What is AI?

Artificial Intelligence (AI) is defined by Furze (2022) as machine intelligence that is comparable to that of a human. There are two different types of AI:

1. General/strong AI - currently only available/possible in books/films (think Skynet in the Terminator films).
2. Narrow/weak AI - such as newsfeed algorithms or systems used to drive autonomous cars.

What is Generative AI?

Generative AI or Gen AI is an example of a narrow or weak artificial intelligence capable of generating text, images or other data using what are known as generative models, often in response to prompts. Gen AI models learn the patterns and structure of their input data and then generate new data with similar characteristics. After a Gen AI model is ‘trained’ on a dataset, it can make predictions or generate outputs based upon new (unseen) input data. But as Mike Sharples from the Open University comments, the models produce continuations not facts i.e., they are not knowledge bases (and should not be treated as such).

What are LLMs?

Large Language Models (LLMs) are a specific type of GenAI. ChatGPT is an example of an LLM and stands for Chat Generative Pre-Trained Transformer. ChatGPT has been given lots of written text as its data source. It can generate coherently structured writing such as natural-sounding text or complete sentences and paragraphs. The more data inputted into the model, the 'better' writing it produces. LLMs use billions of webpages as their data source but they are

How does Generative AI software work?

The user enters a prompt and the LLM creates new text based upon the prompt and lessons it has ‘learned’ from written text it has previously been 'given.' ChatGPT is one example of a text-based generative AI that can generate text. Other types of generative AI can generate images or other data types from text and there are some that can generate text from image data sources. Essentially, Generative AI models learn the patterns and structure of their input training data and then generate new data that has similar characteristics.

How can Generative AI be used ethically in teaching and learning?

CTE adheres to the Russell Group principles on the use of generative AI tools in education (see the right-hand column for a copy of these).

• Content repurposing
• Supporting educators in the moment

Issues with AI:

• Positionality and power
• Synthetic media
• Social mobility issue (prioritise things that can't be done by AI
• Environmental issues - power used to undertake AI activities

Principled refusal - things we can offer in HE

Aberystwyth University resources

WEIRD acronym - lenses

Subject experts should be experts in using AI in their subject.

Definition of AI tools - are we including Grammerly or Google translate generative?

Further resources

• Policy - Institutional Approach to the use of Artificial Intelligence and Academic IntegrityLink opens in a new window
• Overview - Principles of good assessment design that support the use of artificial intelligenceLink opens in a new window
• Case Studies - ANTF Gen AI Innovation ProjectLink opens in a new window
• Paper - WIHEALink opens in a new window

Russell Group principles on the use of generative AI tools in educationLink opens in a new window