professionals globally to understand shifting expectations for entry-level software engineers. The study, published on arXiv, reveals that while both sectors agree on which core competencies matter, they sharply disagree on whether graduates possess them.

Where Schools and Industry Diverge#

Universities frame their AI policies almost entirely through the lens of academic integrity—asking whether students should be permitted to use these tools at all. In contrast, companies have largely moved past that question. Industry adoption of GenAI is now near-universal, with organizational policies focused not on blocking the technology but on governing how teams should deploy it responsibly.

Interview practices remain unsettled. Neither educators nor employers have settled on clear standards for assessing GenAI competency during hiring assessments, and there is no consensus on whether candidates should actively demonstrate AI tool mastery.

Shared Ground on Essential Skills#

Despite their policy differences, both populations identified overlapping priorities:

• Critical evaluation of AI-generated output—the ability to spot errors and limitations in what tools produce
• Responsible and effective use of GenAI tools—applying them appropriately without over-relying on them
• Independent learning and adaptation—staying current as technology evolves rapidly

Both are shifting toward GenAI-resistant assessments that emphasize observable, real-time problem-solving and higher-order thinking rather than tasks that models can easily complete.

The Readiness Gap#

The study uncovered a notable disagreement on graduate preparedness. Hiring professionals reported significantly larger capability gaps than educators did—suggesting that schools perceive their students as better prepared than employers actually find them. This discrepancy is concentrated on foundational, non-AI-specific skills rather than GenAI-focused competencies, pointing to broader weaknesses in core engineering fundamentals.