Breda O’Brien: ChatGPT is the ultimate plagiarism tool. How can teachers respond?

Nick Cave thinks it produces songs that are a “grotesque mockery of what it is to be human”. The Atlantic magazine thinks it signals the death of high-school English. New York high schools have banned it from their school networks. Irish universities such as Trinity College are debating its implications. The National Academic Integrity Network is conducting seminars about it.

What is generating all this fuss?

ChatGPT is a large language model trained by OpenAI. It is capable of generating human-like text based on a given prompt or context. It can be used for a variety of natural language processing tasks such as conversation generation, language translation, and text summarisation.

As you may have guessed, the previous paragraph was written by ChatGPT. The prompt was very basic: what is ChatGPT? The formal, staid language may make you wonder what the big deal is. I then asked it to write a definition that includes ChatGPT’s implications for second-level English classes.

Why does no one give up their seat on public transport any more?

---

Did no alarm bell tinkle from RTÉ's editorial structures about its programme on abortion services?

---

Critics are right to worry that hate speech laws are open to abuse

---

Leaving Cert reform plans may result in irreparable damage to exams’ reputation

---

“ChatGPT is a game-changing technology that can make writing essays and analyzing literature a breeze! Developed by OpenAI, it can understand and respond to prompts in a way that mimics human interaction. This means it can help you generate ideas for your essays, analyze literature, and even help you with grammar and sentence structure. Imagine having a writing partner that can help you with your assignments, 24/7.”

ChatGPT’s enthusiasm for its own potential may not be shared by teachers. Where ChatGPT sees partnership, teachers see plagiarism.

The AI essay now described a dress bought by a grandmother who never got to see her granddaughter wear it and how donning that dress is a bittersweet experience, linking both love and loss

Just for the craic, I entered a personal essay title from a previous Leaving Cert English Higher Level paper into ChatGPT. “Write a personal essay about your clothes, what they mean to you and what they say about you.”

In less than a minute, it produced a perfectly respectable and grammatically correct essay in the classic five-paragraph style beloved of American educators. The spelling and cultural references were American but with a tiny amount of tweaking, it would have received a H4 (60-69 per cent). The essay improved remarkably when prompted to include more descriptive details, personal anecdotes and reflection.

The essay now described a dress bought by a grandmother who never got to see her granddaughter wear it and how donning that dress is a bittersweet experience, linking both love and loss. Another five minutes of tweaking and additional prompts brought the essay up to an H2. I could not get it to an H1, but I wonder whether that reflected my inbuilt prejudice because I knew I was correcting an AI-generated essay? Or is Nick Cave right when he says that the best writing “requires my humanness”?

Teaching tool

Leaving Cert students spend hours crafting a single personal essay. (Some of them attempt to recycle the same personal essay no matter what comes up on the exam paper, but that is another story.) Who could blame them for taking the easy route?

So is it the death of English teaching? Oddly, my immediate response to the first ChatGPT essay attempt was that it would be an excellent teaching tool. Bring it into class, put students in pairs to grade it and then discuss what was good and what could be improved

Students are often too self-conscious to allow their own work to be discussed like this, or feel hesitant about providing critical feedback to their peers. (Maybe that is just girls?) An AI has no feelings to hurt.

Banning or even detecting it may prove impossible. It provides fresh essays for everyone, even with the same prompt. Students can also direct it to contain, say, five common errors in teenage writing

Popular educational blogger Matt Miller of Ditch That Textbook, gives 20 ways ChatGPT can be used positively including, for example, letting the bot take one side of a debate and a student the other.

Lots of educators agree it has positive potential as well as petrifying pitfalls. ChatGPT often contains inaccurate information. It also does things such as making up plausible but bogus references and sources, but it will continue to evolve and improve. Banning or even detecting it may prove impossible. It provides fresh essays for everyone, even with the same prompt. Students can also direct it to contain, say, five common errors in teenage writing in order to conceal AI origins.

Irish second-level students still take pen-and-paper terminal exams, which is some help. However, ChatGPT has serious implications for the proposed review of the Senior Cycle, which will rely much more heavily on non-examination elements such as essays and coursework. The integrity of these alternative forms of assessment will be impossible to guarantee. It would be reckless for any proposed reform to fail to take that into account.

Miller makes another important point: devising assessments that prevent students from exploiting ChatGPT may break many teachers.

Teachers are frayed and exhausted after the pandemic but yet again, educators are on the front lines, facing developments before the rest of society has to wrestle with them. For example, it may render some jobs, such as data entry, customer service and language translation obsolete. (On the other hand, ChatGPT assured Hugh Linehan his column is safe.)

Society as a whole, not just educators, must grapple with the far-reaching effects of a technology that will rival the impact of the printing press, search engine, and smartphone combined.