evidoria

← Back to browse

Good practice Imported

University of Guyana's AI-in-Education Policy — Dual-Track Grading and Mandatory AI Disclosure

Guyana · Georgetown · See the Guyana profile · See the Georgetown profile

Evidence: Descriptive / self-reported Top 28% 60/100 · Ask Evidence Copilot about this practice

Guyana's national university requires students to disclose generative-AI use, grades AI-assisted and unassisted work on different rubrics, and flags its own Turnitin AI-detector as unreliable rather than treating it as proof of cheating.

25%
Turnitin AI-detector reported failure/false-positive rate
30%
Minimum share of identity-verified assessments recommended

Details

Maturity
Established
Promoter
University of Guyana
Period
August 2023–present
Keywords
higher education, academic integrity, AI governance

Context

Effective August 2023, the University of Guyana adopted a formal AI in Education Policy applied campus-wide and reviewed every six months by an AI Working Group led by Dr. Morris-Martin, drawing volunteers from the Faculty of Education and Humanities and the Inclusion Unit.

Objectives

The policy requires students to disclose generative-AI use and grades AI-assisted and unassisted work on different rubrics, while explicitly guarding against over-reliance on unreliable AI-detection tools.

Activities

The policy's own example rubric grades work differently depending on whether generative AI assisted it -- e.g. Argument scored 5/8, Discussion 10/12, Citation 5/7 for AI-assisted versus unassisted submissions. Students must sign a clause permitting the university to analyse their work with plagiarism and AI-detection tools, and the policy recommends that at least 30% of a programme's assessments carry full identity-verified components, such as proctored exams or presentations. It also includes accommodations letting students with disabilities use assistive AI tools.

Conclusions

The policy explicitly warns that Turnitin's AI-detection plug-in is 'not 100% reliable,' citing a 25% failure/false-positive rate drawn from its own cited sources, and states it should not be the sole grounds for an academic-misconduct finding. The policy's real-world operation is independently corroborated by a March 2025 peer-reviewed case study (Bissessar, Equity in Education & Society, SAGE) surveying University of Guyana graduate students' generative-AI use and plagiarism perceptions.

Implementation
Indicative cost
Low (< €50k) — Volunteer-staffed working group; no budget figures disclosed in the source material.
Time to results
Short (< 1 year) — Effective August 2023; reviewed every six months by the AI Working Group.
Staffing & skills
AI Working Group led by Dr. Morris-Martin, volunteers from the Faculty of Education and Humanities and the Inclusion Unit

Conditions for success

  • Clear student disclosure requirement for generative-AI use
  • Dual-track grading rubric distinguishing AI-assisted from unassisted work
  • At least 30% of assessments carrying identity-verified components
  • Explicit caution against relying solely on an AI-detector for misconduct findings
  • Disability accommodations for assistive AI tools

Common failure modes

  • Turnitin's AI-detector carries a 25% failure/false-positive rate per the policy's own cited sources, risking wrongful misconduct findings if used as sole evidence

Where it fits

Governance type
university-wide academic policy
Scale
single national university (Guyana)
Income level
n/a

Replication kit

Reusable artefacts from this practice — as published by their sources.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful