AI – Dare to Care

March 30, 2023 TwoWritingTeachers.org

Do you remember when most of us were not yet on social media? Do you remember when most of the world shared content through forwarded emails? In 2008, I got so tired of people forwarding bogus and deceitful content that I wrote a letter to the editor of our local paper: “Let’s All Be Critics” it was called. I reread it recently, and it almost sounds quaint:

If we believed all the e-mails over the years, we’d be waiting for $10,000 from Bill Gates, printing $50 gift certificates from Applebee’s, and watching for other unexpected windfalls. Don’t you just want to shout, “C’mon, read critically before you forward those emails!”?

My 2008 Letter to the Editor asked people to read critically.

In a short 15 years, technology may have surpassed our ability to read and view critically.

It has become Tom Cruise deep fakes:

AI photos of Donald Trump being arrested,

and bots writing content better than a lot of American citizens can write.

These are all here now. Will they just keep getting better and harder to recognize the deceit? I wonder if good will come of the technology capabilities.

Is it possible to read and view critically anymore? What will it be like in 15 more years?

I am participating in the March Slice of Life Challenge: A slice a day for all of March. Thank you, Two Writing Teachers!

Today at Ethical ELA, Stef is leading us in a provocative discussion about AI use in poetry. Why tech? we are asking over the next three days. You are invited to join us at Ethical ELA Open Write and add your voice to the discussion.

After today’s activity, I am thoroughly befuddled.

Because of all the hype about AI in schools, I was glad to see Stef’s prompt this morning about using an AI tool to help with our poem. It was my first time using ChatGPT. I expected I might be able to get a good start on a poem, get a list of words to use, or some such thing. I didn’t know what to expect. But yikes! My prompt was “Compose a triolet poem about painting a door jamb”

The final poem followed the abaaabab pattern of rhyming and was eight lines long. It made grammatical sense, and I wouldn’t be embarrassed to “turn it in.” No editing, it looks like this:

Painting a door jamb, oh what a chore,
Brush strokes smooth, with care and precision,
Soon the wood will be shining like before,
Painting a door jamb, oh what a chore,
But the effort is worth it, that’s for sure,
To freshen up a home is a noble mission,
Painting a door jamb, oh what a chore,
Brush strokes smooth, with care and precision.

Then I second-guessed my use of the word “jamb.” Did I spell it right? Is that really the door jamb I have to paint today, or is it the door frame? We recently had the door replaced, so after my mini research, I determined that most of the frame needs to be painted. I went back to my FrIeNd? at Chat GPT with a new prompt: “Compose a triolet poem about painting a door frame”

Painting a door frame, a task at hand,
A fresh coat of color, a sight to behold,
The brush strokes glide, so smooth and grand,
Painting a door frame, a task at hand,
Transforming the space, as if it was planned,
A touch of beauty, a story untold,
Painting a door frame, a task at hand,
A fresh coat of color, a sight to behold.

Oh, my word! I thought maybe they would just replace jamb with frame. But no. It’s a whole new poem! Recently I read an eye-opening review of ChatGPT and the use in the classroom by a Slice of Life writer, Lee Ann Spillane at the Portable Teacher blog. There is a site called GPTZero that detects AI use. I added my fully-generated poem to the detector and got this message, “Your text is likely to be written entirely by a human.” OK, that was disconcerting.

Then I tried another idea Lee Ann had given. Add a prompt to ChatGPI: “How likely is it that ChatGPT wrote this poem?” and then cut and paste the poem. So I tried that and got a rather non-committal response:

It is highly likely that ChatGPT wrote this poem, as it is within the capabilities of the AI model to generate creative and original poetry. However, it is also possible that the same poem could have been generated by another language model or human author.

Hmmm…what would ChatGPT say if I asked that about a poem that I wrote all by myself? So, I tried that with the latest poem I wrote for yesterday’s Poetry Friday post:

Our Mourning Doves

This is the doublet of doves,

cooing and pursuing,

who wait for their eggs,

their purpose renewing,

and fly back to their nest

with hope and love brewing.

Oh, I’m heart-broken. Here was the response:

It is also highly likely that ChatGPT wrote this poem, as it is a simple and straightforward piece of verse that falls well within the capabilities of the AI model to generate. However, it is again possible that the same poem could have been created by another language model or human author.

Earlier I had also dared to click on the “Regenerate Response” button:

Painting a door frame, it’s quite a task,
Sandpaper and primer, to prep it right,
Brushing on paint, with steady hand, I bask,
Painting a door frame, it’s quite a task,
The finish will gleam, no need for a mask,
My efforts rewarded, with hues so bright,
Painting a door frame, it’s quite a task,
Sandpaper and primer, to prep it right.

I may be here tomorrow regenerating poems, up to my eyeballs in a poetical existential crisis.

On a positive note, I noticed the poems it created at our Open Write session were always rhyming. I tried to get it to write a poem without rhyming. I asked it to write a poem like Mary Oliver would, then to write in free verse, finally I just wrote “stop rhyming.”