Tuesday, January 24, 2023

What College Professors Should Know About ChatGPT

 Or... How Educators Lost to the Robots

The following is adapted from a talk I gave to the Harding faculty yesterday.

Good afternoon!  I've been given a few minutes to share with you what I know about ChatGPT.  I was wondering why Associate Provost Dana Steil asked me to give this talk.  Then I remembered this email I sent him and my Comp Sci colleagues on Dec 2, just two days after ChatGPT was released by OpenAI. The subject line read: We're in trouble...

I had fed my programming class assignments to ChatGPT, and it had produced perfect or near perfect solutions in seconds. Of course I was quite alarmed and wanted to get my colleagues' attention.

(Side note: I sent a similar email to my colleagues at zyBooks, and it sparked a lot of great discussion.)

Let me backup just a little for those of you who may not be familiar with ChatGPT. ChatGPT is a chatbot created by OpenAI, an AI research laboratory in San Francisco. A chatbot is an application that can chat with you in a human-like manner. GPT stands for Generative Pre-trained Transformer, which is a technical explanation as to how it was created. Basically researchers created a neural network, which is loosely modeled after the human brain. They fed the neural network billions of words from books, conversations, and articles from the web and trained it to make connections between words and even ideas.

You can use ChatGPT by going to chat.openai.com. Here's a screenshot of me asking ChatGPT to summarize in three sentences what ChatGPT is.


ChatGPT is capable of more than just chatting with a human. It can:

  • write poems, entire essay, or research paper on any number of topics.
  • solve difficult math problems
  • write programs in a number of programming languages

The ChatGPT responses are so good that most of us are unable to tell they are AI-generated.

I asked ChatGPT about it's significance: How significant is the invention of ChatGPT?  It responded: "ChatGPT represents a major advancement in the field of natural language processing."  That is an understatement.  ChatGPT is the best AI chatbot ever released to the general public.  Many technologists suggest that ChatGPT is probably going to be one of the most transformative technological innovations we've seen, right up there with the invention of the Internet and the iPhone.

Back to my alarmist email.  When I figured out that ChatGPT could write all the programs for my students, I wondered if it was going to make human programmers obsolete?  Would it make my programming classes obsolete or me as a teacher obsolete?  Many of you may also be feeling some amount of fear in how ChatGPT is going to affect your fields.

The Bible is full of passages encouraging us to be strong and courageous.  Christians should not be people who live in fear.  But I'll admit that I'm still struggling with that fear.  I was curious what ChatGPT thought: Should faculty be scared of ChatGPT?  This is the response I got:

"It is understandable that some people may be concerned about the potential impact of advanced language models like ChatGPT on their jobs or fields of expertise. However, it is important to remember that ChatGPT is a tool, and like any tool, it can be used for a variety of purposes, both positive and negative."

I really liked this response.  Like any new technology, we as faculty need to have the proper perspective that ChatGPT is just a tool.  Rather than avoid the tool, we need to train our students how to use such tools for positive purposes.

But many of us are of course worried that students will use these tools for negative purposes, like asking ChatGPT to complete their writing assignments for them.  So I asked ChatGPT: How do faculty stop students from cheating with ChatGPT?

Here are six ideas it suggested:
  1. Use plagiarism detection software: These tools can help detect when a student has submitted work that is not their own.
  2. Set strict guidelines for using AI-based writing assistance: Clearly communicate to students that using AI-based writing assistance is not allowed and what the consequences will be if they are caught.
  3. Use original topics: Assign topics specific to the class and not easily found online.
  4. Encourage students to think critically: By teaching students to think critically, they will be better equipped to identify and challenge the limitations of AI-based writing assistance.
  5. Have human interaction: Have a face-to-face or phone conversation with the student to discuss the topic and their writing process.
  6. Monitor student's work for inconsistencies: Look for inconsistencies in writing style, language, or tone that may indicate the use of AI-based writing assistance.

Currently, Turnitin won't help you since they are only comparing submissions with known texts. But I saw a beta version that Turnitin is working on that did an amazing job of identifying AI-generated text. Of course the next version of ChatGPT is likely to bypass the detection, forcing modification of the tool in a never ending cat-and-mouse game.

The biggest advice I can give is to monitor the process instead of just evaluating the final product. For example, require students to use Google Docs for writing assignments since Google Docs tracks every single keystroke, making the entire history of the document available for faculty to see.

Many of you have probably discussed detection and deterrents in your departments, and if you haven't already had those discussions, I encourage you to do so soon. It's unlikely that you can just keep doing what you've always done.

Here are some resources that some of you have passed on to me that you thought were really helpful in grasping what this tool is and what to do about it:

I wasn't sure how to end this talk, so I asked ChatGPT to do it for me: Write a funny ending to the conclusion of a talk about ChatGPT. Here's what it said:

"In conclusion, ChatGPT is a game changer in the world of natural language processing, it's like a magic genie that can answer all your questions. But don't worry, it's not going to take over the world, yet. Thank you for listening and I hope you enjoyed the talk, now go and create some hilarious chatbot or a virtual assistant that will entertain us all. And remember, always keep an eye on your virtual assistant, you never know when it might turn into a Skynet."