Introduction

This guide is intended to provide a brief overview of and resources about Generative Artificial Intelligence, and Large Language Models, in particular ChatGPT. It is not a list of best practices for using these tools but should be used as a starting point for conversations about how these tools.

ChatGPT has gained a lot of attention since launching in late 2022, but what is it exactly? 

ChatGPT is a generative artificial intelligence (AI) chatbot that has been developed by OpenAI. It is an example of a Large Language Model (LLM) which includes Bard and Gopher from DeepMind (Google), and BLOOM from BigScience. It is a variation of OpenAI's GPT-3 (Generative Pretrained Transformer-3) and has been fine-tuned specifically for conversation response generation and trained on a large dataset of text, allowing it to generate responses. ChatGPT was trained on 175 billion parameters with Gopher from Google being trained on 280 billion parameters. These parameters include internet sources up until June 2021 such as digitised books from The Gutenberg Project and Bibliotik as well as approximately 61GB of Reddit and Wikipedia pages. 

But what does all that really mean? Well, we asked ChatGPT to describe itself in layman terms and this is what it advised: 

ChatGPT is a computer program created by OpenAI that can understand and generate text like a human. It has been trained on a huge amount of text, so it can respond to questions and have conversations in a way that sounds natural. For example, if you ask ChatGPT a question, it will use what it has learned to generate a response that makes sense and is relevant to your question. This can make it easier and more enjoyable for people to communicate with technology.

Even simpler put it's an online AI chatbot that can answer questions, explain complex topics; and write code, emails and assignments. Or at least try to. 

The answer to how ChatGPT works is in its name, specifically GPT, or Generative Pre-trained Transformer. 

• Generative reflects its ability to produce text that resembles natural human language text
• Pretrained means that the model has been trained on a specific dataset; this includes a large amount of conversational data 
• Transformer refers to the machine-learning architecture 

All together this means that ChatGPT works by predicting answers to questions, based and all of the data it has been fed -- similar to studying and cramming before a test a high school math test. However, unlike humans, ChatGPT doesn't change its answers based on feedback, yet.

ChatGPT didn't come out of nowhere though. The technology that runs ChatGPT has been around for much longer, just with much less publicity. While the model for ChatGPT is not new, its functionality is based on OpenAI's GPT line of AI models, the interface, layout and user experience are, and this is what has led to its rapid adoption. Upon opening the app, the interface has strikingly similar to other messaging and chat software which allows users to feel familiar and safe using it. 

In late January a paid version of ChatGPT was released, called ChatGPT Plus, original only to US customers with plans to release worldwide. On February 10, it was released in Australia. ChatGPT Plus costs USD $20 per month (plus gst), or approximately AUD $31 per month.  

ChatGPT Plus subscribers receive access to ChatGPT, even during peak times when other users are prevented from accessing; faster response times; and priority access to new features and improvements. OpenAI have also announced they are actively exploring options for lower-cost plans, business plans, and data packs for more availability. 

Thank you to Harper College Library for allowing us to reuse parts of their guide Teaching and ChatGPT.