Chatting with ChatGPT: Deep Dive in Five with Tom Moriarty
Imagine starting a conversation with your computer — or better yet, an app on your phone. Already do that? Now imagine that it writes back in full sentences, leveraging online data and computing speech patterns such that the app appears to have a voice of its own. ChatGPT, a language generation model created by OpenAI, has drawn attention since its launch on November 30 because of its potential to disrupt how and why people write.
San José State University’s College of Humanities and the Arts is hosting two events next week to address some of the most burning questions about ChatGPT. On Monday, Feb. 20, San José State Assistant Professor of Technical and Professional Writing Sara West will be in conversation with Education Professor Roxana Marachi for Quick Bites: Chat GPT: Help or Hype?
The following day, the college, in partnership with Office of the Provost, the Center for Faculty Development and Writing Across the Curriculum program, will host a town hall to invite SJSU faculty to discuss the opportunities and threats posed by ChatGPT. Panelist Tom Moriarty, director of San José State’s Writing Across the Curriculum program and professor of English and rhetoric, agreed to answer a few questions — aided in part by AI — in advance of next week’s events.
For the uninitiated, what is ChatGPT? What does it do?
Tom Moriarty (TM): Just for fun, I thought I’d ask ChatGPT to answer this question. Here is what it said:
“ChatGPT is a language generation model developed by OpenAI. It’s based on the Generative Pre-trained Transformer (GPT) architecture, which uses deep learning to generate text.
“The model has been trained on a large dataset of diverse text content and can generate text that mimics human language patterns. When it is used in a chat interface, users can input a prompt or question, and the model will generate a response based on the input it was trained on.
“ChatGPT can be used for a variety of natural language processing tasks, including answering questions, generating text, chatting, and more. It’s a powerful tool for generating human-like text and can be used for applications such as customer service chatbots, language translation, and content creation.”
How can ChatGPT assist students and educators alike?
TM: I think ChatGPT is great for what I would think of as low-level tasks. It can look up general answers to questions you might have, and it can generate a couple of paragraphs to get you thinking about something you might want to write. But its responses to questions and tasks are very formulaic and, not surprisingly, machine-like. They lack voice and personality, and they lack creativity and insight.
Which is not surprising, if you think about it. Like ChatGPT said above, its job is to generate text that mimics human language patterns, not to offer anything new. So if you need to generate some routine questions to put on your next exam, or if you need to answer those questions as a student, ChatGPT is your go-to gal. Or guy. Or bot.
But if you need something new and original – if you want your students to apply what they’ve learned, for example, and not just regurgitate what they read in the textbook; or if you want to write something worth reading – you’re best left to your own devices. And not your electronic devices. Your mind.
What concerns do you have about ChatGPT?
TM: I’m not really concerned about it. I know everyone’s worried about students cheating and asking ChatGPT to do their homework for them. But I think that problem is quickly going to be moot. I’m sure there are companies furiously working on ChatGPT detectors that they’ll sell to colleges and universities, for the low, low price of $23 a month.
That’s what happened when the Internet was the next big thing that was going to ruin education for all of us. Students could cut and paste content from the web straight into their papers, and how would we know? Turnitin.com makes a bunch of money scanning papers for plagiarism. I’m sure a ChatGPT Detector will be available soon, possibly from OpenAI, the creators of ChatGPT. (Talk about a vertically-integrated business model! Sell the problem to the students and the solution to the teachers.)
What do you wish your students understood about the relationship between writing and artificial intelligence?
TM: Writing is generative and creative. Artificial intelligence applications like ChatGPT are imitative and mimetic (which is a fancy word for imitative). Writing makes you smarter. Writing makes you unique. ChatGPT makes you sound like everybody else. Which, I guess, makes you forgettable.
If you could use the tool to write anything you’d like, what would you ask it to do?
TM: I’d ask ChatGPT to respond to the thinly-veiled insult in my previous answer. Here’s what it said:
“As a language model AI, I don’t have personal opinions, but I can tell you that writing is indeed a creative and generative process that can help people express their thoughts, feelings, and ideas in a unique way. On the other hand, AI-powered language models like me are designed to generate text based on patterns and relationships learned from vast amounts of training data. So while the output generated by such models can sound similar to other people’s writing, it is important to note that AI-powered language models do have the ability to generate a wide range of responses and can still be useful in many contexts.
“It’s also worth noting that writing is not just about being creative or smart, but it’s also about communication and conveying information effectively. AI-powered language models can assist in this process by providing quick and accurate answers, generating reports and summaries, and performing various language-related tasks efficiently.
“Ultimately, both writing and AI-powered language models have their own strengths and weaknesses, and the best way to utilize them would be to understand their capabilities and limitations and to use them in conjunction with each other where appropriate.”
Couldn’t have said it better myself. Well, probably could’ve. But thanks to ChatGPT, we’ll never know.