The debate over AI safety is increasingly framed as a technical problem, but a recent IEEE Spectrum article argues that the real failure is sociotechnical. Engineers under pressure to move fast may not fully assess risks, and a growing number of researchers—including Anthropic CEO Dario Amodei—have called for industry to slow down. That tension, the article contends, points to a deeper gap: engineering education has not taught students to weigh workplace pressures, ethical rules, and social consequences as core professional competencies.
Current accreditation and licensing standards do mention public welfare, but requirements are often generic and minimally enforced. Ethics instruction, where it exists, is usually relegated to introductory or non-engineering courses focused on codes of conduct and academic integrity, disjointed from the complex realities of engineering practice. Data cited in the article show that only about 4 in 10 US engineers recall receiving ethics training within actual engineering courses, and roughly 3 in 10 say they never received any training on public welfare responsibilities at any point in their careers.
The authors propose two complementary instructional approaches. One integrates one-hour sociotechnical modules into foundational engineering courses, such as an "Introduction to Circuits" module that uses a hospital power prioritization problem to link electrical concepts with ethical judgment. The other is a standalone course designed to prepare students as "public welfare watchdogs." The intent is to make clear that ethics is not an add-on but a required competency—though the authors note resistance persists, with some reviewers dismissing integrated ethics training as distracting from technical content. As they put it, responsible engineering requires more than technical expertise alone.