Below is a collection of AI resources at Queen's, including guides for teaching and learning, prompt engineering, and more.
Centre for Teaching and Learning
Guidance for Teaching and Learning
Considerations and guidance around curriculum design in response to the ever-growing popularity of AI tools and Large Language Model (LLM) software.
Generative AI and Teaching
Read an open letter to the Queen's community on generative AI and teaching.
Update: the state of AI at 成人大片
Read a report on the results of a recent survey on AI use and needs at Queen's.
成人大片 Library
Library Collection of Resources for AI
AI is appearing in our everyday lives and in our research tools. What do we need to know to use and evaluate its outputs effectively? Learn more on the .
Library Guide on Prompt Engineering
Prompt engineering is the practice of creating and refining a description, question, and or instructions to enable an AI tool to construct the information most relevant to your topic and present it in a desired format.
Library Guide on Teaching and Learning
Read about the Queen's Library's recommendations for ethical AI use and citation analysis at Queen's.
Research and Research Administration (RARA)
Guidelines for the Responsible use of AI
Read through the RARA document of guidelines for the responsible use of AI
Student Academic Success Services
Student Academic Success Services has built an all-around resource for undergraduate learners called Academics 101, which has an informative section on AI.
You鈥檝e heard of , but it's only one of many generative artificial intelligence (GenAI) tools. These tools are built around large language models (LLMs). LLMs function similarly to the predictive text function you've likely encountered when using your phone, email, or internet search. Like predictive text, LLMs use mathematical modelling and probability to predict the next word or phrase in a sentence based on a given context.
- Dr. Gunnar Blohm provides tips on controlling and securing your research data while using AI.
- Jeff Glassford details the AI tools available at Queen's and their benefits.
- Dr. Kelley Packalen provides strategies for students and instructors to maintain academic integrity while using AI.
- Dr. Xiaodan Zhu describes human involvement in AI use at Queen's.
- Ricardo Smalling details the different types of data and how to use them alongside AI tools.
- Dr. Carolyn Lamb describes concerns surround AI's environmental impact and potential inequity and UNESCO's recommendations for mediating these concerns.
Sample Use Cases and Decision Making
Operations
These examples illustrate how common operational tasks can be supported by AI while maintaining appropriate safeguards for data, privacy, and institutional responsibility. Each use case models the reasoning process for selecting the right tool based on the sensitivity of the information involved, the workflow context, and the level of accountability required.
These tasks involve only information that is already public or intended for public release. External AI tools such as ChatGPT may be used, though enterprise tools such as LibreChat remain preferred.
How to read the table:
Each row describes four aspects of one use case:
Use Case and Sample Prompt: the name of the use case and a ready-to-use prompt you can adapt by filling in the bracketed placeholders.
Why AI Would Be Useful: the value AI adds to this task, including time savings and quality improvements.
Decision Logic for Tool Selection: the data sensitivity and workflow factors that determine which tool is appropriate.
Recommended Queen鈥檚 AI Tool: the specific tool, or category of tools, you should use for this work.
How to use the sample prompts:
Replace anything in [square brackets] with details specific to your situation.
Add the actual content (notes, data, draft text) the AI should work from. The prompt tells the AI what to do; you still supply the inputs.
Always review and verify the output before using it. AI drafts speed up your work; they do not replace your professional judgment.
Match the tool to the data sensitivity using the Decision Logic column.
Treat the sample prompts as starting points. The more clearly you describe the audience, tone, format, and constraints, the better the output.
| Use Case and Sample Prompt | Why AI Would Be Useful | Decision Logic for Tool Selection | Recommended AI Tool |
|---|---|---|---|
Public Event Promotion Brainstorm Sample prompt: 鈥淲e are promoting a public [event type] on [date] for [audience]. Please generate 10 catchy titles, 5 taglines, and 3 short social media posts (under 280 characters) that highlight [key message]. The tone should be [tone, e.g., warm, professional, playful].鈥 | Accelerates creative ideation by generating a wide range of titles, taglines, and potential promotional copy, helping teams test messaging quickly and strengthen audience engagement with minimal effort. | Inputs are limited to information already publicly available or intended for public release. No personal data, internal strategy, or operational information is included. Outputs are creative and non-sensitive, so enterprise security controls are not required. | External AI tools such as ChatGPT are acceptable. Enterprise tools such as remain preferred but are not required from a data protection perspective. |
Public Literature Summary for Background Reading Sample prompt: 鈥淧lease summarize the following publicly available sources on [topic] into a 1-page brief for [audience, e.g., a project team unfamiliar with the area]. Identify the 3 to 5 key themes, areas of agreement, and any disagreements between sources. List each source you drew from.鈥 | Quickly synthesizes large volumes of public material into accessible summaries and key themes, saving time while improving clarity and consistency of background materials for learners or stakeholders. | All source materials are publicly available and contain no institutional or personal data. Outputs are informational and intended for public use. Accuracy verification remains a human responsibility rather than a data security issue. | External AI tools such as ChatGPT are acceptable. Users must verify accuracy and attribution. |
How to read the table:
These tasks involve internal, personal, licensed, or otherwise sensitive university data. Work must stay inside institutional tools (LibreChat or M365 Copilot Chat under the institutional licence with single sign-on), with the exception noted for library-licensed content, which can be used only with AI features built into the vendor鈥檚 own platform.
Each row describes four aspects of one use case:
Use Case and Sample Prompt: the name of the use case and a ready-to-use prompt you can adapt by filling in the bracketed placeholders.
Why AI Would Be Useful: the value AI adds to this task, including time savings and quality improvements.
Decision Logic for Tool Selection: the data sensitivity and workflow factors that determine which tool is appropriate.
Recommended Queen鈥檚 AI Tool: the specific tool, or category of tools, you should use for this work.
How to use the sample prompts:
Replace anything in [square brackets] with details specific to your situation.
Add the actual content (notes, data, draft text) the AI should work from. The prompt tells the AI what to do; you still supply the inputs.
Always review and verify the output before using it. AI drafts speed up your work; they do not replace your professional judgment.
Match the tool to the data sensitivity using the Decision Logic column.
Treat the sample prompts as starting points. The more clearly you describe the audience, tone, format, and constraints, the better the output.
| Use Case and Sample Prompt | Why AI Would Be Useful | Decision Logic for Tool Selection | Recommended AI Tool |
|---|---|---|---|
Meeting Minutes to Action List Sample prompt: 鈥淏elow are the raw notes from a [length] internal meeting on [topic]. Please produce: (1) a short summary, (2) a table of decisions made, (3) a table of action items with owner and proposed due date, and (4) any open questions that still need a decision. Flag anything ambiguous rather than guessing.鈥 | Rapidly transforms raw notes or transcripts into decisions, action items, owners, and timelines, reducing administrative burden and improving follow-through and accountability. | Meeting notes would likely contain internal discussions and sometimes confidential information. Outputs affect operational decisions. Enterprise security and auditability are required. | or |
Library Content Summary for Study and Research Sample prompt: 鈥淲ithin [vendor platform鈥檚 built-in AI feature], please summarize the key findings, methods, and limitations of the articles I have selected on [research topic], and identify common themes across them.鈥 | Quickly synthesizes content into accessible summaries and key themes, and assists with efficient searching for students and researchers. | Library-licensed content within databases and e-resources is governed by license agreements with content providers. | Library-licensed content within databases and e-resources can be used only with AI tools that the vendors have made available within their own platforms. Currently, our licenses do not permit extraction of content for use with external AI tools or products (this includes LibreChat and Copilot, which are still considered external to these vendors). Note: this is an active area of negotiation with content vendors and AI providers. |
Email Triage and Drafting for High-Volume Inboxes Sample prompt: 鈥淚n Copilot for Outlook: Please review my unread inbox and group messages into [urgent / awaiting reply / FYI / can be archived]. For each urgent item, draft a brief reply in a [warm, professional] tone that acknowledges the request and proposes a next step or timeline.鈥 | Could reduce inbox overload by prioritizing messages, drafting consistent responses, and surfacing exceptions, allowing staff to focus attention on higher-value work and complex cases. | Emails often contain personal information, internal operations, and sensitive data. Work occurs directly in Microsoft workflows. Tool should remain inside the institutional tenant. | Our Chat or Microsoft Flow to connect Copilot to the application in question. Only the institutional license and only with single sign-on. |
Policy or Procedure Simplifier Sample prompt: 鈥淭he text below is an internal [policy / procedure / guideline] on [topic]. Please produce: (1) a 150-word plain-language summary, (2) 8 to 10 FAQs a typical staff member or student might ask, and (3) a one-page checklist of what someone needs to do to comply. Flag anything ambiguous that should be clarified by the policy owner.鈥 | Improves accessibility and uptake of policies by translating dense documents into clear summaries, FAQs, and practical checklists that support consistent interpretation and compliance. | Draft policies and interpretations are internal documents and may change before publication. Data must remain inside institutional systems. | |
Training Module Outline and Quiz Builder Sample prompt: 鈥淚 am building a [length] training module on [topic] for [audience, e.g., new staff in X unit]. The learning objectives are: [list]. Please produce: (1) a module outline with sections and timing, (2) two short scenarios learners could discuss, and (3) a 10-question quiz with answer key (mix of multiple-choice and short-answer).鈥 | Speeds the development of high-quality learning materials by generating structured outlines, scenarios, and draft assessment items that can be refined by subject matter experts. | Training content often includes internal processes, examples, or institutional context. Draft material should not leave university systems. | |
Data Cleanup and Standardization Guidance Sample prompt: 鈥淚 have a spreadsheet with the following columns: [list columns and example values]. The data has these issues: [e.g., inconsistent date formats, mixed-case names, duplicate rows]. Please recommend a step-by-step cleanup approach, including specific Excel formulas or Power Query steps, and a set of validation rules to prevent the issues from recurring.鈥 | Improves data quality and reduces manual effort by generating formulas, validation rules, and cleanup approaches that make datasets more reliable for analysis and reporting. | Datasets typically contain internal operational or personal data. If working directly inside Excel, native integration improves workflow efficiency. | or |
Survey Theming and Insight Extraction Sample prompt: 鈥淏elow are de-identified open-ended responses to the question: 鈥淸survey question].鈥 Please identify the top 5 to 7 themes, estimate the share of responses for each, provide 2 to 3 representative (de-identified) quotes per theme, and flag any responses that suggest urgent concerns needing follow-up.鈥 | Accelerates insight generation by organizing qualitative feedback into meaningful themes and summaries, supporting faster evidence-informed decision making. | Survey data may include personal opinions, sensitive feedback, or identifiable information. Data protection and confidentiality are required. | |
Drafting Department Communications in Queen's Voice Sample prompt: 鈥淧lease draft a [memo / announcement / newsletter blurb] from [unit] to [audience] about [topic]. Key messages: [list]. The tone should be [warm, plain-language, accessible]. Keep it under [word count]. Include a clear call to action and a contact for questions.鈥 | Increases consistency and efficiency in communications by producing well-structured drafts aligned with tone, accessibility, and institutional standards. | Drafts may reference internal initiatives or unpublished information. Output often lives in Word or SharePoint for review cycles. | or |
Knowledge Base Article Generator from Repeated Questions Sample prompt: 鈥淏elow are [N] questions our team is asked repeatedly, along with the typical answers we provide. Please convert each into a knowledge base article with: a clear title, a 1-sentence summary, step-by-step instructions, common pitfalls, and links/contacts for further help. Use plain language.鈥 | Reduces repeat inquiries and onboarding burden by converting common questions into clear, reusable how-to articles and troubleshooting guides. | Source material reflects internal processes and systems. Content often lives in SharePoint or service platforms tied to institutional access controls. | or |
Operational Process Mapping and Bottleneck Finder Sample prompt: 鈥淏elow is a description of our current [process name], step by step, including who performs each step, the system used, and approximate time taken. Please: (1) produce a clean process map, (2) identify likely bottlenecks, rework loops, and single points of failure, and (3) recommend 3 to 5 improvements ranked by likely impact and effort.鈥 | Makes complex workflows visible and measurable by identifying inefficiencies, rework, and improvement opportunities that support service quality and operational efficiency. | Process details reveal internal operations, staffing patterns, and system dependencies. Requires secure handling and internal-only analysis. | or |
First Draft of a Business Case for an AI Pilot Sample prompt: 鈥淧lease draft a first version of a business case for an AI pilot that would [proposed solution] to address [problem] for [audience]. Use the following sections: problem statement, proposed solution, expected benefits, risks and mitigations, governance and data considerations, success metrics, timeline, and rough cost estimate. Flag assumptions you are making so I can validate them.鈥 | Accelerates early-stage planning by generating structured drafts that clarify the problem, value proposition, risks, governance, and evaluation approach, enabling faster decision cycles. | Business cases include internal strategy, budgets, risk assessments, and governance assumptions. Must remain within institutional tools. | or |
Summarizing Survey Data or Other Analytics Sample prompt: 鈥淎ttached/below is de-identified [survey / analytics] data on [topic]. Please produce: (1) a short narrative summary of what the data shows, (2) the top 5 quantitative findings with the relevant numbers, (3) the top 5 qualitative themes with representative (de-identified) quotes, and (4) any outliers or emerging signals worth flagging for follow-up.鈥 | First-pass organizing of both quantitative and qualitative survey data into coherent summaries, key themes, and patterns. Helps identify trends, outliers, and emerging signals that may not be immediately visible, enabling faster, evidence-informed decision making while reducing manual analysis burden. | Survey data may include personal opinions, sensitive feedback, and potentially identifiable information. Even when de-identified, there is a risk of re-identification or exposure of sensitive institutional insights. Analysis should occur within secure, institutionally approved environments that ensure data protection, confidentiality, and appropriate governance. | or , depending on where the data resides (e.g., Excel, Forms, or SharePoint). Only institutional licenses with secure authentication should be used. |