Artificial Intelligence Frequently Asked Questions

​

NATIONAL COLLEGE OF IRELAND

Artificial Intelligence
Frequently Asked Questions

Practical guidance for staff and students

Draft reference guide

August 2026
 

About this guide

This guide brings together practical questions on the responsible use of artificial intelligence across NCI. It should be read with the Artificial Intelligence (AI) Policy, Responsible Use of AI Procedure and Approved AI Tools and Services Register.

The Register is the authoritative source for the current approval status and conditions of individual tools. Where a use requires evaluation, submit an AI Tool and Use-Case Evaluation Request through the IT Helpdesk.

Quick rule: If the proposed use involves personal or confidential information, transcription, integration, automation or a significant decision affecting a person, do not proceed until you have confirmed that the exact tool and use are approved.

Contents 

1. General AI Use - approved tools, data, disclosure, integrations and incidents
2. Student Learning and Assessment
3. Teaching, Marking and Academic Integrity - materials, assessment, feedback and AI detection
4. Research - research data, writing, transcription, peer review and authorship
5. Human Resources - recruitment, employee information and employment decisions
6. Coding and Digital Content - coding assistants, generated code and media

1.General AI Use 

Which AI tools can I use?

Check the Approved AI Tools and Services Register before using an AI tool for NCI work. Approval applies only to the specific product, license tier, configuration, purpose and data identified in the Register.

Can I use a free or personal AI account?

Yes, but only for public information or Routine Low-Risk Institutional Content. Personal or consumer accounts must not be used for personal, confidential, commercially sensitive or other Protected College Data.

Does paying for an AI tool mean it is approved?

No. A paid subscription does not mean that the tool has been evaluated or approved by NCI. Procurement, installation, integration and processing of Protected College Data require the appropriate approvals.

What should I do if the tool I want to use is not listed?

Check whether it falls within a permitted Level 1 category in the Register. If approval is required, submit an AI Tool and Use-Case Evaluation Request through the IT Helpdesk.

How do I know whether my proposed use requires approval?

Approval is normally required where the use involves Protected College Data, transcription, repeated controlled processing, procurement, installation, integration, automation or a significant decision affecting a person. If in doubt, submit an AI Tool and Use-Case Evaluation Request through the IT Helpdesk.

Can I trust information produced by AI?

Not automatically. AI can generate inaccurate information, invented references, biased content and convincing but incorrect conclusions. Users are responsible for checking important claims against reliable sources.

Can I use AI-generated references or citations?

Only after locating and checking the original source. AI tools may invent publications, authors, quotations, page numbers and links. An unverified AI-generated citation must not be included in academic or professional work.

Do I need to disclose that I used AI?

Material AI use should be recorded and disclosed where required. Routine spelling, formatting or low-risk drafting assistance does not normally require disclosure unless a research, contractual or local rule says otherwise.

Can I upload someone else's unpublished or non-public work to an AI tool?

No, not without the person's knowledge and appropriate authority from the relevant author, rights holder or institutional owner. This includes unpublished student work, staff documents, research papers, draft publications and other original material.

Can I install an AI application, agent or browser extension?

No: Not on an NCI-managed device without IT approval. Browser extensions and agents may access webpages, documents, email, credentials or other information beyond what the user intends to share.

Can I connect an AI tool to my email, OneDrive, Teams or another NCI system?

No: Not without approval. Connectors and integrations may give the AI system access to substantial amounts of College Data and must undergo technical, security, privacy and governance evaluation.

Can I use meeting transcription or AI-generated meeting notes?

This is currently under review and should not be used until a determination is made by the AI Governance Committee.

Can I publish AI-generated content on behalf of NCI?

Only after appropriate human review and normal communications or business approval. The reviewer must check accuracy, tone, accessibility, confidentiality, copyright and whether the AI contribution should be disclosed.

What should I do if I accidentally enter protected information into an AI tool?

Stop using the tool and report the incident promptly. College protected data or Personal-data incidents should be reported to the DPO, security concerns to IT and suspected policy breaches to the AI Governance Committee. Do not assume that deleting the conversation has removed the information from the provider's systems.

2.Student Learning and Assessment

Please see the Acceptable Use of Generative AI in Teaching, Learning and Assessment Policy 

And guidance provided by the library: Use of AI in Teaching and Learning Student Guide

3.Teaching, Marking and Academic Integrity

Can lecturers use AI to prepare teaching materials?

Yes: For drafting, brainstorming, examples and lesson planning using public or Routine Low-Risk Institutional Content. Academic staff remain responsible for accuracy, accessibility, copyright, academic quality and alignment with learning outcomes.

Can AI generate quiz or assessment questions?

It may assist with generic ideas or draft questions. Academic staff must validate the difficulty, accuracy, fairness and alignment of the final questions.

Can lecturers upload student work to AI for marking or feedback?

No: Student work contains Protected College Data and original unpublished content. AI must not replace academic judgement or the approved marking process.

Can AI determine a student's mark?

No: A significant academic or assessment decision may not be based solely on AI. The authorised academic marker and examiner remain responsible for applying the marking criteria and determining the final result.

Can AI-detection software prove that a student used AI?

No. AI-detection results must not be treated as conclusive evidence of misconduct. Concerns should be considered through dialogue, assessment evidence and the established academic-integrity procedure. Please see the Academic Misconduct Procedure

Can staff design an assessment that requires students to use AI?

Yes: Where AI use supports the learning outcomes and has received the necessary academic, technical and governance approvals. Students must receive clear instructions on permitted use, disclosure, data protection and available alternatives.

4.Research

Can I use AI to help with a literature review?

Yes:  For tasks such as generating search terms, identifying themes or structuring an approach, using public or lawfully accessible information. Researchers must conduct and verify the actual literature search and must not rely on AI-generated references.

Can AI help draft a research paper or funding application?

Potentially, subject to NCI, funder, publisher and research-integrity requirements. The researcher remains responsible for the intellectual contribution, accuracy and authorship and must disclose material AI use where required.

Can I upload research-participant data?

Only where the exact tool and research use have been approved. Research-participant data may require research ethics approval, data-protection screening, DPO consultation and a DPIA. Data should be anonymised wherever possible.

Can I use AI to transcribe research interviews?

Only following appropriate evaluation and approval. Participants must receive the required information, the lawful basis and research consent arrangements must cover the processing, and access and retention must be controlled.

Can I upload a confidential manuscript, thesis or grant application that I am reviewing?

No: Not without appropriate authority and confirmation that the relevant review, funder or publisher rules permit it. Peer-review and unpublished materials are confidential and may also be protected by copyright.

Can AI be named as an author of research?

No. AI cannot take responsibility for the integrity of research or meet the normal requirements for authorship. Human authors remain accountable for all content and must acknowledge material AI assistance where required.

5.Human Resources

Can I use AI to draft a disciplinary letter?

Yes, but only as a drafting aid. You may use AI to create a generic structure or improve wording without entering identifiable employee or case information. If personal, disciplinary or other confidential information is required, use only a system approved for that specific purpose and data category. HR and the responsible manager must review the letter, and AI must not determine the disciplinary outcome.

Can I upload employee data into an AI tool?

Only where the exact AI system, configuration, purpose and type of employee data have been expressly approved by NCI. Employee information is Protected College Data and must not be entered into consumer or personal AI accounts. Use only the minimum necessary data and anonymise or pseudonymise it where possible. Data-protection screening and DPO consultation may also be required.

Can I use AI to summarise interview notes?

Only where the tool and recruitment use case have been approved to process applicant data. The summary must be checked against the original notes by a member of the interview panel. AI must not be used to score candidates, rank applicants, infer personal characteristics or recommend an appointment.

Can AI help screen CVs?

Not as a routine or unapproved use. AI screening, scoring or ranking of applicants may be a high-risk employment use under the EU AI Act. The use requires formal evaluation and may require AI Governance Committee, data-protection and equality review, a DPIA and appropriate applicant information. Recruitment decisions must always include meaningful human review.

Can I use meeting transcription for HR investigations?

Not unless the transcription tool and the specific HR investigation use have been approved. Microsoft Teams transcription and Intelligent Recap are currently listed as under review in the Approved AI Tools and Services Register. Any future approved use would require notice to participants, a documented lawful basis, restricted access, approved retention and DPO advice or a DPIA where required. The transcript must be checked and must not replace the formal investigation record or human judgement.

Can I use AI to draft performance review comments?

Yes, for wording and structure, provided the comments reflect the manager's own assessment. Generic or effectively anonymised information may be used in a tool permitted for low-risk drafting. Identifiable employee information may only be entered into a system approved for the relevant purpose and data category. AI must not independently evaluate performance, allocate ratings or recommend employment action.

Can AI make an employment or disciplinary decision?

No:  A decision about an applicant or employee may not be based solely on AI. The accountable manager must make the decision and provide meaningful human review, explanation, correction and appeal where applicable.

Can I use AI to create generic HR templates?

Yes. AI may be used to draft generic letters, policies, interview questions or guidance where no personal, confidential or case-specific information is included. The user must review the output for accuracy, fairness, legal compliance and suitability before use.

Can I use AI to draft a job description or job advertisement?

Yes, provided no personal or confidential information is included. HR must review the final content to ensure that the requirements are accurate, inclusive and free from discriminatory or biased language.

Can AI generate interview questions?

Yes, as a drafting aid. HR and the interview panel must review the questions to ensure they are relevant to the role, consistently applied and compliant with employment and equality requirements.

Can AI score interview answers or rank candidates?

No: Not without formal evaluation and approval. Scoring, ranking or recommending candidates may constitute a high-risk employment use. AI must not make or materially determine a recruitment decision.

Can I paste an employee's email into an AI tool to help draft a response?

No: Not into a consumer or unapproved AI tool. An employee's email may contain personal data, confidential information and original unpublished content. Where AI assistance is required, remove identifying details or use an approved system in accordance with its conditions.

Can I use AI to draft a response to a grievance or complaint?

Only as a carefully controlled drafting aid. Case details must only be entered into a system approved for the relevant purpose and data category. HR and the accountable manager must verify the accuracy, fairness and tone of the response and approve the final decision.

Can I upload occupational health or medical information?

No: Not without specific approval. Health information is special-category personal data and requires enhanced protection. The DPO must be consulted, and data-protection screening or a DPIA may be required. Only the minimum information necessary should be processed.

Can AI analyse employee absence or attendance records?

No: Attendance and absence records are Protected College Data. AI must not be used to profile employees or recommend employment action.

Can I use AI to summarise staff survey comments or analyse anonymised employee feedback?

Yes: Only where the information has been effectively anonymised so that individuals cannot reasonably be identified from the text or context. Removing names alone may not be sufficient. Raw or identifiable responses require an approved system and appropriate data-protection controls.

Can AI help prepare HR reports or workforce analysis?

Yes, but only within an approved use and using the minimum data necessary. Reports must be checked for accuracy, bias and misleading conclusions. Profiling, predictions or decisions affecting employees based on AI is not allowed.

Can I use AI to translate an HR communication?

Yes, for generic communications that contain no personal or confidential information. A competent person should check important translations before they are issued.

Can I use my personal ChatGPT, Claude or other AI account for HR work?

It depends: Only for public information or Routine Low-Risk Institutional Content that contains no employee, applicant or other Protected College Data or Personal Data.

Do I need to tell an employee that AI was used?

Routine drafting assistance does not normally require a separate declaration.

Can AI take minutes at an HR meeting?

This is currently under review and should not be used until a determination is made by the AI Governance Committee.

Can AI identify employees who may be underperforming or likely to leave?

No: Predicting performance, behavior or retention may involve profiling and could significantly affect employees. It should not be used.

Can AI draft a reference for an employee?

Yes: As a wording aid, provided the information is accurate and the system is approved to process the relevant employee data. The authorised referee remains responsible for the content and must personally review and approve the final reference.

6.Coding and Digital Content

Can I use an AI coding assistant?

Yes: for generic, public or open-source coding assistance within the Register conditions. Do not enter credentials, security details, non-public system architecture, proprietary code or restricted repositories unless the exact use is approved.

Can I use AI-generated code without checking it?

No. Generated code must be reviewed and tested for functionality, security, licensing, accessibility and maintainability before it is used.

Can I use AI to create images, audio or video?

Yes: for appropriate low-risk purposes, subject to copyright, privacy and disclosure requirements. Do not create deceptive content or use an identifiable person's image, voice or likeness without appropriate authority.

Related NCI Documents

Artificial Intelligence (AI) Policy: sets NCI's principles and mandatory requirements.  

Responsible Use of AI Procedure: sets the evaluation, approval, control, monitoring and reporting process.

Approved AI Tools and Services Register: records current tool status, permitted uses, data conditions and evaluation routes.

Status note: AI services, supplier terms and institutional approvals can change. Always consult the latest controlled documents before using an AI tool.

Version Control