• Generative AI refers to technology that can create new content, such as text, images, code, audio, or video. 
• Put very simply, AI text generators work by predicting the next word in a sequence. They do not "understand" your questions in the way that another person would.
• AI text generators, such as ChatGPT, are trained on a large amount of text scraped from the Internet. This training is supplemented by annotation and feedback testing done by humans.  AI image generators, such as Midjourney and DALL-E are trained in a similar way.
• With some exceptions, we don't have a lot of information about how the models that drive these tools were trained. That means that, in most cases, users have no way of knowing what information an AI tool has had access to. 
• Many other applications are being developed that make use of generative AI technology. In fact, AI technology is already at work in many of the tools that you use.

Some limitations and causes for concern:

• Some AI text generators, such as Bing Chat and Google Bard have access to the Internet, but the free version of ChatGPT does not. ChatGPT Plus, which requires a paid subscription, is introducing new capabilities and plugins that do access the Internet.
• Most AI tools capture and reuse a significant amount of user data. You should not share private information with ChatGPT or similar programs.
• All AI text generators can, and often do, produce plausible sounding but false information (known as "hallucinations"), and by their nature will produce output that is culturally and politically biased.
• There are a number of ethical and societal concerns about generative AI, from the way these tools were created, to the way they are being applied now and in the future.

Sourced and adapted from Brown University's Generative Artificial Intelligence Guide and JISC's Generative AI - A Primer.

Artificial Intelligence (AI): A field of study within computer science, focused on the development of computer systems that can accomplish tasks typically associated with human intelligence. These tasks include speech recognition, route mapping, decision making, etc.

Bias: The training data of an AI model can skew the output, leading it to generate inaccurate or offensive material.

Chatbot: A program designed to communicate with humans in a natural manner, sometimes to facilitate providing information or completing tasks.

Chat Generative Pre-trained Transformer (ChatGPT): A chatbot developed by OpenAI. ChatGPT is a transformer type of AI that is designed to mimic conversations using natural language processing, through which users can write prompts to generate text-based responses.

Generative AI: A model of artificial intelligence that can generate new content such as text, images, video, etc., through pattern recognition, by examining large amounts of training data and creating material that contains similar characteristics to identified patterns in the dataset. Examples include ChatGPT, Claude, Midjourney or DALL-E.

Hallucinations: Instances where a generative AI model generates output that contains inaccurate or irrelevant information, especially when it may look correct. For example, when asking ChatGPT (or any text-based generative AI) to generate a list of citations for a topic, the citations it provides may look accurate but the source material associated with the citation may not actually exist when searching for it.

Large Language Model (LLM): An AI model that receives large amounts of training data that establishes the capacity for it to respond to conversational queries. AI such as ChatGPT, Bard, or Claude use LLM.

Natural Language Processing (NLP): The programmed capacity to understand conversations and respond in kind.

Prompt: A structured text-based query that asks a generative AI to generate new content in the form of text, image, video, etc.

Prompt Engineering: The process of refining prompts to elicit more desirable results from generative AI.

Training Data: The development of a generative AI model involves the input of specific types of data, often in large amounts. This process is referred to as “training” and it determines the content output of the specific model. For example, if developing an AI that reviews artwork specifically, the AI model will be trained only on data containing artwork.

Sourced and adapted from Austin CC Research Guide, Brown University's Generative AU as a Research Tool, and School Library Journal.