Year 8: Healthy Lifestyle Analysis – Hidden Sugars
VIDEO: Introduction to Gen AI & Using CoPilot in Yr 8 PDHPE Task 1
In this video, Mr. (Library) Jones summarises the key ideas from this page and demonstrates the process of using CoPilot for the Yr 8 PDHPE assessment task.
If you missed the lesson in the library, or if you would like to revise any parts of the instructions given, please watch the video. Timeline of subjects covered:
Start to 9 min 30 sec: What is Generative AI? / What are some of the risks of using Gen AI? What precautions should you take?
From 9 min 30 sec: How to write your CoPilot prompt for your Yr 8 PDHPE Task 1
From 17 min 40 sec: Entering the prompt into CoPilot, evaluating the CoPilot response, and creating follow-up prompts to improve the CoPilot response.
Introductory Notes
This webpage has been created to support Year 8 students undertaking the PDHPE assessment task "Year 8: Healthy Lifestyle Analysis – Hidden Sugars". The full assessment task notification is available in your Google Classroom for PDHPE.
This assessment task requires students to use the generative artificial intelligence (Gen AI) program called CoPilot (https://copilot.microsoft.com/) from Microsoft.
The ethical and safe use of generative artificial intelligence is addressed on this webpage; however, you must understand the following:
Never put sensitive information or personal data into Gen AI tools (like CoPilot): Do not enter personal information about yourself or anyone you know. Sharing personal information can compromise a person's privacy and safety; legal issues can occur from inappropriate sharing. Examples of information about yourself or another person that should not be shared include:
Your name, home address, or phone number,
Passwords or login details; your school timetable,
Financial information like bank card numbers,
Medical information about you or people you know family
Private photos or videos of yourself or others.
Notes on Privacy & NSW Department Of Education Guidance on the use of Gen AI (Click to read)
The following resources will help you understand issues relating to your privacy when using CoPilot:
Microsoft CoPilot: Privacy Statement
NSW Department of Education: Guidelines regarding the use of generative AI:
NSW Government: Generative AI: Basic Guidance
Glossary
Artificial Intelligence (AI): Any intelligence demonstrated by machines, in contrast to the natural intelligence displayed by humans and other animals. These machines use mathematical models that identify and encode patterns in data sets, which can then perform predictions on new situations which they haven’t encountered before.
AI-assisted writing: When artificial intelligence software utilizes existing content to predict, modify or create text based on input that a user supplies it. Certain tools may create novel bodies of text, while others may reword existing text in the case of AI paraphrasing tools.
AI-generated text: Text created by artificial intelligence based on vast amounts of data of existing content from the internet.
AI paraphrasing: AI paraphrasing refers to the use of AI techniques to rephrase or rewrite a given piece of text in a way that preserves the original meaning of the text while using different words and phrases.
Algorithm: A set of instructions or computations that a machine follows in order to learn how to do a particular task.
Chatbots (or “bots”): A program designed to help human users with simple tasks and communicate via voice or text commands in order to feel like a human-to-human conversation.
ChatGPT: ChatGPT (short for Chat Generative Pre-trained Transformer) is a chatbot launched by OpenAI in November 2022. It is a Large Language Model with both supervised and reinforcement learning techniques. CoPilot can produce a body of unique text from a user’s specific input based on existing content from the internet.
Generative AI or Gen AI: This is a broad label that's used to describe any type of artificial intelligence that uses learning algorithms to create new digital images, video, audio, text, or code.
Hallucinations: AI data is based on numerous sources including sites like Wikipedia and Reddit that contain mistakes. CoPilot can generate entirely new text, that is convincing, irrelevant, nonsensical, and wrong.
Large Language Model (LLM): Artificial intelligence that has been trained on massive quantities of text data, or datasets, to produce human-like responses to natural language inputs.
Machine intelligence: An overarching term for different types of learning algorithms, including machine learning and deep learning.
Machine learning: A subset of AI that is specifically focused on developing algorithms that will help machines to learn and change in response to new data, without the help of a human being.
Natural Language Processing (NLP): Technology that allows machines to determine–via text or by voice–what humans are saying.
Prompt: A prompt is natural language text describing the task that an AI should perform.
(This glossary is copied and adapted from: https://www.turnitin.com/blog/ai-terms-education-a-glossary-of-what-you-need-to-know)
More glossary terms from the NSW Department of Education can be found at A Common Understanding: Simplified AI Definitions from Leading Standards. This is not required reading, but it is interesting if Gen AI is very interesting to you.
Introduction to Artificial Intelligence
Let's Start With Generative Artificial Intelligence
Generative AI is a sophisticated computer program designed to understand and generate human-like text. It works by analysing vast amounts of written material, including books, articles, and online content, to learn patterns and structures of language and concepts.
When you interact with CoPilot by asking a question (providing a prompt), it uses its learned knowledge to formulate a response. Think of it as tapping into a massive database of information, where CoPilot sifts through data to provide relevant answers delivered in human-like text responses.
More Types of Artificial Intelligence
Artificial Intelligence (AI) can be categorised into three types: Narrow AI, General AI, and Super AI.
Current tools like CoPilot and ChatGPT are classified as Narrow AI.
General AI and Super AI are currently only theoretical and we do not know if they will ever be created.
Types of AI: Narrow AI, General AI & Super AI (Click here for optional reading)
Narrow AI: Also known as Weak AI
Focuses on one narrow task and cannot perform beyond its limitations
Examples include facial recognition, speech recognition/voice assistants, driving a car, or searching the internet
Narrow AI is the current state of AI technology (including CoPilot) and is becoming increasingly common in our day-to-day lives as machine learning and deep learning methods continue to develop.
General AI: Also known as Strong AI
Refers to where AI will (or may) be in the future
Learns from its surroundings and responds to them itself
Can think like humans without being programmed by humans
Machines with strong AI will be able to exhibit self-awareness and emotions
General AI is theoretical and not yet made.
Super AI:
Surpasses human intelligence and can perform any task better than a human
Also known as Artificial Super Intelligence
Develops self-awareness by learning on its own
Super AI is also theoretical and not yet made.
Risks of Generative AI in Learning
Note: In this assessment task you are required to to use CoPilot. In this context we are intentionally using Gen AI (CoPilot) to learn more about how it operates, we are learning how to critically evaluate the responses produced by Gen AI, and we are learning about how to operate it more skillfully.
The following list of risks of using Gen AI refers to the general use of Gen AI for personal learning, homework and assessment tasks.
Generative AI can make significant mistakes. It can't tell fact from fiction and responses can be a mix of truth and story.
If you do not acknowledge that you have used Gen AI, it is cheating.
Students build academic skills through doing school work and assessments. Letting Gen AI generative AI do the work takes away that opportunity to grow and learn your academic skills.
Letting Gen AI do the work removes your personality, your voice, and your perspectives from the learning product.
Gen AI will not be with you when you do not have access to the internet. That includes times when you will rely on your own ability to read, analyse, interpret, generate ideas, think critically, think reflectively, and communicate.
Strategies for Safe & Ethical Use of Generative AI
1. Safeguard Personal Information
The NSW Department of Education (2023b) states that it is important that you keep your personal information private.
If you sign up for an account you need to provide some personal information. Whenever ever possible, reduce the amount of information you provide in the sign-up process.
When you are putting prompts into generative AI do not include personal information like your name, address, class, school, family details, photographs of yourself or any information from which you, another student or other individual can be identified. A safe prompt can include information like “I am a Year 12 Biology student in NSW…” More personal information that you include in prompts can be added to the database of information that the generative AI tools hold.
2. Verify and Edit AI Generated Content for Accuracy and Suitability
The section above titled “Risks & Benefits of Gen AI in Learning” explains that it is important for you to check any information provided by Gen AI is correct. Find the information from a reliable source, then cite and reference the new source.
3. Use Effective Prompting
What is a prompt?
A prompt is a message or instruction entered into the Gen AI guiding it to create something specific or help you with a task.
Writing a clear and effective prompt improves the quality of output.
Prompts can be simple, they can also be complex to the point of being an exercise in logic and planning to undertake complex and difficult tasks.
Types of Prompts (Click here for optional reading)
Following is a list of types of prompts and illustrative examples. These are included as part of a general introduction to Gen AI and almost none of these are relevant to your PDHPE assessment task.
For this PDHPE assessment, you will create a prompt that is a more detailed version of a Specific Instructions Prompt, this will be described further down this page.
Some prompts below include sections in square brackets, E.g. "I am at a Year [INSERT YOUR YEAR] level". If you copy a prompt like this, edit it to suit your situation, E.g. "I am at a Year 12 level".
Adjusting prompts to meet a need for a higher or lower level of language or conceptual difficulty is a chance for you to customise learning content to suit you, when used well, this can be like having a private tutor who caters to your specific needs.
Here are examples of types of prompts:
Direct Question Prompt: Asking a straightforward question.
Examples: "What is the capital city of New Zealand?"
Specific Instructions Prompt: Providing detailed directives.
Examples: "Describe the water cycle in 3-5 sentences."
Input/Output Prompt: Giving a specific input and expecting a calculated output.
Example: "Given these numbers [1, 3, 5, 7], what's their sum?"
Evaluative Prompt: When you have already made an initial prompt and received a response, an evaluative prompt is when you respond and ask follow-up questions.
Example: "Based on your last answer, evaluate the benefits of exercise."
Self-Critique Prompt: Asking the generative AI to evaluate its own outputs.
Examples: "Review your last response and suggest improvements."
"Review your last response, cross-check this against multiple information sources that are credible, suggest improvements and explain why they are improvements."
Role-play Prompt: Setting a scene for the AI. Examples:
"You are a teacher helping a student. Teach me…."
"As an assistant, recommend three productivity improvement methods."
You are William Shakespeare, think, talk, and write as though you really are William Shakespeare. I want to have a conversation with you, prompt me with a question, I will respond, and I need you to convincingly continue the conversation by responsively responding to my answer and ask follow up questions until I type "END".
Tutor-me Prompting: This is an extension of role-play prompting in a specific context. Prompting the AI to act as a tutor, teach a concept, ask you questions to check for your understanding of that information, give you feedback on your response, and continue this process until you tell it to stop. Example:
“ChatGPT, can you quiz me on the [INSERT TOPIC]? I am at a Year [INSERT YOUR YEAR] level, use language and concepts appropriate to this level. Please ask me one question at a time, starting with a broad overview and then diving deeper. After I provide my answer, give me feedback on what I have correct and how I can improve. If answer a question wrong twice please explain the information at a simpler level and then check for understanding with another question. Keep asking questions and providing feedback until I type 'STOP'.”
Recursive Inquiry: Engaging the AI in a dialogue until it gains sufficient information to perform a complex task.
Example: "ChatGPT, I need help with my project. Please ask questions until you understand the task thoroughly, then assist me."
(“Types of Prompts” was in part copied & adapted in parts from Machine Learning Mastery (2023)).
Tips on Writing Prompts
One small tip is that it is easier to draft your prompts in a separate Word Doc, that way you can edit your prompts easily, save them for future reference, and importantly, keep them as evidence if you need to demonstrate how you did an assessment task.
You can improve your prompt writing by experimenting, reading sample prompts like those above, and even by having ‘conversations’ with Gen AI asking for its help in how to write or refine prompts.
One structure, that is suggested by Danny Liu of Sydney University, suggests using clear steps in creating a prompt (2023). A prompt can describe:
Role (ChatGPT act as…)
Task (summary of what the AI needs to do)
Requirements (what the completion needs to include, contain, be, etc)
Instructions (what the AI should do to act on the prompt)
Designing Your Prompt For the PDHPE Assessment Task:
In step #1 of your PLANNING TEMPLATE, create your prompt by starting with the structure of "Role, Task, Requirements, & Instructions". Then copy sections of text from the assessment task notification. You will need:
text from the task description,
details of what is required in the report,
descriptions from the Assessment Criteria 2
Tip: Use the descriptions from the "A" section beginning with "Demonstrates excellent knowledge and understanding of - the effects of sugars..."
2. Enter your prompt into CoPilot (https://copilot.microsoft.com/), then read the response from CoPilot and evaluate it against your prompt, then evaluate the response against the assessment task. After this, you have a few options:
If you think the response does not match what you have described in the prompt either revise the first prompt or add your secondary requirements as a follow-up prompt.
In step #2 of your PLANNING TEMPLATE, record your new prompt or record any additional prompts that you enter into CoPilot. This back-and-forth cycle of entering prompts, evaluating the response, and entering additional prompts to seek a better CoPilot response is an important part of working with any Gen AI.
In your assessment task, assessment task 'Criteria 1' calls for you to demonstrate an "excellent understanding and knowledge of all steps of the planning process". With this in mind, each time you revise your prompt and record it at step #2 of your planning template, include some notes explaining why you have made this additional prompt or revised prompt and what you intend on achieving- this is a way to demonstrate your understanding of the planning process.
When the CoPilot response achieves your goals, make sure you are able to identify the source of the key ideas (5) that you are going to use in steps #4 and #5 of the PLANNING TEMPLATE, go to the source (webpage) of the information (this process is demonstrated in the video further below at 21 min 50sec) and then go through the CRAAP evaluation process; this is how you will "be able to demonstrate that you use "reliable and accurate sources".
From this point in the process, follow steps #3 and #4 in the PLANNING TEMPLATE and continue as described in the assessment task.
CRAAP Testing for Evaluating Websites and Information
Below is a copy of the CRAAP Testing Guide that you are required to use for evaluating and rating the information and webpages that CoPilot identifies as the source of information.
A demonstration of using this tool in conjunction with using CoPilot can be seen in the video that can be found further below on this webpage
CRAAP-Testing.pdfClick here to download a copy of the CRAAP PDF's.
Optional Tools for Evaluating Media Bias and Fact-Checking
The following websites are tools that can be used to check the quality of information sources, they tend to evaluate news websites (sources of information) for bias, and sometimes accuracy. Using these tools is not a part of the assessment task; these are only provided if you are curious about digging deeper into how you can evaluate sources of information for reliability.
Media Bias/Fact Check (MBFC News), founded in 2015, is an independent online media outlet that evaluates the level of bias of a media publication. MBFC News is dedicated to educating the public on media bias and deceptive news practices. There are currently 2300+ media sources listed in their database.
FactCheck.Org (use the search button, enter a subject or fact that you are investigating)
"We are a nonpartisan, nonprofit “consumer advocate” for voters that aims to reduce the level of deception and confusion in U.S. politics. We monitor the factual accuracy of what is said by major U.S. political players in the form of TV ads, debates, speeches, interviews and news releases. Our goal is to apply the best practices of both journalism and scholarship and to increase public knowledge and understanding."
The FactCheck website also has a SciCheck page that allows you to search for articles on specific topics; controversial topics are more likely to be featured here.
AllSides has a specific page that presents a current news topic and then provides you with three different articles representing that news story, one right bias, one centre, and one right. This is an invaluable resource that allows an easy comparison of how differently one story can be told depending on an organisation's bias. View at: https://www.allsides.com/unbiased-balanced-news
AllSides also features a page that has a variety of tools that you can explore on topics including Media Bias Chart, Misinformation & Disinformation explainer, and the Rate Your (own) Bias page. View at: https://www.allsides.com/schools/resources