Coding Errors in Leaving Certificate Grading Algorithm Caused Inaccurate Scores in Ireland

Errors in Irish Department of Education's algorithm to calculate students’ Leaving Certificate exam grades resulted in thousands of inaccurate scores.

Risk classification

• Primary risk domain: 7 AI system safety, failures, & limitations
• Primary risk subdomain: 7.3 Lack of capability or robustness

The custom-built grading algorithm failed to perform reliably due to coding errors, leading to incorrect grade calculations for thousands of students.

Additional risk subdomains

• 7.4 Lack of transparency or interpretability: The grading algorithm's code was proprietary and not open source, and the system suffered from 'inscrutability', meaning it could not justify why a specific grade prediction was made.

Causal factors

• Entity: AI
• Intent: Unintentional
• Timing: Post-deployment

The grading errors were caused by unintended software bugs in the deployed grading algorithm, which produced incorrect grade calculations for thousands of students.

EU AI Act risk tier

• Risk tier: 2 High Risk

High Risk: The system is an educational grading algorithm used to determine high-stakes Leaving Certificate results, which directly affects students' access to higher education, fitting the EU AI Act's definition of high-risk AI systems in education.

AI system and alleged parties

• AI system: Leaving Certificate calculated grades system
• AI purpose: Automatic Skill Assessment; Resource Allocation
• Behaviour type: Tool
• Alleged developer: Irish Department of Education and Skills
• Alleged deployer: Irish Department of Education and Skills
• Alleged harmed parties: Leaving Certificate exam takers, Irish Department of Education and Skills

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Substantial. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Negligible, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Substantial, indirect Substantial
• Civil rights: direct Minor, indirect Minor
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Minor
• Epistemic: direct Negligible, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Differential treatment

Reported: The report explicitly describes unfair and incorrect grading outcomes affecting thousands of students.

Directly caused: The algorithm directly assigned incorrect grades to over 13,000 students, with some receiving lower grades and others receiving higher grades than they should have.

Indirectly caused: Some students potentially lost university places they rightfully earned to peers who received incorrectly inflated grades.

Civil rights

Reported: The report explicitly describes how the grading errors impacted students' academic outcomes and access to higher education.

Directly caused: The algorithm directly compromised students' rights to a fair and accurate academic assessment, which is critical for their educational progression.

Indirectly caused: The resulting grading chaos disrupted the fair allocation of university places, impacting students' right to equal access to higher education.

Psychological

Reported: The report explicitly describes emotional and life-affecting distress caused indirectly by the grading errors.

Indirectly caused: The grading errors caused significant distress, frustration, and anxiety for over 6,000 students who received lower grades than they deserved, disrupting their immediate life plans.

Inferred additional harm: Widespread psychological stress and anxiety likely affected the broader cohort of students and their families due to the systemic uncertainty and perceived unfairness of the automated grading process.

People affected

• Occurrences reported: 1
• People reportedly harmed: 6000
• People reportedly exposed: 13943

Potential causes

Management

• Severe Time Constraints: Urgency of the situation prevented thorough testing of the system.
• Resourcing Constraints: Limited resources impacted the ability to adequately develop and test code.

Technology

• Coding Errors in Algorithm: A single line of code contained two errors affecting predicted grades.
• Custom-Built System Lack of Maturity: The software was custom-built and not previously tested or used elsewhere.
• Algorithm Inscrutability: The system could not explain or justify its grade predictions.

Data Inputs

• Incorrect Subject Inclusion: CSPE subject results were wrongly included in the grading calculations.
• Incorrect Subject Selection: Code substituted the worst two subjects instead of the best two.

Human Factors

• Developer Coding Mistakes: Human error led to two critical bugs in a single line of code.

Process and Methods

• Insufficient Software Testing: The grading code was not sufficiently tested before release.
• Lack of Code Transparency: The code was not open source or reviewable by external experts.

Regulatory Environment

• Lack of Algorithmic Transparency Laws: Absence of regulations requiring corporate algorithm transparency.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide clear, consistent, and detailed factual accounts of the grading algorithm's errors, the developer involved, the scale of the impact, and the resulting societal disruption. There are no major conflicting accounts.
• Ambiguities identified: The exact code and implementation details of the algorithm were not made open source, preventing independent public review.
• Alternative interpretations: None. The incident is clearly a software failure in a high-stakes public grading system.

An unintentional coding error in Ireland's national grading algorithm led to incorrect results for thousands of students, disrupting university admissions and highlighting governance challenges in automated public decision-making. The incident possesses minor national security implications, primarily affecting government service delivery and public trust.

• Overall national security impact: Minor
• Scope: Single nation
• Primary target: Ireland
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Long-term. The incident represents a policy and algorithmic governance challenge rather than an active national security crisis.
• Autonomy: Human-controlled. The system functioned as a deterministic tool to calculate grades based on pre-defined formulas, with final distribution managed by government officials.
• Novelty: Evolved capability. While software bugs are common, using a custom automated grading algorithm to replace national exams at this scale represented an evolved policy challenge.

Impact by dimension

• Physical security: Negligible. No physical systems, critical infrastructure, or kinetic safety mechanisms were affected by this software grading error.
• Information security: Negligible. No intelligence compromise, classified data theft, or foreign information operations were associated with this incident.
• Sovereignty: Minor. The incident temporarily disrupted the execution of a key government function, specifically the national education assessment and university placement system, due to software errors.
• Economic security: Negligible. There was no strategic technology theft, financial market manipulation, or direct threat to national economic competitiveness.
• Societal stability: Minor. The grading errors impacted academic outcomes and caused significant public distress for thousands of students, but did not threaten overall societal stability or civil liberties.