From ELIZA to today's AI writing tools: What have we done?

[sue_frantz]

Many of us who have been around psychology for a while remember ELIZA (or at least remember learning about ELIZA), arguably the first chatbot. Jumping ahead a few decades, with paraphrasing tools users copy/paste text into the tool, click a button, and the words—after being hammered by a thesaurus—are spun into new text. The tools have been marketed to freelance writers. The freelance writer can write one article, copy/paste it into the tool, click the button to get their next text, then tweak the new text to make sure it sounds good. They now have a new-enough article they can sell to someone else without it being picked up by plagiarism detectors as duplicated content. Some of our students have been using these same paraphrasing tools, although sometimes without the proofreading and tweaking. The results can be, frankly, hilarious. My favorite is a history essay that kept referring to President Shrub.

Computing technology, however, continues to march on. And now we have AI tools that will write a pretty good essay based on a sentence fragment. We even get fake references. Plagiarism checker tools will not flag the content because the sentences have been newly generated. Just not by a human.

The newest tool is called ChatGPT. Anyone can try it out, although at the time of this writing, the website says, “We’re experiencing exceptionally high demand. Please hang tight as we work on scaling our systems.” While we wait for them to have enough servers to handle the demand (from our students?), read an essay that Mike Sharples had the ChatGPT tool create on learning styles (Sharples, 2022). Sharples gave the AI this prompt: “The construct of ‘learning styles’ is problematic because.” Here’s a paragraph from that AI-generated essay—and, yes, the Dunn and Dunn reference is fake, although it is based partially on fact.

In a review of this research, Dunn and Dunn (1997) found that the evidence does not support the idea that people have a particular learning mode that is dominant or that they process information in a particular way. In their study, the authors found that when students were given a particular instructional treatment, they could be classified in a particular learning modality category. But when they were given a different instructional treatment, they were classified in a different category. In a follow-up study in which they gave students a number of instructional treatments, Dunn and Dunn found that students could be placed in a particular category in some treatments, but not others (Sharples, 2022).

As I pasted in that paragraph from a four-paragraph essay, I wondered if I would be violating copyright law for the amount that I was using. And then I thought, “Wait! No one actually wrote this!” Surely if a monkey does not own the copyright to their selfies (Nauro, et al. V. David John Slater, et al., 2016), then a computer program does not own the copyright to their generated text.

To make this even creepier, Sharples writes, “’Just as a student can generate an essay in seconds, so a teacher can assess it. Add ‘Here is a short assessment of this student essay:’ to the end of the text and GPT-3 writes a review.” Now that I’ve concluded that a computer cannot hold the copyright, I have no qualms sharing the entire AI-generated review with you.

The essay discusses the idea of learning styles, and how they may be developed over time. It cites research by Dunn and Dunn which suggests that learning styles are not fixed, and that people can learn in different ways depending on the situation.

The essay is well-organized and flows smoothly. It makes a clear argument and supports it with evidence from research.

One potential criticism is that the essay does not discuss the implications of learning styles research, or how this research can be used to improve learning. However, this is not necessarily a flaw, as the essay is only meant to be a brief overview of the topic (Sharples, 2022).

Sharples suggests that instructors use AI writing tools as a teaching tool rather than struggling to find ways to detect or counter their use. I’m less optimistic.

Stephen Marche, writing in The Atlantic, also has concerns. “The essay, in particular the undergraduate essay, has been the center of humanistic pedagogy for generations. It is the way we teach children how to research, think, and write. That entire tradition is about to be disrupted from the ground up… Neither the engineers building the linguistic tech nor the educators who will encounter the resulting language are prepared for the fallout” (Marche, 2022).

I have lots of questions and no answers.

• What have we done to create a generation of students who are more interested in completing an assignment for a course grade than learning?
• What happens when college degree recipients hit the workforce and are unable to write? Will the AI tools cover for them there, too?
• Does writing make us better thinkers? What happens if we stop writing?
• Is this the beginning of the end of online education? Are we going to turn back the clock and return to in-class, hand-written assessments? If so, will cursive writing make a comeback because it’s a faster way to hand-write? And if we all return to the classroom, what are the implications for who has and who does not have access to higher education?
• Or will we require students to do a Cloze test on all of their assignments to prove that they did indeed write them themselves—or at least were able to memorize their AI-generated essay well enough to convince their instructors that they wrote it?
• Or will we do what Sharples suggest, such as use AI to generate essays, and then ask students to critique the essays and then edit them so they are better (Sharples, 2022)?

Lots and lots of questions. No answers.

References

Marche, S. (2022, December 6). The college essay is dead. The Atlantic. https://www.theatlantic.com/technology/archive/2022/12/chatgpt-ai-writing-college-student-essays/672...

Nauro, et al. V. David John Slater, et al., No. 24, 28 (U.S. District Court, Northern District of California January 28, 2016). http://s3.amazonaws.com/cdn.orrick.com/files/naruto-v-slater-motion-to-dismiss-feb-2016.pdf

Sharples, M. (2022, May 17). New AI tools that can write student essays require educators to rethink teaching and assessment. Impact of Social Sciences. https://blogs.lse.ac.uk/impactofsocialsciences/2022/05/17/new-ai-tools-that-can-write-student-essays...