Since ChatGPT’s release, it has gained global popularity, with around 100 million users using it monthly. Schools around the U.S. have implemented policies regarding using generative artificial intelligence tools for schoolwork.
OpenAI released ChatGPT on Nov. 30, 2022, as a successor to InstructAI, their previous AI model. ChatGPT unlocks new limits of AI technology by interacting with inputs in a conversational form rather than an instructional form.
According to OpenAI, “The dialogue format makes it possible for ChatGPT to answer follow-up questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests.”
In order to make ChatGPT work, OpenAI trained it using Reinforcement Learning from Human Feedback (RLHF), a system where human AI trainers act out both sides of the user and AI interaction. Trainers give prompts and model-written suggestions in order to help them respond. Data is then collected on the interactions, and by using a reward-based system, OpenAI can fine-tune ChatGPT.
According to OpenAI, ChatGPT is not without limitations. OpenAI states, “it sometimes writes plausible-sounding but incorrect or nonsensical answers.” It can mistake or misunderstand a prompt if the wording is slightly off. If a question or prompt is too ambiguous, the AI will guess what the user meant rather than asking clarifying questions.
Due to ChatGPT having the capability to write well-written and structured information, there has been an increase in worry amongst schools across America. In Seattle, schools have banned the use of ChatGPT in school.
WCC faculty and staff have been figuring out how to prepare for the spring quarter. Currently, WWC is leaving the use of ChatGPT up to the teacher’s discretion though these policies can change in the future. Staff encouraged teachers to include a section in their syllabus dedicated to the uses of generative AI.
Computer science instructor Darrell Criss has allowed it in his class as he believes it can be used as a tool. “It’s good for researching stuff, it’s good for getting started on stuff, but again you want to be able to go back and double check.” Criss believes that his students are still responsible for the quality of work they turn in, even if allowed to use AI.
Criss still wants his students to function without it, saying, “I want to make sure the students know how to do this on their own. So for example, when they finish my class, do they have a certain level of understanding so they can get by in the next class and the next class?”
While faculty and staff make decisions regarding ChatGPT, the Writing Center has been preparing for students’ potential using ChatGPT or other chatbots.
“We haven’t been screening for it, but we’ve been playing around with it. We had a meeting where we dedicated a whole hour to playing around with ChatGPT and similar services, ” said Khaenin Rutherford, a Writing Center staff member.
During this meeting, the staff explored ways to see if ChatGPT has been used by running the AI-generated response through GPTZero, an AI detector. However, it is not a fool-proof system.
“We found out that changing a few things or the phrasing you ask ChatGPT will have a drastic difference in the level of plagiarism GPTZero censors,” Rutherford said. “By just changing the phrasing, I’ve been able to get it to not recognize machine writing at all.”
The Writing Center has not seen students use it so far, but they have not been testing papers with GPTZero because of its inaccuracy. However, OpenAI has developed its own AI classifier for human-written and AI-written texts to find better ways to distinguish the two.
When it comes to the uses of ChatGPT as a tool for students, Rutherford mentioned, “I see it as a tool that can easily be used to create a problem. It has good applications, I’ve seen people use it to generate ideas and then use it as a springboard into some really cool writing.” he continued to say “it can be leaned on too heavily too in a creative way, I feel like if you use it too much you could lose some of your own creative ideas.”
When asking the AI itself about the negatives of using ChatGPT for schoolwork, it stated, “While ChatGPT can be a useful tool for students, it is important to use it in moderation and with caution. Students should be encouraged to develop their own critical thinking, research, and writing skills and use ChatGPT as a supplement rather than a substitute for their own efforts.”
These negatives can affect a student when trying to find work after college as they might not have been able to develop the critical thinking and social skills needed. For instance, Criss mentioned that when students graduate and go in for a technical interview, they may be asked to show their understanding of a certain topic and will not have ChatGPT to help them.
Even with these potential risks, many students still use ChatGPT as a helper for school work. One junior at Western Washington University is a fan of the generative AI model: “ChatGPT has been a great help when it comes to formulating ideas for an essay or a project I am working on. I have even used it to write cover letters for me or write out a text.”
ChatGPT and the growing market of generative AI is a factors that can change the way students learn. However, it is not without its costs. As AI continues to develop and schools develop new ways to adapt to its uses, we will see it continue to rise.
Companies such as Google and Alibaba have begun developing their own versions of generative AI to compete with ChatGPT. With the progression of ChatGPT, we will see more companies begin to develop their systems as the world moves further into the age of artificial intelligence.
