Artificial intelligence (AI) and the library

Before you use any AI tool on an assignment, check with your instructor on their policies on use of AI to make sure they are OK with it. There may be information in the course syllabus or assignment instructions, but ask if you don't see it. You'll always want to make sure that you are using AI as a tool and not a substitute for doing your own work.

AI assistants like ChatGPT can help with many tasks during the writing process:

• Brainstorming topics. Do you have a general idea for a topic area you want to write about? AI might be able to suggest more specific topics within that area. You could ask what the current issues are for researchers in a field, or ask it to summarize various points of view on controversial issues.
• Suggesting search terms. You'll come up with keywords when you need to search for sources, but if you're not finding much, AI might be able to suggest additional or alternate search terms. (You can always ask a librarian for help with search strategies, of course!)
• Drafting an outline. When it comes to the actual writing, sometimes staring at a blank page trying to get started is the hardest part. You can ask AI to draft an outline for your paper, giving it important concepts you want to include. It probably won't be perfect, and maybe not even good, but you can use it as a starting point to improve with your own ideas.
• Drafting a summary or abstract. Given the text of your paper, AI can draft a summary of it. Again, you'll want to review and improve any text the AI suggests.
• Checking grammar and Improving writing. Are there sections of your paper where the writing seems awkward or not quite right? Ask AI to provide suggestions for rewording them. But be sure that the finished paper reflects your own voice.

AI chatbots are designed to be conversational, so you can talk to them just like you might text a person. But there is a skill to developing effective prompts in order to get the clearest and most relevant results. In his article "The CLEAR path: A framework for enhancing information literacy through prompt engineering," Leo Lo (2023) proposes a five element framework for constructing effective prompts:

1. Concise: brevity and clarity in prompts
2. Logical: structured and coherent prompts
3. Explicit: clear output specifications
4. Adaptive: flexibility and customization in prompts
5. Reflective: continuous evaluation and improvement of prompts

The last of these may be the most important. ChatGPT in particular remembers your conversation for context rather than responding to each prompt in isolation. So you can ask additional questions, revise the prompts, and give it additional information or requirements to improve the results.

Remember that you are responsible for the writing you include in your assignments! The Large Language Models (LLMs) that power AI chatbots are trained on huge amounts of text, but may not have access to current information, and they may "hallucinate," producing incorrect factual claims or nonsense. For example, when AI is used in academic research, it commonly produces imaginary citations for books and journal articles that do not exist.

Before you rely on information provided by AI, make sure you read laterally. That is, instead of just reading up and down the text from the AI, move sideways to another tab to verify the claim. Can you confirm the information with sources found in library databases, or can you find a reliable source for it with a quick Google search?

Chatbots, with the right prompts, are very good at generating text content and ideas, but keep in mind that they are (mostly) not reliable as information retrieval systems. You will still need to do research to provide reliable information to support your claims.