Ana Maria Cornea / Gavel Media

ChatGPT: Blessing or Curse?

Artificial intelligence, while not the world-ending, hyperintelligent force depicted in popular media such as Space Odyssey or Terminator, has progressed a great deal since its theoretical inception with Alan Turing’s models of computation in the early 20th century. One of the more popular current applications of AI is in generating original writing, with ChatGPT being one of the best and most popular models for this purpose. To explain further, ChatGPT is a conversational language model developed by OpenAI. It was trained on a large corpus of text data from the internet and can generate text responses in a variety of styles and topics. It can be used for applications such as chatbots, customer service, and language translation, among others. To realize how well this model emulates human writing, one only needs to look at the preceding three sentences, which were written by ChatGPT when prompted with "What is ChatGPT?"

As one can imagine, many innovative students both at Boston College and around the world have turned to this software for all manner of academic writing. Though raw content from ChatGPT is fairly bland and formulaic, papers written entirely by ChatGPT have earned B+ grades in a double-blind study using professors from the University of Pennsylvania’s highly prestigious Wharton School of Business. 

On February 2, the Center for Teaching Excellence (CTE) hosted a panel discussion where faculty members proposed strategies for dealing with and integrating ChatGPT into academia. One conclusion reached was that the experience of learning must be "recentered" on primary research areas such as current events, presentations, and oral exams, areas where ChatGPT either does not have up-to-date knowledge or cannot help at all. Essays in the future of literary academia may be less focused on reasonably supporting theses, something that ChatGPT is skilled at, and more about discerning the intention and emotion behind fictional or philosophical works. The CTE has prepared a tentative list of grading policies and strategies for integrating AI into the classroom setting. Boston College currently has no official policy regarding AI content in academic work, but using ChatGPT without citation would be considered plagiarism. Educators can combat this by using for-profit services such as Originality.AI that claim to be able to detect chatbot usage with 94% accuracy.

ChatGPT does have its limitations, including not being able to reference events or works of media more recent than September 2021, not generating responses to prompts deemed offensive or dangerous, and writing responses that align more with what the model believes a person would say than what is factual. For example, when prompted with, "An old man talks to his father. Who is the older man?" ChatGPT responds with "The old man." It cannot properly cite sources, and there may be copyright violations when prompts involve pre-existing intellectual properties ("Write an explanation of cellular respiration in the style of George R.R. Martin’s ‘A Game of Thrones’"). Additionally, ChatGPT is currently a demo open to the public for research purposes, but OpenAI is already rolling out a paid subscription under the name "ChatGPT Plus", and initial valuations of the tool are coming in at a modest $29 billion. Many students are already financially stretched enough without paying for AI software that might be able to be detected in writing.

A panel speaker compared the advent of AI to Socrates' words on the popularity of books and writing: "In fact, it will introduce forgetfulness into the soul of those who learn it: they will not practice using their memory because they will put their trust in writing." Anytime new modes of learning are introduced, there are always drawbacks and trade-offs. It remains to be seen if AI will prove to be the end of education as we know it, but it will doubtless have an impact.

ChatGPT can be accessed for free at chat.openai.com/chat

Comments