AI for Latecomers
A Beginner's Guide for College Faculty
Sorry for the long break between posts. The fifth installment of my series looking at AI through a pedagogical lens was always going to be about AI, but I was struggling with what to include and how I could contribute to a topic that is already saturated. I decided to write the guide I wish I had as I began researching generative AI. What was originally going to be one post about AI has evolved into a series addressing a range of topics:
Getting to Know AI
Creating an AI Policy
Teaching Your Students About AI
Using AI to Help You Teach
AI Wrap Up (all the things that didn’t fit anywhere else)
Enjoy!
Getting to Know AI
ChatGPT has been publicly available for about 2 years and has had a huge impact in a lot of ways. Its exponential evolution has made it a challenge to keep up with what you might need to know. Regardless of how you feel about AI, it isn’t going anywhere If you didn’t try it when it first came out (or haven’t since the early days), it may feel daunting to know where to begin. Hopefully this post can help you get started.
What is AI?
AI stands for Artificial Intelligence, and is a field of study that has been around since the 1950s. When we hear AI, many of us think of ChatGPT or maybe an image generator, but there is so much more to it than that. Below is a brief introduction to some different types of AIs and AI tools. If you want more information, the AI Pedagogy Project from metaLAB (at) Harvard is a great resource for learning more about AI and seeing what other people are doing with it.
LLMs – LLMs are Large Language Models. They are trained (typically) on large volumes of information found on the Internet, and they use that training to generate text. They are not search engines (except for Perplexity), and they are known to “hallucinate,” (i.e. make stuff up) but overall, the content that they generate is effective if not exciting or interesting.
While all of the LLMs perform similar functions, and most have free and paid tiers, they do have their strengths and weaknesses. Below is a quick look at each of the major text generators.
ChatGPT – ChatGPT is the original LLM. When it came out, it was not connected to the Internet, but its newest iteration is not limited to the pre-2021 training data. It is great for brainstorming, generating quiz questions, and explaining complex concepts (Learn more about How to use ChatGPT)
CoPilot – CoPilot comes with Microsoft 365 and uses ChatGPT-4o, so if your school has a Microsoft subscription you might be able to use it for free. (What Is Copilot?)
Gemini – Gemini is Google’s answer to ChatGPT, it also powers the AI summary that sometimes shows up at the top of a Google search. It is useful for research and data analysis, although it has had some high profile gaffs (remember it recommending glue to keep the cheese from sliding off of your pizza) (What is Google Gemini?)
Claude – Claude was created by Anthropic to be a “helpful, honest, and harmless” assistant. It is less well known than the other LLMs and tends to be more conservative in its responses, which can limit creativity. It is probably the most ethical choice among the available AIs (What Claude AI Can Do?). It has some pretty interesting tools that we’ll talk about in a later post.
Perplexity – Perplexity is an AI powered search engine that will respond to your prompt and cite its sources. It is great for research and current events, and while it is less prone to hallucinations, you still need to be careful. (Try Perplexity for Search)
Image Generators – These are another category of generative AI that converts a text prompt into an image. The quality of the image will depend on your prompt, but you can get some surprisingly good results with a little bit of effort. The images that I use for this blog are always generated with AI (mostly because it gives me an excuse to play around). The training of AI image generators has raised issues from artists over copyright (I’ll address this in another post). Some examples of AI image generators are:
ChatGPT (DALL-E) – DALL-E is the image generator associated with ChatGPT. It is available with premium subscriptions, although free users can create 2 images per day.
CoPilot – The image generator in CoPilot is free (with limitations). It can be less customizable than some of the other image creators
Midjourney – This is not a free tool (although you can get a free 7 day trial) and is typically used by serious image creators
Stable Diffusion – This is an open-source image generator that has a free trial, but requires a subscription for long term usage. It is powerful but there can be a learning curve associated with it
Flux – is a newer, free image generator, although as with the other image generators, upgrading to paid will get you more features.
To give you a sense of how an AI image generator works, I’m sharing the output from the prompt: “draw a picture of a confused professor trying to figure out AI” for ChatGPT, CoPilot, Stable Diffusion and Flux. The same prompt was entered for each AI. What do you notice about the images? (hint: we’ll be talking about bias in a later blog post)
Other Types of AIs – There are other types of Ais that don’t fit neatly into the above categories. The list below is just a small subset of those I have found most useful.
NotebookLM – NotebookLM is an experimental design tool from Google. It allows you to upload content (could be your notes, papers, presentation, etc) then use the Notebook as an AI that gives results based on the content that you provided. It can summarize (and provide a podcast version of the summary), generate quiz questions and so much more (Make an AI notebook).
Poe – Poe is a chatbot platform. It allows you to access multiple AI models (including ChatGPT and Claude), you can create your own custom bots, for example a teaching assistant for your course that answers FAQs. (Make your own AI bot)
Elicit – Elicit is an AI research assistant that can help with literature review and data collection. It can help you find relevant research papers and summarize their findings. (Elicit)
Udio and Suno – Udio and Suno are both tools for creating audio experiences (text to voice, music, etc) with AI. Both have limited access for free (Udio, Suno)
Where Else Do We Find AI?
AI has infiltrated much of what we do with technology on a daily basis. Places where you or your students will encounter AI include:
Writing assistants - Writing Assistants like Grammarly can be used to help improve writing in many ways including grammar, spelling style, tone and clarity, it also has built in generative AI that can help with creating content, so it is more than just a grammar checker (do what you will with that information…)
Search engines -- In addition to the AI search engine Perplexity, traditional search engines (like Google) have also begun incorporating AI into the search results, for example, you may have noticed AI summaries at the top of your search results pages. While these can be helpful, the AI doesn’t know how to distinguish between serious and non-serious information (again, remember the glue on pizza…)
Social Media – Social Media use AI algorithms to personize your experience and help to curate your feed and see content and ads targeted to your specific interests. Facebook and Instagram have added the AI Meta to their search function as well
Voice Assistants – Even before ChatGPT, voice assistants like Siri, Alexa and Google Assistant have used natural language processing (a form of AI) to respond to spoken requests
Using AI
Since it is pervasive in our lives, it is definitely time to address it. If you haven’t created an AI policy for your courses, you should, and you should do it from a position of authority, not fear. The rest of this blog is designed to jump start your understanding of how AI works (the next blog will look at creating an AI Policy). My hope is that you will do more than just read, that you will actively engage with AI using the prompts provided in this section.
What to Use?
Any of the AIs listed above will work for playing around. Most of them will either require you to create an account (ChatGPT, Claude) or use an existing account (CoPilot/Microsoft, Gemini/Google). Perplexity does not require you to have an account to use it, but some of the functionality will not be available to you without one.
I’ll take a closer look at privacy concerns in a later blog post, but just be aware that these tools will collect your data as you are using them and may use your data to train future AIs. Never put any confidential information into the AI.
First, Have Some Fun
Fun #1 -- To get your creative juices flowing, try this Connect Random Things exercise from Jon Ippolito at University of Maine. You will ask the AI to find a connection between two seemingly unconnected things (you can choose from the list in the exercise). How did it do? Could you have done better? This exercise shows how AI can use its “training” in creative and surprising ways. It can think outside the box and help us to do the same.
Fun #2 -- Next, pick a topic that you teach and ask AI to describe it in the voice of someone completely different. (For example: Explain the constitution of the United States as if you were Elmo from Sesame Street or Rewrite the Odyssey in the style of the Harry Potter). Again, the goal is to see how creative the AI can be, but this is an exercise that can help you view something from a different angle which could evolve into something you could use with your students.
An Example of Getting Creative
Look for new and different ways to explain things like the Backward Design as a sitcom example below. ChatGPT wrote the script and the video was made with Fliki, all in less than 10 minutes with minimal editing. While the video is a little strange, the point is to show you how easy it can be to create with AI.
But What Can It Do For Me?
There are so many things that you can use AI for to help you in your teaching that I am planning an entire blog post about it. For the purposes of testing things out, I asked ChatGPT to come up with some sample prompts for you to try. I made prompts for a variety of disciplines. Use them to get a sense of what AI can do then adapt them to your specific area and see what you get.
Prompts to use AI to Support Your Teaching:
English Professor Prompt: "My students are reading The Great Gatsby. I want to have a class discussion where we focus on how the themes of wealth and social class contribute to the narrative. Suggest three discussion questions I can use in my literature class to help students engage more deeply with these themes."
Computer Science Professor Prompt: "Explain how recursion works in Python with a simple example, and provide suggestions for common errors students might encounter when implementing recursion. Additionally, suggest an in-class activity to help students understand recursion better."
Biology Professor Prompt: "Create an overview of the cell cycle, including each of the phases (G1, S, G2, M) and the key processes that occur during each. Suggest how I can use this information to create a quiz that tests both knowledge recall and application of concepts."
Spanish Professor Prompt: "Generate a list of key grammar points that intermediate Spanish learners struggle with, such as subjunctive mood and verb conjugations. Provide suggestions on how I can design a lesson plan around these grammar points with examples and activities."
Prompts for Assignments You can Use with Students
English Professor Prompt: "I teach English Literature, and I’m interested in integrating AI into my classroom assignments. Can you recommend 5 assignment ideas that require students to use AI as part of their work and include opportunities for them to reflect on the experience? Each assignment should leverage AI tools in ways that support literary analysis, creativity, or research skills."
Computer Science Professor Prompt: "Design a programming assignment for students where they create a basic to-do list application in Python. Outline the steps involved, key learning objectives, and include a stretch goal for students who are ahead of the curve."
Biology Professor Prompt: "Create a lab report assignment prompt for students who have just completed a microscope lab where they studied cell structures. Include sections for hypothesis, methods, results, and analysis. Suggest what students should focus on in each section."
Spanish Professor Prompt: "Create a role-play assignment prompt where students simulate a conversation between a tourist and a local in a Spanish-speaking country. Provide a scenario, key phrases they should use, and grammar points that they should demonstrate in their dialogues."
The generic prompts above are intended to give you a taste of what is possible when you collaborate with ChatGPT. There is a lot of potential both in the classroom and out. Regardless of how you feel about it, it’s important to know at least a little something about it to be able to discuss it with your students.
Where to Learn More
Below are a few of my favorite resources available to help you explore AI a little more:
Why A.I. Isn’t Going to Make Art – This great article in the New Yorker by Ted Chiang is a great look at how AI isn’t going to replace us. I love the sentiment and wanted to include it to give you a break from all of the technical stuff
AI Pedagogy Project – This is a great site because it has a guide, links to a lot of resources and a searchable database of assignment ideas
Learn With AI – This Toolbox from the University of Maine offers a lot of ways to learn about, think about and use AI
Accessibility and AI Collection – This resource from the University of Virginia Teaching Hub looks at the intersection of AI and accessibility, with plenty of articles to check out
My Recommendation
Ethan Mollick is a professor at the Wharton School at the University of Pennsylvania. He has been using and writing about AI since day 1 and his posts are always thoughtful and engaging. I have learned a lot from his posts over the past 2 years. Here is a link to his Substack: One Useful Thing and its companion site: More Useful Things: AI Resources