AI and information literacy

Learn Prompting defines "prompt engineering" as "the process of communicating effectively with an AI to achieve desired results."

Using AI tools effectively requires that the user know the right questions to ask, and how to phrase them for the best results. Vague or generic questions generate vague or generic results. (In other words, garbage in, garbage out.)

To get better results from AI tools and chatboxes, include these components in your prompt:

Who am I?

What's the context?

What's my goal?

What do I want the AI to do?

Example:

I am a community college student writing a research paper for a writing course. We have been discussing topics related to social justice. I want to make sure that my paper relates directly to issues that affect the lives of community college students and that I can find information that will help me to persuade other students. Please brainstorm 3 research paper topic ideas. Limit your descriptions of each of them to a paragraph. Make sure that each description is unique from the other so that I have some diverse ideas to consider. 

This prompt model and example are based on the work of Dominic Slauson. Find more prompt examples following this model on his website, coedaptivelabs.com.

Beware of hallucinations, or the chatbot confidently stating incorrect or made up information! If you are unfamiliar with the topic, check the chatbot's work. Treat the chatbot like a brand new assistant who is eager to please but doesn't know what they don't know.

Evaluating all information for credibility is highly recommended, regardless where you find it. This is true for generative AI responses, especially given the information presented above. There are many different tools, checklists, and strategies to help you evaluate your sources. None of them are black-and-white checklists for determining if a source is credible and if you should use it.

Here are two strategies for evaluating information provided by generative AI tools:

1. Read "laterally"

Don't take what ChatGPT tells you at face value. Look to see if other reliable sources contain the same information and can confirm what ChatGPT says. This could be as simple as searching for a Wikipedia entry on the topic or doing a Google search to see if a person ChatGPT mentions exists. When you look at multiple sources, you maximize lateral reading and can help avoid bias from a single source.

2. Verify Citations

If a generative AI tool provides a reference, confirm that the source exists. Trying copying the citation into a search tool like Google Scholar (make sure it is linked to the Library's database collections!). Do a Google search for the lead author and the publication.

Second, if the source is real, check that it contains what the AI response says it does. Read the source or its abstract.   

Different AI tools are good at different things, and many tech reviewers have published comparisons between ChatGPT, Microsoft Copilot (Bing Chat), and Bard and others. Remember that all of this is changing quickly! Bottom line: look around for current reviews and comparisons.

Chatbot Arena - Created by researchers at UC San Diego and UC Berkeley, Chatbot Arena is a benchmark platform for large language models (LLMs) that features anonymous, randomized battles in a crowdsourced manner.

The best AI chatbots: ChatGPT and other noteworthy alternatives (ZDNet, 6/21/2023) - Compares Bing (selected as best overall), ChatGPT, Perplexity AI, Jasper, YouChat, Chatsonic by Writesonic, Google Bard, and Socratic by Google.

ChatGPT vs Bing Chat vs Google Bard: Which is the best AI chatbot? (ZDNet, 5/30/2023)

ChatGPT Vs. Bard Vs. Bing: What Are The Differences? (Search Enginge Journal, 4/4/2023) - Includes comparison table and comparisons across test prompts.

Awesome-LLM - GitHub repo of featuring curated list of papers about large language models, frameworks for LLM training, tools to deploy LLM, courses and tutorials about LLM and all publicly available LLM checkpoints and APIs:

Open LLM Leaderboard - The Open LLM Leaderboard aims to track, rank and evaluate LLMs and chatbots as they are released. Anyone from the community can submit a model for automated evaluation on the GPU cluster.