Educating the Agentic Engineer: Curricula, Collaboration, and Continuous Learning in the AI Era
Citation. Alenezi, M. (2026). Educating the agentic engineer: Curricula, collaboration, and continuous learning in the AI era [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2607.29610
Summary. This conceptual synthesis proposes the ACCEL framework for educating engineers who direct, verify, and remain accountable for agentic-AI work rather than only producing technical artifacts themselves. It defines five competency pillars—intent specification, orchestration and delegation, verification and critical evaluation, ethical governance and accountability, and adaptive self-directed learning—delivered through integrated curricula, hybrid collaboration, and continuing learning. Its central pedagogical claim is that AI autonomy should expand only as learners demonstrate the verification competence needed to supervise it.
Key claims
- ACCEL frames agentic-engineering capability as five interdependent competencies: specification and problem framing, orchestration and delegation, verification and validation, ethical governance and accountability, and adaptive self-directed learning.
- The proposed delegation–verification learning loop makes students specify intent, delegate with guardrails, inspect and validate agent output against acceptance criteria, integrate or re-delegate, and document the rationale; accountability checks gate irreversible actions.
- The proposed curriculum moves through four stages from restricted AI generation and foundational unassisted problem solving to supervised delegation and full multi-agent orchestration, with greater AI autonomy contingent on demonstrated verification competence.
- The paper argues that assessment should weigh specifications, delegation rationales, verification records, governance artifacts, and reflection more heavily than a final artifact that agents may increasingly be able to produce.
Evidence & limitations
- This is an integrative conceptual preprint, not a new empirical study or validated competency model; its framework and curricular prescriptions require independent implementation and outcome evidence.
- The paper synthesizes evidence from engineering education, computing education, human–AI interaction, human factors, and learning sciences, but its claims about the agentic-engineer archetype and ACCEL framework are the author’s conceptual integration.
- Public provenance was verified by downloading the public arXiv PDF; its SHA-256 exactly matches the ingested Drive PDF. The arXiv API was temporarily unavailable (HTTP 503), so submission metadata was confirmed from the PDF and public arXiv landing/PDF rather than the API response.