SDC AI Guidelines

Learning Objectives

Start defining key terms: LLM and Machine Learning.
Read through the AI guidelines for SDC
Explain how we can use AI in acceptable ways whilst we’re learning on the SDC
Give examples of how we can check we’re using AI in acceptable ways for our learning journey

What is AI? (5 minutes)

Start forming a definition of AI on a shared board.

Facilitator should popcorn around the group asking: what kind of definitions can we come up for “Artificial Intelligence”? Write down any common themes.

What does the group think of this definition: A system we cannot imagine writing an algorithm for.

Discuss any similarities or differences this definition has with the definition created by the class.

The point is, the definition of AI is fuzzy and hard to nail down, it seems to change as technology change. Similarly industry is constantly moving and shifting, ask 5 different industry professionals and they’re likely to have different approaches and opinions on AI use. Their companies most likely have different policies on its use.

Let’s nail down two fundamental ideas (max 20 minutes)

For each of the following questions:

First, spend some time discussing people might answer the question. Note down on a shared Miro board what ideas the group has.
Then discuss the defined idea written below the question - does it match what the group said?

Spend 5 - 10 minutes per question depending on class needs.

1. What is machine learning?

Broadly defined as a form of training that allows an algorithm to learn patterns through training with existing data - to the point where the algorithm can make more accurate predictions on new data.

2. What is an LLM?

A large language model, is a type of model trained, through machine learning, to predict the next word in a sentence. It’s often trained on huge amounts of data, often the internet. It is often seen to understand and generate human-like text content. Importantly, LLMs work by predicting the words that should follow an input, like a massive autocorrect. Knowing this: how important is the quality of the prompt you give it?

Read + discuss the SDC AI guidelines (max 15 minutes)

Individually, for maximum 5 minutes read through the guidelines for AI use on the SDC.

🕹️Class Discussion, 10m

After the 5 minutes is up, set a timer for max 10 minutes. As a class discuss the following questions:

What skills are we trying to build during the SDC? How is this different to the ITP?
How could AI interact with these skills? Where could it disrupt and where might it assist us in our learning journey?
What is an example of acceptable AI use?

There might be multiple different approaches to the questions. Discuss and see if there is a consensus or disagreement across the group.

How has AI tooling worked for you so far (15 minutes)

We’re at a stage in our learning journey where we have more foundational skills. We can now engage more meaningfully with AI tooling whilst keeping a critical eye on what is helpful or not to our own goals.

As a group, share your experiences of AI tooling use in and out of your CYF journey so far. Consider the following questions:

Has it ever helped you in navigating something you previously would not have been able to do?
Has you ever used AI tooling and been more confused or “in the dark” than when you started or when you had to revisit the topic?
How have your uses of AI matched with the guidance given? Would this use be acceptable or not?
If it was not acceptable use of AI - what could we change to make that use more acceptable?

This is a good time to share tips that help with learning. E.g. how to make VSCode less “predictive” so you can write code without being prompted all the time, or how to prompt an LLM to be more “teachy” and not give answers.