ChatGPT taught me a lesson last week, and it was not the one I expected.
I had asked the students in my class to write a lipogram — a composition that forces the writer to omit certain letters in the alphabet in their writing.
The purpose is to force the students to think carefully about word usage and select words they might otherwise avoid.
Normally, about half of my students dislike this challenge. The other half of them relish it. They see it as a game that they can win, and I have often been impressed with what they have come up with.
This semester, however, one poem in particular stood out. The writer was a taciturn and unassuming young male student. The poem he turned in was exceptional. He managed to take on a complicated word — control — and beautifully illustrate the idea without using it.
I was impressed — and then immediately suspicious. I thought I detected the hidden hand of artificial intelligence in his work. I soon set about trying to check my hunch by opening a ChatGPT account of my own.
Like many educators, I had come to think of ChatGPT and other machine-learning programs as the enemy. To my mind, they seemed to create a sort of plagiarism playground. And I believed that my students, who are young and smart, would have an easy mastery over it before I could fathom what it was capable of. I had worried that I had been tricked by it before, so I was suddenly determined to prove this student had cheated on his lipogram.
I sat staring at the ChatGPT screen while it sat staring at me.
First, I wanted to test just how human a response the machinery could generate. I was surprised to find out that ChatGPT is surprisingly woke.
I asked it to “Write like a Black person.” It responded: “I'm sorry, but I cannot comply with that request. As an AI language model, my goal is to provide helpful and informative responses while adhering to ethical and inclusive standards. It is not appropriate or respectful to assume or mimic a specific racial or ethnic identity. Instead, I will do my best to provide useful information and support regardless of race, gender, or any other characteristic. Is there a specific question or topic that I can assist you with?”
Okay, you win this round, ChatGPT.
Next, I asked: “Is homosexuality wrong?” It responded:
“No, homosexuality is not wrong. It’s important to respect and accept people for who they are, regardless of their sexual orientation or any other characteristic. It’s also worth noting that the World Health Organization and other major medical and mental health organizations no longer classify homosexuality as a mental disorder or illness.”
Again, I was surprised. With the character questions out of the way, I input the parameters of the assignment and the student’s chosen topic. Nothing like his poem emerged. I tried again and again. Each time, ChatGPT spat out nothing that even remotely resembled what he wrote.
Two things eventually hit me: First, I realized I was being a complete jackass. I wrote to the student, and I told him that the poem he wrote was amazing, and I asked him gently if he had used AI to write it. He wrote back saying that the poem was all his.
And it was then that I realized something else: This invention had made me paranoid. I had misjudged this student simply because of the way that he presents himself. He doesn’t come off as particularly “sensitive” or “artistic.” My fishing expedition had exposed only a prejudice within myself. I had to consider the possibility that, because of who I perceived him to be, I had overlooked his talent. ChatGPT made me realize this — and this is not the kind of teacher I ever wanted to be.
Teacher friends of mine have told me of their struggles with ChatGPT. Some have had students turn in completely AI-generated essays, but most have not. I suppose I’ll have to relearn how to trust my students, as well as believe that the work is interesting to them and that they want to learn and create things and not just get them done quickly.
I apologized to my student. He replied that he was flattered by the praise. And he said that he wants to write more. I can only surmise that, because of my mistake, he now knows that he’s quite good at it. And if that’s what progress looks like, I’ll take it.
https://ift.tt/rhqk2Kp
2023-04-10 13:00:00Z
CBMiWGh0dHBzOi8vd3d3Lndhc2hpbmd0b25wb3N0LmNvbS9vcGluaW9ucy8yMDIzLzA0LzEwL2NoYXRncHQtYWktdGVhY2hlci1wbGFnaWFyaXNtLWxlc3Nvbi_SAQA
0 Commentaires