ChatGPT is a form of generative artificial intelligence. Educators are intrigued by its potential classroom applications.
Artificial intelligence is not your father’s pocket calculator, but educators are comparing a new wave of machine-learning to the leaps in efficiency offered by that now-familiar instrument.
Like a calculator allows for accurate and efficient processing of mathematics, so too might new forms of artificial intelligence help students process writing, research and ideas, said Jill Sible, associate vice provost for undergraduate education at Virginia Tech.
“Just like the calculator years ago, that was forbidden in any sort of quantitative test,” Sible said. “Now, it’s a pretty universal tool that’s allowed in many classes, and on most exams.”
The humble calculator might be taken for granted as a smartphone application today, but its introduction to classrooms in the 1970s caused quite some upheaval.
With artificial intelligence finding its way into classrooms now, Sible said educators at Virginia Tech are as curious as they are concerned, if not outright excited about the possibilities.
People are also reading…
“There’s definitely a buzz on our campus,” Sible said. “Lots of curiosity. Lots of questions among our faculty and our administrators.”
In particular, an online bot called ChatGPT, backed by billions of dollars from tech giant Microsoft, has in recent weeks brought public attention to new possibilities of artificial intelligence.
Made by a startup called OpenAI, the bot is capable of answering for itself. Just ask it a question, or type it a prompt.
ChatGPT said it is trained on “a massive amount of text data,” and “can handle a wide range of topics and tasks.”
“ChatGPT is a type of artificial intelligence model that can generate human-like text in response to prompts, allowing it to chat with people, answer questions, and perform other language tasks,” ChatGPT said. “It’s good for tasks that require understanding and generating text in a conversational manner.”
It’s an example of generative AI, part of a new wave of machine-learning technology that can be trained to use troves of digital data to conjure new sentences, sounds, images, sitcom plots, or in theory just about anything else, said Sylvester Johnson, founding director of the Virginia Tech Center for Humanities.
“That generative part is important, because what it means is that the software can generate an image, or a composition like a sound or music, or some act of language,” Johnson said. “It’s not repeating what it has been given, it’s taking what it has been trained on and generating its own content.”
Johnson, whose work as a humanities scholar includes studying technology, said generative AI has potential to impact many aspects of life, including learning.
“Students went home for winter break back in early December, and that’s about when this ChatGPT got released,” Johnson said. “It’s just a stark difference from a month and a half ago to now.”
He said that difference is because learning institutions are grappling with how to be sure students’ human intelligence—and not some artificial technology’s—are the brains completing writing assignments.
“The biggest challenge that we’re seeing in educational institutions is of course around writing assignments,” Johnson said. “And that’s not just writing essays. That’s also writing code.”
Johnson said OpenAI has at least tried to establish guardrails and encourage responsible use of its technology. Earlier this week, the startup released an AI detection tool to help educators catch bot-assisted plagiarists.
“ChatGPT is game-changing in a way that other similar chat software has not been, because it’s able to maintain a conversational thread… it can remember what it just talked about,” Johnson said. “So you can refer back to something in the conversation and it will go deeper… you can build out a set of responses to very complex things, by just asking one piece at a time.”
Looking ahead, Johnson said he expects universities to amend honor codes to account for advances in artificial intelligence.
But Sible, who works with the academic integrity office at Virginia Tech, said it is too soon to tell what kind of policies might need changing, or whether changes are needed at all.
“Obviously, it presents a new challenge with respect to academic integrity,” Sible said. “Although in some respects, not terribly new.”
A small subset of students try to cheat, and that’s always been true, she said.
“So in some ways, that hasn’t changed,” Sible said. “There’s just a new piece of technology in the mix.”
She said ChatGPT has been cause for excitement among teachers considering how to incorporate artificial intelligence into their curriculum. But of course there are challenges and constraints to consider.
“Our honor code absolutely already highlights that students’ work should be their own, and that they need to follow the rules set by that assignment and that professor,” Sible said. “This new technology really fits within the umbrella of policies that we already have.”
Cheaters beware, in the case of ChatGPT, the technology is still a work in progress. Its listed limitations include occasionally generating incorrect information, harmful instructions, biased content, and “limited knowledge of world and events after 2021.”
And the bot doesn’t cite sources in its writing, Johnson said.
“One of the challenges is that the generative AI, of which ChatGPT is an example, is so good,” Johnson said. “It’s not infallible, it can make errors that are really obvious to human beings… but most of it is actually really good.”
Outside the classroom, a bevy of ethical considerations await society as artificial intelligence continues to improve and become more commonplace. Johnson said adopting a public interest approach to artificial intelligence is a good way forward.
“How do we get it to work in a way that gives us a better society?” Johnson said. “What’s going to be good not only for growing our private capital, but also for strengthening our public institutions and benefitting the public good?”
These issues are very much on the minds of people working in government, private industry, academia and beyond, far more so today than even in 2015, Johnson said.
“I hope that we get less inequality, because this could really expand the precarity of people’s lives,” Johnson said. “A lot of people who thought they were going to be OK in our economic environment could suddenly find themselves unable to participate gainfully in the labor economy.”
Unlike the uproar in classrooms caused by early calculators of the 1970s, many educators today are more interested in artificial intelligence as a tool to improve learning outcomes, Johnson said.
“If you’re an engineering student, you’re taking very advanced forms of math. You’re not being tested on whether you know the answer to 7 + 12,” Johnson said. “You need to use the calculator in order to do something that’s more sophisticated.”
The bot brain behind ChatGPT aligns with that idea. The bot said artificial intelligence, “should not be used in a way that undermines the educational process or takes away from the valuable experiences and interactions that students need to thrive.”
“I hope that students use AI as a tool to enhance their learning and deepen their understanding of various subjects, as well as to facilitate the discovery of new information and knowledge,” ChatGPT said. “Additionally, I hope that AI can assist students in developing their critical thinking, problem-solving, and creativity skills, by providing them with instant access to information and new perspectives.”
ChatGPT has been blowing up on the internet. Now, the Artificial Intelligence tool is turning out to be a lot smarter than people thought.
Want to see more like this?
Get our local education coverage delivered directly to your inbox.
