Faculty optimistic about AI writing tools
March 7, 2023
There’s a lot of discussion among educators, marketers, journalists and writers on how their worlds will be impacted by AI content creation tools. ChatGPT is one such tool that’s been growing in popularity. Users enter a prompt and, much like how AI art programs like Midjourney and Stable Diffusion produce a relevant image, ChatGPT provides text. This text can range from a sentence or two to a work spanning the length of a novel.
Social media is awash with various accounts of students using ChatGPT to cheat on their coursework. From students writing short answers to entire essays, educators are concerned about how they’ll be able to cope with the increasingly popular technology.
“I was very worried, until I tried it,” said Andrew Tolle, an Eastfield English instructor. “There is no way this would ever be able to replace human work.”
Tolle has students write analyses on various works, including poems, plays and novels. As a demonstration to the class, he issued a poem as a prompt to ChatGPT and asked it to write an analysis.
“They could see it has some interesting thoughts. If that had been submitted, it wouldn’t have even got a 50,” Tolle said. I was very worried, until I tried it.” — Andrew Tolle, English instructor
I was very worried, until I tried it.”
— Andrew Tolle, English instructor
The reason ChatGPT would not get a passing grade is because it is prone to slipups. Tolle noted how the program misidentified a character in a short story as a third-person protagonist multiple times. While it may struggle with long-form analysis and essay writing, the Chatbot might be effective at small-scale assignments. This raises the question of whether students will use these tools as a shortcut, reducing their student engagement with coursework.
“When you decide you’re no longer going to think critically to write a paper and to synthesize ideas and to analyze ideas, your ability to do it will become lost,” said Tuesday Hambric, a psychology and learning framework instructor at Eastfield.
Hambric argued that this technology in its current form could hinder student learning because they aren’t arriving at the answer through problem solving.
It could come down to how the instructor adapts the technology and engages students.
“If the instructor is teaching them how to use it and helps them understand the implications of using it in a negative way, I think it could help them,” Hambric said. “Technology is just going to happen. It’s already here. So it’s one of those things where instead of screaming bloody murder, we’ve got to ask ourselves, ‘how can we embrace this technology to better help students?’”
Tolle said a benefit to AI writing tools is helping students understand difficult concepts.
“ChatGPT provides a level of access to certain conventions of writing that may not be accessible to folks immediately,” Tolle said.
The writing styles may be academic or professional. It may also aid students who speak English as a second language and struggle to write it.
“I see artificial intelligence as a great equalizer,” marketing instructor Wade Hyde said. “Those folks who might’ve had at-risk students, English was not their primary language, the enormous jump from not only having to learn English but then learning the nuances of writing skills, it takes time. AI will help accelerate that process,”
Tolle said AI can also be used to dismiss certain ideas for argumentative essays. For example, he said ChatGPT refused to make an argument about how vaccines cause autism because it couldn’t find reliable evidence.
Voice is the way an author’s personality comes out through their writing. AI technology could also be a way for students to find their own.
When implementing a prompt into an AI chatbot, it has the potential to give a starting point to students in forming and organizing their thoughts, Tolle argues. How can we embrace this technology to better help students?” — Tuesday Hambric, psychology instructor
How can we embrace this technology to better help students?”
— Tuesday Hambric, psychology instructor
“If they see it can be said that way, it can give them more confidence,” said Tolle.
When it comes to identifying whether a student has used an AI, Tolle examines the voice. He says that the more general the analysis is the more he’d question whether this was written with AI tools. This could lead to students not finding their voice if they’re using it for assignments.
When it comes to combating the technology, Tolle argues it might be more beneficial to educate oneself on how it works.
“I want to get ahead of students thinking they can just use it, copy and paste and turn it in because I know they’re gonna fail, even if I don’t catch it,” said Tolle.
He says that in better understanding the technology will allow him to help his students understand how to use it to improve their own writing.
“We have a job to do. Regardless of what is going on outside of the classroom, we still have to meet our student learning outcomes,” Tolle said. “If some development occurs that changes how we can ensure they meet those student learning outcomes then we need to adapt to it,” Tolle said.
There is a possibility that many content writing jobs, like for press releases, blog posts and social media posts, may be automated via AI technology.
“Society had the same fears about computers, and the same fears about automobiles and airplanes,” Hyde said. “There’s potential for some outplacement of jobs. Likely the people who were displaced by that will be outsourced or find other opportunities and maintain the devices that replaced them,” Hyde said.
Hyde argues this technology could enhance these writing jobs rather than automate them outright. He compared it to similar writing tools like Grammarly. Rather than being a replacement, it would merely correct human mistakes.
“I’m astonished to this day that newspapers like the New York Times still have grammar errors. How is that possible?” Hyde said. “It’s not very often that they do, but it’s human error. I think by leveraging technology you end up having a better-quality product.”
Hyde said there might be some writing jobs that could be automated outright, such as obituary writing.
After the recent shooting at Michigan State University, Vanderbilt University (capitalized) issued a statement generated by ChatGPT calling for community support. It was seen as tasteless by many and contained inaccuracies citing multiple shootings when only one occurred. Peabody Dean Camilla Benbow issued an apology.
Recently CNET, a website dedicated to tech reviews, news, articles and blogs, had been quietly using AI technology to write its articles for months. When discovered, the site Futurism was able to determine whether certain information was inaccurate.
“The drawback isn’t the technology, although the technology can always be improved. The drawback is that CNET was not editing their own articles” Hyde said. “You still have to proof your work, so somebody wasn’t doing their job. It’s a misuse of the tool, not the tool itself,”
Rather than putting the onus on the technological flaws, humans may be able to work in tandem with the strong points of the tech to put out a better product. Society had the same fears about computers.” — marketing instructor Wade Hyde
Society had the same fears about computers.”
— marketing instructor Wade Hyde
“If you have someone else ensuring what is produced is spoken about correctly, factually correct, follows the style that you as an institution have adopted – those are the things AI hasn’t caught up with yet,” Tolle said.
Hyde said a potential benefit is that the technology will allow writers to focus more on being creative and get rid of the tedious parts of writing.
“Using this technology does not decrease humans’ ability to be creative or humans’ want or need to write something,” Hyde said, “It’s more likely going to enhance their ability to do their jobs. Where it may replace some rudimentary tasks, the person will likely be doing something more advanced.”
Journalists have also expressed concerns about being replaced, especially in light of the recent CNET scandal. Hyde argues that journalism is far too sophisticated a job to be supplemented by an AI given the amount of deep research involved.
CNET did have some errors in the AI-generated articles like basic math being wrong. Given the quantity of work AI can produce it would likely be beneficial to hire staff to do the proper editing and research that needs to be done. Hyde says writers, editors and researchers well versed in using AI-based tech are more likely to be hired, thus would be wise to learn the technology.
“What it will probably do is just take what the person wrote and generate a lot more news releases and they can spend some time editing the material generated and be able to do other tasks AI cannot do,” Hyde said. “For example, AI cannot pick up a phone and call you to pitch a story.”
Hyde also argued that the onus is on the humans using this technology to make sure their information is factual and correct since, like Tolle pointed out, it is subject to error.
“Buzzfeed is not going to be able to say this is just going to be created by an AI and go straight to the website without anybody looking at it first,” Tolle said.