The World of AI!

Artificial Intelligence (AI) is rapidly changing our world, and education is no exception.This guide is designed to help you, as a Scottish educator, understand the basics of AI, machine learning, and generative AI, and explore how these technologies might impact your practice. We’ll break down the jargon and focus on practical applications relevant to Scottish Education.

Don’t worry if you’re new to this! This page is a starting point, and we’ll focus on clear, simple explanations. We’re here to help you feel confident and informed.

We have made an AI Readiness Self Evaluation tool which has been designed to help schools understand where they currently are, and what they might want to prioritise in terms of AI use in their setting. You can find the tool below.

Demystifying the Terminology

“All GenerativeAI is AI, but not all AI is GenerativeAI”

What’s the Difference? AI, Machine Learning, and Generative AI Explained

Artificial Intelligence (AI)

AI is the broad concept of machines that can perform tasks that typically require human intelligence. 

Artificial Intelligence is not a new technology, in fact, as far back as 1956, the term Artificial Intelligence was first used. To learn more about the Evolution of AI, visit our dedicated page.

Think of AI as any technology that can carry out task that would normally require human intelligence.

For example, imagine a calculator that can do more than just add and subtract. It can analyse large datasets to find patterns – that’s a simple form of AI.

Key Message: AI is the overarching term.

Machine Learning (ML)

Machine learning is a subset of AI where computers learn from data without being explicitly programmed. They improve their performance over time.

Think of it as a pupil learning from practice tests. The more tests they take, the better they understand the material. ML systems do the same with data.

Examples for Education:

• Systems that can analyse pupil performance data to identify areas where they may need extra support.
• Software that can help to mark multiple choice tests.

Real-Life Examples:

• When a streaming service recommends new shows based on your previous viewing habits, that is a form of AI that improves with the more shows you watch.

• When an online retailer recommends products based on your previous purchases or those made by others who bought a similar item it using Machine Learning to build an overview of what you may wish to purchase next.

Key Message: ML is how AI learns from data.

 

Generative Artificial Intelligence (genAI)

Generative AI is a type of AI that incorporates Machine Learning to create new content, such as text, images, music and various forms of media type. It learns patterns from existing data and then generates something new based on a user prompt.

GenerativeAI is becoming more commonplace in our everyday lives,. The same online retailer that uses Machine Learning to recommend products; can also use AI-powered features to provide a short paragraph right on the product detail page that highlights the features and customer sentiment mentioned across written reviews to help customers determine at a glance whether a product is right for them.

Imagine you’re teaching a dog a new trick. You don’t just show the dog what to do, you also give it treats when it does well and correct it when it makes mistakes.

Reinforcement Learning from Human Feedback (RLHF) is a way to train genAI models (like the ones that power tools such as ChatGPT) by giving them feedback on their responses. The more feedback the tool receives, the better the result gets over time.

Here’s a simplified breakdown:

1. Initial Training: The AI model learns the basics of language from a huge amount of text data.
2. Human Feedback: People provide feedback on the AI’s responses, telling it what’s good, bad, or needs improvement.
3. Learning and Adapting: The AI uses this feedback to learn and get better over time, just like your dog learning from your guidance.

This helps the AI understand what kind of responses are helpful, safe, and appropriate in different situations. It’s like having a continuous loop between humans and the AI tool, making it more aligned with our expectations.

How does genAI relate to Education?

“Imagine a tool that can help pupils brainstorm ideas for a creative writing project or generate different versions of a historical narrative.”

Education examples:

• AI Tools that can create personalised learning materials.
• AI Systems that can generate different creative writing prompts.
• AI Tools that can generate summaries of complex texts.

Key Message: GenAI creates new content.

How AI Tools Support Teaching & Learning:

Personalised Learning:

• AI tools can help tailor learning experiences to individual pupil needs, providing targeted support and resources.
• This can provide teachers with more time to focus on the individual needs of their learners.

Assessment, Monitoring & Tracking:

• AI tools can assist with automated marking of certain assessments, providing quicker feedback to pupils.
• AI tools could provide insights into pupil progress, helping to identify areas for improvement.

Administrative tasks:

• AI tools can help streamline tasks like scheduling, data analysis, and communication, freeing up teachers’ time.

Creative & Innovative approaches to teaching & learning:

• GenAI tools can be used to create engaging learning materials, such as interactive quizzes, simulations, and creative writing prompts.
• It can support pupils to explore new ways of learning and creating.

Teacher comments that show AI tools being used:

Ethical Considerations and Responsible Use

Thinking Critically about AI

Data Privacy & Security:

• It’s essential to consider how pupil data is collected and used by AI systems. Adhere to data protection regulations.

Bias & Fairness:

• AI systems can sometimes reflect biases present in the data they are trained on. Be aware of potential biases and strive for fairness.
• Gen AI tools produce outputs based on a data set, and of course the prompts. 
• There is obviously scope for bias in both areas.
• If the training data is unbalanced or has certain patterns (like underrepresenting certain groups), the model picks that up bias and reflects it in its outputs. Prompts, on the other hand, act like a steering wheel—they guide how the bias in the dataset shows up in responses.
• For example, if a model was trained on data that mostly portrays CEOs as men, a simple prompt like “create an image of a CEO” might give you a male-dominated response. But if you tweak the prompt to “create an image of a CEO who is a woman of colour,”  you can help counteract that built-in bias.
• Still, prompts can only do so much- if the dataset itself is skewed, the bias is already baked in and therefore the outputs will show that.

Digital Literacy & Critical Thinking:

• Educate pupils about the capabilities and limitations of AI, promoting critical thinking and digital literacy skills.
• Ensure that pupils understand the ethical implications of using AI.

Maintaining Human Connection:

• AI should be seen as a tool to support, not replace, human interaction and connection in education.

Understanding the terminology

There are a number of associated acronyms associated with AI that can seem overwhelming. To make this easier to follow, we have collated the key terms into a single page, which you can visit below: