By Emma Reilly
Editor-in-Chief
What if a computer told you that the character it most relates to in Mary Shelley’s “Frankenstein” is the creature?
Professor of English, Director of Writing, and Co-Director of the Barbara and George Cromwell Center for Teaching and Learning Dr. Sean Meehan experienced just that while experimenting with the artificial intelligence chatbot ChatGPT.
Is the AI referencing the fact that it was made by humans? Is it alluding to a desire to seek revenge? Is it calling us, the Dr. Frankensteins of this world, monstrous?
These questions, though unanswerable, are nothing if not intriguing.
Launched in Nov. 2022 by the research laboratory OpenAI, ChatGPT is raising ethical questions in pedagogical circles and beyond.
John W. Allender Associate Professor of Ethical Data Science Dr. Kyle Wilson spoke to students and faculty about how ChatGPT came about, what it is, and what it can be used for on Jan. 25. The event was organized by the Cromwell CTL.
ChatGPT is circulating in the media for a number of reasons. Namely, it is highly innovative and reflects rapid growth in the realm of AI development.
“AI development is speeding up,” Dr. Wilson said. “When suddenly stuff worked…a flood of money came into the field and a flood of interested people. The number of people that will sit around and work on a dumb idea that hasn’t worked for 40 years but might one day is much smaller than the number of people that want to get in on the next big thing.”
According to Dr. Wilson, a state-of-the-art statistical model laid the foundation for ChatGPT’s development. Though it is based on the same basic framework as services like text prediction or Grammarly, ChatGPT’s complexity allows it to interpret language conversationally.
ChatGPT was trained using 45 terabytes of text from the internet, text which included Reddit feeds and fanfiction, according to Dr. Wilson.
As Dr. Wilson explained at the talk, ChatGPT can be used to create headlines in the style of The Onion, to explain highly complex mathematical concepts while mimicking Snoop Dogg, or even to communicate with a deceased loved one.
“It’s appropriate to step back and feel some wonder about these things, whether they’re going to wreck society or improve it,” Dr. Wilson said. “The response of wonder is very real and fitting, I think.”
Yet, the implications of this technology, especially for students and educators, are difficult to ignore.
In recent weeks, The Atlantic deemed ChatGPT “the end of high-school English,” while the New York Times investigated the ways in which college professors are adapting to AI’s prevalence.
Questions of how to regulate AI use in academic settings and how to adapt teaching and learning processes to a world in which AI is widely accessible are gathering interest here at Washington College.
Senior Jack Goembel shared an anecdote about his brother, who is an educator.
“He thought something might be up because the student didn’t really know much about their essays when questioned. But otherwise, he couldn’t tell and wasn’t sure if the student was using AI,” Goembel said. “If the student hadn’t admitted to using ChatGPT, he could have gotten away with it.”
Chair of Anthropology and Archaeology, Associate Professor of Anthropology, and Co-Director of the Cromwell CTL Dr. Emily Steinmetz was surprised to discover that her middle-school-age son was experimenting with ChatGPT, though not for class assignments.
Senior Emilee Cramer is concerned with AI’s potential to impact not just classrooms, but labs and field work.
“You can look up a scientific question and get back a reasonable answer, but with no access to the source of that answer, and there’s already so much misinformation in the scientific field,” Cramer said.
The next Cromwell CTL machine learning talk will take place in the Faculty Lounge in Hodson Hall from 11:30 a.m. to 12:30 p.m. on March 2.
The students in Dr. Wilson’s machine learning course will be hosting a teaching series this semester, with further details to come.
Photo by Olivia Long
Photo Caption: Dr. Kyle Wilson delivered the first talk in a machine learning series last week.