Ethics, Equity, and Academic Integrity in the GenAI Classroom: A Framework for Higher Computing Education

Generative AI's Ethical and Societal Impacts in Higher Computing Education

Generative AI is reshaping how we teach and learn computing. Alongside the benefits, it raises real questions about equity, academic integrity, bias, and who holds decision-making power in the classroom.

A new systematic review synthesizes research and university policies to propose an Ethical and Societal Impacts-Framework for higher computing education. It's a practical lens for course design, assessment, and institutional policy.

What the systematic review found

• Out of 94 review studies identified, only six centered on GenAI in higher education, and just three addressed ethics directly - a clear gap.
• A 71-paper analysis showed heavy emphasis on tool capabilities while grounding ethical concerns in the ACM Code of Ethics.
• In 21 papers, teachers used LLMs for assignment generation and evaluation; students used them as on-demand tutors.
• Key risks: academic integrity, biased outputs, unclear data provenance, over-reliance leading to shallow learning, and fairness concerns with AI detection tools.
• Key opportunities: multilingual support, culturally relevant content, and more accessible learning materials.
• Equity gaps are growing as access to high-quality tools and paid tiers varies widely across students and institutions.

Why this matters for educators

• Assessment integrity and feedback workflows are changing fast.
• Detection tools can misfire; fair process and transparency matter.
• Policy, consent, and data handling must be explicit - especially around student data and third-party tools.
• Faculty development and student onboarding are now essential parts of course setup.

Ethical and Societal Impacts-Framework (practical view)

• Academic integrity: What proof of learning will you accept (process logs, oral checks, version history)? What's permitted vs. prohibited?
• Equity and access: Do all students have comparable access and bandwidth? Are there low-cost or offline alternatives?
• Bias and fairness: How will you surface and correct biased outputs? Are students trained to critique model responses?
• Data provenance and IP: Do students understand training data limits, licensing, and citation for AI-assisted work?
• Transparency and agency: Are tool limitations and risks stated up front? Can students opt for non-AI paths?
• Assessment and pedagogy: Are tasks authentic, iterative, and hard to auto-complete? Do they reward process as much as product?
• Policy and governance: Is there a clear course policy that aligns with institutional rules and the ACM Code?

Assessment and academic integrity

• Require visible process: commit history, prompt logs, drafts, and reflection memos.
• Use oral check-ins, pair programming, and timed in-class builds to verify individual competency.
• Shift some grading weight to design decisions, testing, debugging, and code review quality.
• Avoid over-reliance on detectors; use them as signals, not verdicts. Provide an appeals path.

Equity and access

• Provide institution-level access to approved tools so students aren't forced into paywalls.
• Offer low-bandwidth options, offline prompts, and text-only workflows for constrained environments.
• Encourage multilingual prompts and localized examples to improve inclusion.

Data, privacy, and consent

• Do not require students to upload personal or proprietary data to third-party systems.
• Disclose data-sharing terms for any tool used. Provide a non-AI alternative path.
• Run "failure mode" drills: What happens if a tool is down, biased, or wrong? What's the fallback?

Faculty and policy playbook

• Publish a clear AI use policy in the syllabus with examples of acceptable and unacceptable use.
• Provide quick-start guides for prompts, critique checklists, and error-spotting strategies.
• Run short workshops for TAs on grading AI-influenced work and documenting process evidence.
• Review course policies each term with student reps; adjust based on actual classroom patterns.

AI and programming education research: how to organize your review

• Define the primary goal: trends, themes (e.g., automated assessment, student perceptions), or longitudinal shifts.
• Tag by keywords and cluster into higher-level areas: student support, AI-assisted coding, integrity, accessibility.
• Track redundancies and gaps to prioritize what truly needs study next.
• Publish a categorized summary that departments can reference when designing new courses.

Quick start: add these to your next syllabus

• AI use statement + examples.
• Process evidence requirements (logs, commits, reflections).
• Integrity procedure (what happens if concerns arise).
• Accessibility plan (tool access, low-cost options, multilingual support).
• Data policy (no personal data uploads; approved tools list; opt-out path).

Further reading

• ArXiv: Ethical and societal impacts of GenAI in higher computing education
• ACM Code of Ethics

Upskill your team

If your department is building AI literacy for faculty or TAs, structured programs help turn policy into daily practice.

• AI courses by job role