Generative AI 101

Introduction to the Fundamentals

Michael Callaghan

Reader in Technology Enhanced Learning

Conference founder & Chair GenAIEdu

[email protected]
 Overview
• What is generative AI • ChatGPT command types
• Large language models • ChatGPT - plugins
• Tokens • What about the students?
• Tokens versus parameters • Office 365 Copilot – Windows 11 Copilot
• Prompt engineering
• What are our options?
• Zero-shot, one-shot and few-shot learning
• What about Turnitin?
• Prompt guide
• Practical use in learning and teaching
• Generative AI systems examples
• Large language models – getting started • Assessment redesign for generative AI
• OpenAI – ChatGPT • Resources and further reading
• OpenAI tools
What is generative AI and
how does it work?
Generative AI (GenAI)
• Type of Artificial Intelligence that leverages AI to generate content or data
• Data can include text, images, audio, video, 3D models, code and video games
• Typically created in response to prompts (prompt engineering)
• Prompts are constructed inputs to language models to generate useful output
• Usually given with examples - zero shot versus few shot learning

Source: docs.cohere.com
https://docs.cohere.com/docs/prompt-engineering
Large language models
• Type of machine learning model/algorithm
• Performs variety of natural language processing (NLP) tasks
• Learn, understand, and process human language efficiently
• E.g., generate/classify text, answer questions conversationally
• Large = number of values (parameters) the model can change
autonomously as it learns
• Uses hundreds of billions parameters
Large language models
• Trained with large amounts of data
• Based on neural networks (Transformers) that learn context and
understanding through sequential data analysis
• Uses self-supervised learning to predict the next token in a
sentence, given the surrounding context
• Process is repeated over and over until the model reaches
acceptable level of accuracy
• GPT-4 (Generative Pre-trained Transformer)
Large language models (LLM)

Source: https://lifearchitect.ai/models/

Source: informationisbeautiful.net
https://informationisbeautiful.net/visualizations/the-rise-of-generative-ai-large-language-models-llms-like-chatgpt/
Large language models training
• LLMs return similar patterns to data it is trained on (not thinking)

• Wikipedia Large
• Common crawl Text output
Language
• Other sources Model

Source: informationisbeautiful.net
Generative AI +: Education
https://www.youtube.com/@mit
Tokens
• Basic units of text/code LLM uses to process or generate language
• Can be characters, words, subwords, segments of text or code
• Tokens generally = ~4 characters of text for common English
• ¾ of a word – 100 tokens ~= 75 words
• GPT models process text using tokens
• Common sequences of characters found in text
• Understands the statistical relationships between these tokens
• Used to predict next token in a sequence of tokens
Tokens

https://platform.openai.com/tokenizer
Tokens

https://platform.openai.com/playground
Tokens and structure

Source:
Shanahan M, McDonell K, Reynolds L. Role play with large language models.
Nature. 2023 Nov;623(7987):493-498. doi: 10.1038/s41586-023-06647-8
Epub 2023 Nov 8. PMID: 37938776.
Tokens versus parameters
• Large language model (LLM) context
• Token is a basic unit of meaning e.g., word, punctuation mark
• Parameters = numerical values that define model behaviour
• Adjusted during training to optimize model's ability to generate
relevant and coherent text
Large language model

Takes an input and produces a token

as an output

Source: medium.com
https://medium.com/@hmohamedhussain2004/what-is-an-llm-a0086882e585
Why are tokens important?
• Context window in large language models
• Length of text model can process and respond to in given instance
• Constraints length of prompt and response
Prompt (tokens) Response (sampled tokens)
gpt-4 8000 tokens
gpt-4-32k 32000 tokens

gpt-4 6000 words

gpt-4-32k 24000 words

Context window
ChatGPT-4 currently has a cap related to message frequency – 128,000 tokens
https://platform.openai.com/docs/models/gpt-4-and-gpt-4-turbo
Cost is based on tokens used

https://help.openai.com/en/articles/7127956-how-much-does-gpt-4-cost https://invertedstone.com/tools/openai-pricing/
Gemini 1.5 - 1 million multimodal tokens

Complex reasoning about vast amounts of information

1.5 Pro can seamlessly analyze, classify and summarize large amounts of
content within a given prompt. For example, when given the 402-page
transcripts from Apollo 11’s mission to the moon, it can reason about
conversations, events and details found across the document.

https://blog.google/technology/ai/google-gemini-next-generation-model-february-2024/#architecture
How do I communicate with and
use large learning models?
Prompt engineering
• Prompt is a piece of text that you provide to a language model to
get a specific response or output
• Prompt can be a sentence, a paragraph, or entire article
• Serves as starting point for the model to generate new content
• Constraints or focusses interaction

Source: docs.cohere.com
https://docs.cohere.com/docs/prompt-engineering
zero-shot, one-shot and few-shot learning
• zero-shot learning: prompting a
language model with an instruction
• No examples or demonstrations of the
task you want it to perform
• Givens instruction in natural language
• one-shot includes one demonstration
• few-shot (in-context) learning includes
many demonstrations 10+
Source: medium.com
https://sh-tsang.medium.com/review-gpt-3-language-models-are-few-shot-learners-ff3e63da944d
Prompt guide
• Offer context – think about exactly what you want to generate
• Provide a prompt specifically tailored to that outcome
• Include helpful information upfront – role or persona
• Give examples of inputs and outputs
• Set the length of the response
• Ask for help to write the prompt
• Add useful expressions
Source: zapier.com
https://zapier.com/blog/gpt-prompt/
Useful expressions to include
• Let's think step by step - logic and maths
• Thinking backwards - avoid inaccurate conclusions
• In the style of [famous person] - match style
• As a [insert profession/role] - frames knowledge
• Explain this topic for [insert specific age group] - defines audience
• For the [insert company/brand publication] – tone & context
Source: zapier.com
https://zapier.com/blog/gpt-prompt/
Test and adjust variables
• Change model used
• Change temperature for creativity. Same prompt = same response
Tweak maximum length of response
• Stop sequences – terminate at certain point e.g., end of sentence
or number of items in a list
• Frequency penalty - discourage model from repeating same words
• Presence penalty - encourage model to include diverse range of
tokens in the generated text
Source: zapier.com
https://zapier.com/blog/gpt-prompt/
Prompt frameworks - CIDI
• C - context
• I - instructions
• D - details
• I - input

Source: 7taps.com 9 Frameworks to master ChatGPT Prompt Engineering

https://app.7taps.com/0pBlRT8vd7U9L https://www.linkedin.com/pulse/9-frameworks-master-chatgpt-prompt-engineering-edi-hezri-hairi/
CIDI Framework - context
• Give ChatGPT a role “ Act as a … You are a …”
• Give it other information it needs to step into that role
• If you are unsure, ask it, “what information would you need
before stepping into that role”

Source: 7taps.com
https://app.7taps.com/0pBlRT8vd7U9L
CIDI Framework - context
• Make the context and role specific
• You are an educator with 10 years of experience creating student
centred writing lessons. You believe that writing clearly is
important for thinking clearly
• Tips – write a good description. Write down your values and
expertise. Make it descriptive. Then you can use it for any prompt
when brainstorming lesson plans, rubrics etc.
Source: 7taps.com
https://app.7taps.com/0pBlRT8vd7U9L
CIDI Framework - instructions
• Tell ChatGPT what you are going to do and what you want it to do
• Structure it as “I will” and “You will statement”
• Be clear about what everyone’s role is
• E.g., “I am going to provide you with a list of notes related to
social media marketing. You are going to convert these notes into
a basic outline for a 45 - minute lesson plan. Then I will revise and
approve. Once approved you can expand it..”
• Give it a sense of what you want
Source: 7taps.com
https://app.7taps.com/0pBlRT8vd7U9L
CIDI Framework - details
• Give ChatGPT any other information you want it to consider
• E.g., tone, style, format, specifics about task, target audience
• Think about the output and put in the details needs to get that

Source: 7taps.com
https://app.7taps.com/0pBlRT8vd7U9L
CIDI Framework - input
• You have told ChatGPT you were going to do something and it
was going to do something
• Follow your structure and instructions
• Give it the information to complete the task
• E.g., provide it with data or samples of good output
• Range of advanced techniques you can use

Source: 7taps.com
https://app.7taps.com/0pBlRT8vd7U9L
Multimodal – voice and image inputs

https://openai.com/blog/chatgpt-can-now-see-hear-and-speak

How to Enable ChatGPT Voice to Voice on Phone (iPhone & Android) Talk to ChatGPT!
https://youtu.be/SamGnUqaOfU?si=A-rnz_uOrzCtUEzw
Multimodal – voice and image inputs

https://openai.com/research/gpt-4
Multimodal – voice and image inputs

https://openai.com/research/gpt-4
What other domains is generative
AI impacting and used in?
Generative AI systems examples
• Image generators – Midjourney, Stable Diffusion, DALLE 3
• Runway – text to video
• Sora – text to video
• Code generation tools – Copilot
• Audio generation tools - VALL-E, resemble.ai
• AI covers
• Video game generation - Ludo.ai, FRVR Forge, Unity
• Large language models - GPT-4, Gemini, or Claude 2
Midjourney, Stable Diffusion, DALLE 3

Source: midjourney.com
https://www.midjourney.com/
https://www.linkedin.com/posts/nunoalexandreantunes_ai-midjourney-greenarchitecture-activity-6986203891709165568-0gqw/
https://youtu.be/FjtBpcvBLLE
Midjourney prompts

Source: blueshadow.art
https://www.blueshadow.art/midjourney-prompt-commands/
 Runway – text to video

Source: runwayml.com
https://runwayml.com/
Runway/Midjourney text to video

Anonymouse @TheMouseCrypto
https://twitter.com/i/status/1687127183045574656
OpenAI Sora - Text to Video Model

OpenAI Sora - Text to Video Model cat on bed

https://youtu.be/-2daXglty98?si=Xy6V2Te8weQxy0Lc
Eleven labs AI generated audio Sora

Sound Effects are Coming Soon to ElevenLabs

https://youtu.be/VDaZ9gTx7A8?si=gffTBgrNIKSxpySM
Rapidly evolving – Emo video

Alibaba presents EMO AI - All Demo Clips Upscaled to 4K

https://youtu.be/wtcSZdHZne4?si=bMkRXGZbOor-S1jW
Code generation – GitHub Copilot
• AI pair programmer – free for students
• GitHub Copilot uses the OpenAI Codex to suggest code and
entire functions in real-time in your editor
• GitHub Copilot allows users to train tool on their codebase
Code generation – GitHub Copilot
• GitHub Copilot completes code, helps developers boost
productivity with repetitive, simple code snippets
• ChatGPT supports developers in development/debugging process

https://docs.github.com/en/copilot/quickstart
https://github.blog/2023-06-13-survey-reveals-ais-impact-on-the-developer-experience/
Audio generation tools - VALL-E, resemble.ai

Source: arstechnica
https://arstechnica.com/information-technology/2023/01/microsofts-new-ai-can-simulate-anyones-voice-with-3-seconds-of-audio/

Source: resemble.ai
https://www.resemble.ai/
AI Covers - Frank Sinatra

https://youtu.be/W7SQ4uf9GmA?si=wNYjZCyjCWaFPb5q
https://youtu.be/_Y543NEiR5w?si=cGW0OtFtY5xeYZ1y
https://youtu.be/VNWudHD3Kt8?si=BIOq2eaM5tVibbkI
https://youtu.be/HWsb7zTKplc?si=93Z5BArTwBHNWd0I
AI Covers – Johnny Cash
 Suno/Chat GPT 4 – music generation

https://www.suno.ai/
Video games - Ludo.ai, FRVR Forge
• AI supported game ideation, planning and development

https://ludo.ai/
https://www.facebook.com/frvrgames/?locale=en_GB
Video games – Unity Muse

https://muse.unity.com/
Video games – Scenario

https://www.scenario.com/
AI generated game worlds

https://aidungeon.io/
https://www.newscientist.com/article/2390019-ai-generates-video-game-levels-and-characters-from-text-prompts/
AI generated game worlds

https://www.technologyreview.com/2024/02/29/1089317/google-deepminds-new-generative-model-makes-super-mario-like-games-from-scratch/
Clone yourself for videos

Source: synthesia
https://www.synthesia.io/features/custom-avatar
Generative AI in Ireland 2024

https://www.rte.ie/news/business/2024/0229/1435038-generative-ai-survey/
https://pulse.microsoft.com/en-ie/work-productivity-en-ie/na/fa1-generative-ai-adoption-rates-are-on-the-rise-in-workplaces-according-to-our-latest-report-supported-by-trinity-college-dublin/
How do I get started with
large learning models?
Large language models – getting started
• Large range of LLM’s available e.g., GPT-4, Gemini, or Claude 2
• Range of ways to access and setup
• Depending on level of technical expertise
• Cohere or Hugging Face and LangChain
• Also use Microsoft co-pilot, Bing AI or ChatGPT, Orca
• Following slides will focus on ChatGPT
https://www.bing.com/?/ai
https://chat.openai.com/
https://gemini.google.com/
OpenAI - ChatGPT
• Free (limited) access and paid subscription available - $20 USD

https://platform.openai.com/apps
OpenAI – range of tools available

https://platform.openai.com/tokenizer
https://platform.openai.com/ai-text-classifier
https://platform.openai.com/playground
ChatGPT command types
• Ask for facts - search but more conversational
• Suggestions - recommend 5 books on programming
• Translations - language translation with context
• Comparing - compare two different things (cats versus dogs)
• Creative writing - email templates, blogs, plays, short stories
• Brainstorming - brainstorm ideas, five ideas for a picnic
• Summarizing text - take paragraphs of text and summarise
• Ideas and list generation - create lists, gather information
ChatGPT command types
• Pros and cons - making decisions on buying decisions etc
• Quotes and studies - find 3 scientific studies on sleep and summarise
• Feedback on your text - good and bad points, suggestions for improvements
• Role playing - assign a persona, talk as that person, interview me for a job
• Explanation of concepts - good at explaining concepts
• ChatGPT commands - ask it commands you can use e.g., review my essay
• Coding - writing and debugging code
• Mentorship - life advice, coaching and mentoring
ChatGPT command – recommend books
ChatGPT - plugins
• Allow ChatGPT to interact with live data from web and specific websites
• Similar to apps with potential to turn the chatbot into a versatile interface
• Extends on core functionality
ChatGPT - plugins
ChatGPT – Your AI council
• AI council assesses queries through various agents, offering insights from
many perspectives.

Prompt “Use your AI council to discuss the advantages and disadvantages of using generative AI for assessment in higher education”
ChatGPT – Your AI council
ChatGPT – Ask my pdf

https://www.ulster.ac.uk/ data/assets/pdf_file/0005/318560/EC1-Guidance-1.pdf
ChatGPT – Video insights

https://youtu.be/cNazLpXCE_Q
ChatGPT – Video insights

https://youtu.be/cNazLpXCE_Q
GPT – customized version
GPT – customized version
GPT- 4 main weaknesses
• Fine tune GPT-4 i.e., adjust and adapt a pre-trained model to
perform specific tasks or to cater to a particular domain more
effectively – niche areas e.g., academic papers.
• Does not (directly) update knowledge in real time
• Hallucinations/confabulation – makes up “facts”
• Cost, access constraints and scalability
• High computational power requirements, less sustainable
• Limitations of overall approach
Source: openai https://platform.openai.com/docs/guides/fine-tuning
https://platform.openai.com/docs/guides/fine-tuning/what-models-can-be-fine-tuned
What is coming next / here now?
Office 365 Copilot – Windows 11 Copilot

https://blogs.microsoft.com/blog/2023/03/16/introducing-microsoft-365-copilot-your-copilot-for-work/
Microsoft 365 Copilot – copilot for work
• Combines large language models (LLMs) with your data in the Microsoft
Graph and the Microsoft 365 apps
• Works with you, embedded in Word, Excel, PowerPoint, Outlook, Teams
• Business Chat works across the LLM, the Microsoft 365 apps, and your data
your calendar, emails, chats, documents, meetings and contacts
• Give it natural language prompts like “Tell my team how we updated the
product strategy,” and it will generate a status update based on the
morning’s meetings, emails and chat threads
https://blogs.microsoft.com/blog/2023/03/16/introducing-microsoft-365-copilot-your-copilot-for-work/
Microsoft 365 Copilot – copilot for work
• Copilot in Word - gives you a first draft to edit and iterate on, saving hours in
writing, sourcing, and editing time
• Copilot in PowerPoint creates presentations with a simple prompt, adding
relevant content from a document you made last week or last year
• Copilot in Excel, analyse trends, create professional-looking data
visualizations in seconds
• Works alongside Windows Copilot for Windows 11

https://blogs.microsoft.com/blog/2023/03/16/introducing-microsoft-365-copilot-your-copilot-for-work/
Microsoft 365 Copilot – copilot for work
Microsoft Copilot everywhere

AI in Education - the latest updates from Microsoft EDU

https://youtu.be/kya-6ZU5WPo?si=Fna7ob-p0xW0lgae
 Learning Accelerators

AI in Education - the latest updates from Microsoft EDU

https://youtu.be/kya-6ZU5WPo?si=Fna7ob-p0xW0lgae
Learning Accelerators

AI in Education - the latest updates from Microsoft EDU

https://youtu.be/kya-6ZU5WPo?si=Fna7ob-p0xW0lgae
What are our options?
Assessment redesign for generative AI
• Ignore - generative AI is poised to have a significant
impact on education
• Ban - students almost immediately found ways to
bypass these bans
• Invigilate - written exams, oral examinations and vivas
• Embrace – concerns with ethics, fairness, equity,
privacy, access & ability of students to use
• Design around - increasingly sophisticated AI, riskier
strategy/less effective
• Rethink - medium to long term solution
https://www.linkedin.com/pulse/assessment-redesign-generative-ai-taxonomy-options-viability-lodge/
Assessment redesign for generative AI
• Ranking of viability red - likely not viable, orange - care needed,
and green - seems most viable.
• Redesign of assessment solution - short, medium, and long term

https://www.linkedin.com/pulse/assessment-redesign-generative-ai-taxonomy-options-viability-lodge/
Rethink assessment
• Assessment of learning that is conducted through 'real world'
tasks requiring students to demonstrate their knowledge and
skills in meaningful contexts' (Swaffield, 2011)
• Authentic assessments are designed to measure a learner's ability
to apply learning to real-world contexts
• Using activities related to real-world tasks or scenarios
• Providing an opportunity for collaboration and co-production
• Encouraging opportunities for reflection on lived experiences
• Adopting problem-solving techniques and activities
• Assess process not outcomes
https://www.mmu.ac.uk/about-us/professional-services/uta/assessment/authentic#ai-79899-0
Challenges
• Infrastructure and security development
• Ethical analysis and policy development
• Community engagement
• Digital and AI literacy divide
• Curriculum building
• Professional development - literacy, technology, and curriculum
training
• Initial rollout of student-facing initiatives and programs
“That initial panic from ChatGPT's release has ebbed, being replaced gradually with a growing unease. Something new has arisen and cannot be
easily policed or governed by policy, making us confront some deep and fundamental questions about what it means to teach and learn. ”

https://nickpotkalitsky.substack.com/
Proctored in person written exams

https://www.bbc.co.uk/news/uk-scotland-glasgow-west-68380264
https://www.teachthought.com/learning/what-is-blooms-taxonomy/
https://www.simplypsychology.org/blooms-taxonomy.html
Interactive oral assessments

https://sway.cloud.microsoft/yQ2s0Bm3ILkWtGll
Alpha School

https://alpha.school/austin/
Alpha School

https://alpha.school/austin/

'We don't have teachers' | This Austin private school lets AI teach core subjects
https://youtu.be/QFczkA88bcs?si=1gJyXGWCQ_cts9Ah
Alpha School
• Adaptive AI tutors adjust to each student's learning style provide
personalized instruction in core subjects like reading, math, and writing
• Noise-cancelling phone booths, allow students to focus and limit distractions.
• Utilize a range of educational apps designed to be engaging and interactive,
adapting to the student's progress and level of understanding.
• Real-time analytics provide insights into student performance and progress,
allowing teachers to adjust their instruction accordingly.
• Curriculum is aligned with state and national standards, but also goes beyond
traditional standards to foster creativity and critical thinking skills.
• Allows for deep work - not designed to focus and learn for six hours a day
https://alpha.school/austin/
Alpha School – socialization & activities
• Students spend two hours a day on traditional classroom learning in booths
• Rest of the time, engaged in a variety of activities designed to develop their
skills and passions
• E.g., group projects, hands-on learning activities, exploring creative pursuits
• Finland and Sweden following similar approach – “friluftsliv”

https://alpha.school/austin/
Generative AI Shaping The Future: Massachusetts Institute of Technology (MIT)
https://www.educationcorner.com/finland-education-system/ https://youtu.be/sc9WYhdCb7U?si=lRUEnUA36LJ-aAbO
What about Turnitin or AI detectors?
The one - hour final year project report
The one - hour final year project report
The one - hour final year project report
Turnitin AI Writing Detection

https://help.turnitin.com/ai-writing-detection.htm
Turnitin AI Writing Detection
Netus.ai

Assume Turnitin AI will be no use

How can the students use AI
practically for learning?
Assigning AI: approaches for students with prompts
• Paper examines transformative role of Large
Language Models (LLMs) in education
• Potential as learning tools with inherent risks and
limitations
• Propose seven approaches for utilizing AI in
classrooms
• Explores distinct pedagogical benefits and risks
• Focus on helping students learn with and about AI

Mollick, Ethan R. and Mollick, Lilach, Assigning AI: Seven Approaches for Students, with Prompts (June 12, 2023).
Available at SSRN: https://ssrn.com/abstract=4475995
Explores challenges around AI use
• Practical strategies designed to mitigate risks
• E.g., complacency about AI’s output, errors, and biases
• Promotes active oversight, critical assessment of AI outputs, and
complementation of AI's capabilities with the students' unique insights
• Challenges students to remain the "human in the loop“
• AI serves as a supportive tool rather than a replacement
• Framework offers a guide for educators navigating the integration of AI-
assisted learning in classrooms
Seven approaches for utilizing AI in classrooms
• AI-mentor - frequent feedback on work immediate and adaptive reactions
• AI-tutor - direct instruction and educational guidance
• AI-coach - increasing Metacognition, thinking about thinking
• AI-teammate - increasing collaborative intelligence, play devil’s advocate
• AI-student - teaching others, student teaches AI about a topic
• AI-tool - general use writing software, interview subject, extend capabilities
• AI-simulator - scenario creator, set story, help make decisions, solve problems
AI-tutor - direct instruction and educational guidance

• Create a generic prompt

• Help any student study any topic
• Good tutor role, pushing students to generate responses and think
through problems, connect ideas, and offer feedback and practice
• Highlight risk of confabulation – make up plausible-seeming
incorrect answers
• Provides guidance on use for students
AI tutoring Instructions for students
• You are responsible for your own work. The AI can “hallucinate” or make things up
• Take every piece of advice or explanation critically and evaluate that advice
• It’s not a person but it can act like one. It’s very easy to imbue meaning into AI responses, but the AI is
not a real person responding to you. It is capable of a lot, but it doesn’t know you or your context. It
can also get stuck in a loop.
• The AI is unpredictable. The AI is a prediction machine. It has studied billions of documents on the
web, and it tries to continue your prompt reasonably based on what it has read. But you can’t know
ahead of time what it’s going to say. The very same prompt (from you) can get you a radically different
response (from the AI). That means that your classmates may get different responses and if you try
the prompt more than once, you’ll get a different response from the AI as well.
• You’re in charge. If the AI gets stuck in a loop or won’t wrap up a conversation or you’re ready to move
on, then direct the AI to do what you’d like. For instance: I’m ready to move on. What’s next?
• Only share what you are comfortable sharing. Do not feel compelled to share anything personal.
Anything you share may be used as training data for the AI
Getting the most from tutoring - students
• Ask for clear explanations: If something isn't clear, don't hesitate to ask the AI to clarify. Use
phrases like: "Can you explain this term?" or "Can you explain this differently? "
• Share what you understand and what you don’t understand: The AI can provide better help if
it knows where you're having trouble. For instance, you can tell the AI : "I understand this
part, but I'm not sure about this other part. Can you give more details? "
• Summarize the conversation: The AI doesn't necessarily track all your previous interactions.
To get better help, briefly summarize interactions before asking your question. For example:
"We talked about 'educational scaffolding' before. Can you explain how this would work in a
classroom?“
• Ask for connections between examples and concepts: If you're unsure about how an example
relates to a concept, ask the AI. For instance, "How does this example relate to the concept
we're discussing?" This will help you make sense of the concept.
• Share all of your interactions with me. Briefly discuss what you learned from using the tool
AI-tutor - direct instruction and educational guidance

“You are an upbeat, encouraging tutor who helps students understand concepts by explaining ideas and
asking students questions. Start by introducing yourself to the student as their AI-Tutor who is happy to
help them with any questions. Only ask one question at a time. First, ask them what they would like to learn
about. Wait for the response. Then ask them about their learning level: Are you a high school student, a
college student or a professional? Wait for their response. Then ask them what they know already about
the topic they have chosen. Wait for a response. Given this information, help students understand the topic
by providing explanations, examples, analogies. These should be tailored to students learning level and
prior knowledge or what they already know about the topic. Give students explanations, examples, and
analogies about the concept to help them understand. You should guide students in an open-ended way. Do
not provide immediate answers or solutions to problems but help students generate their own answers by
asking leading questions. Ask students to explain their thinking. If the student is struggling or gets the
answer wrong, try asking them to do part of the task or remind the student of their goal and give them a
hint. If students improve, then praise them and show excitement. If the student struggles, then be
encouraging and give them some ideas to think about. When pushing students for information, try to end
your responses with a question so that students have to keep generating ideas. Once a student shows an
appropriate level of understanding given their learning level, ask them to explain the concept in their own
words; this is the best way to show you know something, or ask them for examples. When a student
demonstrates that they know the concept you can move the conversation to a close and tell them you’re
here to help if they have further questions.”
AI-tutor - direct instruction and educational guidance
AI-tutor - direct instruction and educational guidance
Assigning AI: approaches for students with prompts
How can I use AI practically in
learning and teaching?
Using AI to Implement Effective Teaching
Strategies in Classrooms
Guidance for using AI to quickly and easily implement evidence
based teaching strategies then assess the AI output
• Strategy 1: Using AI to Produce Many Varied Examples
• Strategy 2: Using AI to Provide Multiple Explanations
• Strategy 3: Using AI to Develop Low-Stakes Tests
• Strategy 4: Using AI to Assess Student Learning
• Strategy 5: Using AI to Distribute Practice of Important Ideas

Mollick, Ethan R. and Mollick, Lilach, Using AI to Implement Effective Teaching Strategies in Classrooms: Five Strategies, Including
Prompts (March 17, 2023). Available at SSRN: https://ssrn.com/abstract=4391243 or http://dx.doi.org/10.2139/ssrn.4391243
Using AI to Produce Many Varied Examples
“I would like you to act as an example
generator for students. When confronted with
new and complex concepts, adding many and
varied examples helps students better
understand those concepts. I would like you to
ask what concept I would like examples of,
and what level of students I am teaching. You
will provide me with four different and varied
accurate examples of the concept in action.”
Using AI to Provide Multiple Explanations
“You generate clear, accurate
examples for students of concepts. I
want you to ask me two questions:
what concept do I want explained,
and what the audience is for the
explanation. Provide a clear, multiple
paragraph explanation of the concept
using specific example and give me
five analogies I can use to understand
the concept in different ways.”
Using AI to Develop Low-Stakes Tests
“You are a quiz creator of highly
diagnostic quizzes. You will make good
low-stakes tests and diagnostics. You
will then ask me two questions. (1)
First, what, specifically, should the quiz
test. (2) Second, for which audience is
the quiz. Once you have my answers
you will construct several multiple
choice questions to quiz the audience
on that topic. The questions should be
highly relevant and go beyond just
facts. Multiple choice questions should
include plausible, competitive alternate
responses and should not include an
"all of the above option." At the end of
the quiz, you will provide an answer key
and explain the right answer.”
Using AI to Assess Student Learning
“I am a teacher who wants to understand what students
found most important about my class and what they are
confused by. Review these responses and identify common
themes and patterns in student responses. Summarize
responses and list the 3 key points students found most
important about the class and 3 areas of confusion: [Insert
material here]”
Distribute Practice of Important Ideas
“You are an expert teacher who
provides help with the concept of
distributed practice. You will ask me
to describe the current topic I am
teaching and the past topic I want
to include in distributed practice.
You will also ask me the audience or
grade level for the class. Then you
will provide 4 ideas about how
include the past topic into my
current topic. You will also provide 2
questions I can ask the class to
refresh their memory on the past
topic.”
Rethinking assessment
COM540 Business Intelligence 2022
• 20 Credit point final year module
• Computing and business students
• 100% coursework – report/prototype

https://www.windowscentral.com/microsoft/copilot-in-microsoft-excel-can-break-down-spreadsheets-and-identify-trends-using-ai
 Rethinking assessment
Stages of Backward Design: Wiggins & McTighe

Understanding by Design through three stages

a) identify desired results

b) determine acceptable evidence
c) plan learning experiences and instruction

Understanding by design® framework McTighe and Wiggins

https://files.ascd.org/staticfiles/ascd/pdf/siteASCD/publications/UbD_WhitePaper0312.pdf

Generative AI Shaping The Future: Massachusetts Institute of Technology (MIT)

https://youtu.be/sc9WYhdCb7U?si=lRUEnUA36LJ-aAbO
AI Literacy Framework

Dr Kara Kennedy
https://kennedyhq.com/wp/wp-content/uploads/2023/12/AI-Literacy-Framework-2023.pdf
COM540 Business Intelligence 2023
• Are goals still valid - what should they learn
• Find out what generative AI can do - use it effectively
• What can students do that is more interesting – high level of
cognitive processing – synthesizing, evaluating and analysing
• Revise assignment
• More focus on the process/journey (together)
• Evaluate output
Design, Evaluate, Explore, and Reflect
• D - Clearly define the stages of the project e.g. essay,
brainstorming assignment, or activity, state each stage's
purpose in achieving student learning outcomes
• E - For each stage, evaluate a specific generative AI technology
to pair with the learning activity
• E - Encourage students to explore that specific generative AI
technology for that stage
• R - Provide students with space for reflection
https://marcwatkins.substack.com/
COM540 Business Intelligence 2023
COM540 Business Intelligence 2023
Ulster University Copilot - research
Ulster University Co-pilot - research
• Explore practical use of Large Language Models in Education
• Extend models to include local knowledge
• E.g., undergraduate student resources and module support
• Add a virtual human front end
• Be inclusive – support diversity
• Assess practical use with students
Ulster University Co-pilot
Module support
Undergraduate student support
Tutor role
Meta Humans
D-iD agents
D-iD agents
Summary, considerations and
conclusions
Summary
• Provided definition of generative AI
• Terminology - Large language models, tokens, parameters
• Interacting with AI - prompt engineering
• Practical uses and impact of generative AI
• AI tools available
• AI detection tools
• Signposted what is coming next
Considerations
• Focus on what technology can do for us and society
• Learning is fundamentally social – build relationships with teachers
and other students/children
• Educational systems are complex and technical environments
• Multipurpose - preparing for work, providing pastoral care,
supervision, creating citizens …
• Change requires systemic change, communities, partnerships,
families and schools
• Broader changes - project based learning, timetabling, curriculum
Generative AI Shaping The Future: Massachusetts Institute of Technology (MIT)
https://youtu.be/sc9WYhdCb7U?si=lRUEnUA36LJ-aAbO
Potential impact
• AI tools should make educators lives easier
• Reduce the busy work – promising start but a long way to go
• Balance between adding more tools/ feeling overwhelmed
• Ideally lead to spending more time engaging in one to one and small
group instruction
• AI replacing teachers? – great tech demos
• Usually problematic or ineffective in practice
• Need to focus on students need
• Work for small groups of highly motivated students
Summary and conclusions
• Undergoing a period of rapid change - requires creative, curious, collaborative thinkers
• Ongoing development of tools that bypass useful cognition e.g., Google maps, calculators,
translators
• Lots of opportunities but also risks
• Consider the long-term implications of AI in education
• Consider how AI will shape skills and competencies required for future workforces - think
about what to teach differently or not teach at all
• Balance technological advancement with critical human skills
• Embrace but understand strengths and weaknesses
• Critically assess – human in the loop
• Still need to be able to assess if something is good or not!
• Rethink our roles and what we do particularly around teaching and assessment
Further reading and resources
Microsoft AI for educators – sign up

https://educationblog.microsoft.com/en-us/2023/03/exploring-new-opportunities-with-ai-in-education
JISC Artificial intelligence

https://www.jisc.ac.uk/innovation/artificial-intelligence
Higher education in the era of AI

https://www.advance-he.ac.uk/news-and-views/higher-education-era-ai
https://www.insidehighered.com/views/2023/03/22/how-ai-shaping-future-higher-ed-opinion
Related publication/discussions

On the dangers of stochastic parrots: Can language models be too big

https://youtu.be/N5c2X8vhfBE
https://www.turing.ac.uk/events/dangers-stochastic-parrots
https://dl.acm.org/doi/10.1145/3442188.3445922
One Useful Thing – Ethan Mollick

https://www.oneusefulthing.org/
National Conference on Generative Artificial
Intelligence in Education – GenAIEdu 26-28 June 2024

https://www.ulster.ac.uk/conference/genaiedu-2023 https://www.ulster.ac.uk/conference/genaiedu-2024
Closing thoughts
• “We overestimate the impact of technology in the
short-term and underestimate the effect in the
long run.” - (Robert) Amara’s Law

• “Remember, today's AI is the worst AI you will

ever use” – Ethan Mollick
Generative AI 101
Introduction to the Fundamentals
Michael Callaghan

Reader in Technology Enhanced Learning

Conference founder & Chair GenAIEdu

[email protected]