Getting started with AI for educators
Source: https://oerc.ox.ac.uk/ai-centre/ai-guides/getting-started-with-ai-for-educators Parent: https://oerc.ox.ac.uk/ai-centre
This page offers guidance and advice for educators getting started with using generative AI in their work:
- What can educators use generative AI for?
- Policy & Guidance
- Tools
- Applications of generative AI in the University
- Tips
- Benefits
What can educators use generative AI for?
Generative AI offers opportunities to enhance teaching and learning in higher education when used with professional judgement.
Educators can use generative AI for these and many other tasks:
- Outlining class content: Suggesting structures or sequencing for lectures and tutorials. Don't consider the output a final version and always adapt for context because AI doesn’t understand you, your students, or your content as well as you do.
- Drafting survey questions: For gathering student feedback or pre-session diagnostics, which you can refine to suit the course aims.
- Communication support: Writing or revising emails or other communication documents - particularly useful for educators who speak English as an additional language or neurodivergent educators.
- Differentiating materials: Adjusting existing content to support learners from different linguistic, cultural, or educational backgrounds.
- Reframing explanations: Brainstorming new ways to explain difficult concepts - especially helpful when students struggle with a particular topic or approach.
- Developing scaffolds: Iterating alternative representations, analogies, or structured prompts for guiding student understanding.
Educators should always review, revise, and contextualise outputs.
You can see how educators and students at Oxford have been using AI with our AI in Education Case Studies.
Policy & Guidance
Oxford’s policy and guidance on AI use in summative assessment sets out expectations for both staff and students.
Tools
University-Supported Chatbots
These tools have enterprise agreements with the University of Oxford. When signed in with your SSO, they can be used for confidential academic data, with data security and privacy protections in place.
- ChatGPT Edu: Developed by OpenAI and includes several features relevant to education. These include: custom GPTs that can be configured and shared with students; voice mode for spoken interaction; canvas for live text collaboration and editing with the LLM; Study Mode for exploring and testing understanding of academic concepts; and a deep research feature that helps synthesise and summarise information across multiple sources.
- Gemini: Developed by Google, Gemini includes several features relevant to education: Gems, which are the equivalent to custom GPTs; a deep research features for finding and synthesising information; and multimodal capabilities for working with text, images, and audio. One standout feature for education is NotebookLM.
- NotebookLM: Acts like a context-aware assistant that can be pre-loaded with readings, videos, and lecture notes. It will link to the source of information in the uploads to make verification simple. Educators can use it to:
- Upload academic materials to explore ideas or pedagogical approaches
- Generate study notes or summaries
- Create mind maps or visual representations
- Turn readings into study notes or quizzes
- Generate interactive audio podcasts on the content
- Interrogate uploaded material and surface connections between sources
- Your Notebooks can be shared with students so they can interact with the materials you've assigned.
- Microsoft Copilot: Microsoft Copilot is available in two versions. The free Copilot chatbot is accessible to all users signed in with their university SSO and can be used for general AI text assistance. The (paid) licensed version, Microsoft Copilot for Microsoft 365, integrates directly into apps like Word, PowerPoint, and Outlook, offering in-context support such as writing assistance, generating slides, transcribing Teams meetings, and summarising meeting notes.
Other AI-powered applications
These tools are useful for a range of educational tasks, but they do not have enterprise agreements with the University of Oxford. They must be used in line with the University's data policy and should not be used for confidential information. Many other AI tools exist, these are just some examples of those the AI and Machine Learning Competency Centre has experience with.
- Claude: Developed by Anthropic, Claude is a general-purpose AI chatbot. It includes features such as "artifacts," which allow users to generate and interact with content (like code, documents, or notes) in a persistent and editable workspace. Claude performs well in long-form content generation, summarisation, and deep reasoning tasks, making it useful for drafting, synthesis, and iterative thinking in educational settings.
- Elicit: An LLM-powered research assistant designed for finding and summarising academic literature. Ask a research question using natural language and it will find relevant studies, generate a report with the information, and extract and display key information about studies.
- Consensus: Similar to Elicit, Consensus is another AI search engine for answers from peer-reviewed literature.
- Gamma: A generative AI tool for creating presentations, documents, and visual explanations. Gamma is useful for turning ideas or AI-generated content into slide decks, visual briefings, or structured documents. It can be prompted directly or used in conjunction with other tools like ChatGPT to present information in more engaging and accessible ways.
- Canva: A design tool with AI-powered features for the creation of visual learning materials, presentations, and infographics. Canva's Magic Design and Magic Write tools can help generate visual templates and textual content, making it easier to produce polished slides, handouts, or explainer graphics.
- HeyGen: A video-generation platform that enables users to create AI-generated videos. It can turn scripts into narrated videos with realistic avatars or translate the language spoken in a video of a user while voice matching and lip-syncing realistically.
- ElevenLabs: A voice synthesis tool that generates realistic speech from text. It supports multiple voices and languages, and is useful for creating audio versions of reading materials or multilingual content to enhance accessibility and engagement.
Applications of generative AI in the University
To explore how generative AI is being used in education at the University of Oxford, you can read the outputs from the AI Teaching and Learning Exploratory Fund, which supported departments and staff at the University of Oxford to explore innovative uses of generative AI in teaching and learning.
You can explore the outputs from those projects here: Teaching and Learning Exploratory Fund Outputs.
Tips
Prompts
To get the most useful output from generative AI, include plenty of detail and context in your prompt. A strong prompt should explain what you're trying to achieve, why you're doing it, and the structure or format you'd like the output to follow. Providing examples or background information helps ensure relevance and clarity.
Example:\ "I’ve written an introduction to semiotics for a first-year media studies course. Some of the students in this group are neurodivergent and/or speak English as an additional language. Rewrite this text to be more accessible without simplifying the academic content—use shorter sentences, define key terms in context, and suggest visual metaphors or examples that could support understanding. Let me know if anything in the original text could be confusing or ambiguous.” (Upload the file you would like rewritten.)
Metaprompting
If you're unsure how to frame your task, you can ask the AI to write the prompt for you. Metaprompting is especially useful for creating clear, well-formatted instructions (ask it to use markdown) for custom GPTs and for setting up complex deep research tasks.
Example: “I’m creating a custom GPT to act as a Socratic discussion partner that helps students deepen their understanding of ethics. Write a prompt that defines its questioning style, tone, and how it should respond to encourage critical thinking and self-reflection. It needs to challenge the user rather than agree and must never give the answer or do the work for them. Write it using markdown so I can copy and paste the instructions into the GPT instructions.”
Dictation
Use the microphone button to speak your ideas rather than typing. Speaking allows faster brainstorming, and AI can work with messy or unstructured language just fine.
Human-in-the-loop
Always verify AI outputs. Core academic skills - source evaluation, bias detection, clarity of reasoning - remain crucial. Generative AI can accelerate production but cannot replace critical judgement.
Benefits
When used thoughtfully with careful and critical oversight, generative AI can support:
- Time savings: Lightening the load of drafting, summarising, or rewording repetitive or formulaic materials to free up time for teaching presence and relational work.
- Personalisation: Adapting content to support varied learning preferences, language backgrounds, and neurodiverse students. Easily offer multimodal formats (text, audio, visual) from the same base material. E.g. Record a lecture (a spoken explanation of a topic) and give the transcript to generative AI to structure a clear and accessible written explanation.
- Student agency and metacognition: Teaching students how to use AI responsibly (e.g., to reformat texts for comprehension) helps them become better learners.
- Fostering critical thinking: When framed as a dialogue tool, AI can act as argument practice or for challenging assumptions, not just content delivery.