Collaborating With AI (Iteration Two – Playing A Poem)

Yesterday, I wrote about a poetry experiment I did with Google’s Gemini AI, but I am still interesting in exploring different angles on that project. I took its words and my words into Keynote and separated them out, so that one side is the AI lines and the other, mine. The music I composed seeks to echo some of the mentions in the poem.

I have another idea for the poem that I am not yet sure I can pull off, giving the reader agency over the poem. More on that, perhaps, for another day.

Peace (in experimental poems),
Kevin

Collaborating With AI (Iteration One – Writing A Music Poem)

Ghost in the Machine
Ghost in the Machine flickr photo by Simon_68025 shared into the public domain using Creative Commons Public Domain Dedication (CC0)

Terry Elliott was exploring the Perplexity AI platform as a “writing partner” for poetry and it responded to a question of his about its own limitations in a way that caught my attention.

“Poetry is a deeply human art form that conveys emotions, experiences, and perspectives in ways that are often tied to the poet’s personal history and emotional landscape. AI, by its nature, cannot replicate these aspects of human experience, which limits its ability to create poetry that resonates on a human level.”

I then went into Perplexity myself, and asked it look at the quote it gave Terry and give me a response. Interestingly, it seemed to take a different view.

While it’s true that AI, by its current nature, lacks personal emotions and experiences, it is capable of creating poetry that resonates on a human level. AI can analyze vast amounts of existing poetry to understand patterns, styles, and emotional content, allowing it to generate new poems that capture the essence of human emotion and experience. Additionally, AI can be used as a tool to inspire and assist human poets in their creative process, leading to collaborative and emotionally resonant works. The intersection of AI and poetry presents an opportunity for unique and thought-provoking artistic expression.

I asked Perplexity to give me examples of where humans and AI worked as collaborators on poetry projects, but the links it shared — while interesting — were mostly research papers about the possibilities. The one link that was quite interesting was a project at the MIT Lab museum in Boston, where they set up a live demonstration of visitors writing poems with a version of ChatGPT. (Note to self: next time I am in Boston …)

I have long been exploring the role of Generative AI as writer, and Terry’s recent explorations presented yet another reason to try collaboration. So I went into Google’s Gemini (the old Bard, now upgraded).

Gemini was very enthusiastic about my suggestion for working on a poem together!

That sounds wonderful! I’d love to collaborate with you on a poem about music.

I gave it some instructions, and told it I wanted to collaborate on a poem about music, by riffing off single lines of a poem (like jazz musicians, I told it). It responded:

Sounds like a stimulating jam session! Let’s riff on this melody of words. Here’s the first line:

A whisper stirs on strings, a lonely cry,

Your turn to take the lead and see where this melody carries us. Remember, feel free to break the rules, improvise, and surprise me with your creative flow!

And so it began, the back and forth  …

Playing A Poem (a collaboration with AI)

(Gemini) A whisper stirs on strings,
a lonely cry,
(me) resin on the bow,
she closes her eyes to sing

(Gemini) A melody unfurls,
a tapestry of sound,
(me) Knitted and knotted,
each note, unbound

(Gemini) Whispers crescendo,
a storm in a teacup’s heart,
(me) The listener settles in, closes eyes,
finds the center of the sound

(Gemini) Harmony drapes them,
a cloak of calm
(me) On the horizon, a trombone
plays soft and low

(Gemini) A distant call,
a yearning melody’s glow
(me) And beneath our feet,
a rhythmic beat, sticks on stone

(Gemini) Mirroring the rhythm,
finding their own space,
(me) In grace, the music
makes light

See entire thread of the conversations we had and some revision conversations.

I still don’t know what to make of it all, but I did find the experiment interesting.

Peace (in the machine),
Kevin

Poem: Erratic Muse

Erratic Muse

This poem originated from an inquiry research article by Wendy T and friends, which they presented at a conference. The inquiry had to do with using GenAI for playful explorations. (see research article – it’s worth your time for the way they reflected on how GenAI might be useful and not for making art).

I popped their article into ChatPDF this morning, and chatted with the GenAI about the piece about human agency, making art and GenAI, and ended with a request for phrases and terms that it could find that I could use in a poem on the theme of the article. (you can do the same, with this link.)

I then took those terms (there were about 20 or so) and whittled them into a poem, adding AI artwork from Adobe Firefly.

Peace (and poems),
Kevin

Google’s New AI Tool: MusicFX

I have long been interested (sometimes, alarmed) by how some of the new machine learning/AI tools might impact the making of music. Mostly, I have not been all that impressed (good thing for human music composers, right?) but I also know that the technology is only getting better. (See some earlier posts)

The results have been awful or weird or un-listenable.

Google just released its new tool called MusicFX in its AI Test Kitchen (so, you know, beta) and, well, it’s a big leap forward to what I was playing around with just a few months ago. You write in text about the kind of music you want, and you can add genre, instrumentation, etc, and the site generates a 30 second track. The few experiments I did sounded decent.

Bad news for human music composers? Maybe.

Peace (and sound),
Kevin

PS — there’s also a new TextFX in the same platform but I can’t for the life of me figure out its value.

AI NotebookLM Experiment: Analyzing Terry’s Poems

NotebookLM

I vaguely remember signing up as a beta user for Google’s NotebookLM AI experiment but then forgot about it until an invite came in my email yesterday. It’s still a bit funky in its workings but the general idea is that it uses Google Bard, with up to 10 texts that you can upload or paste into its system for inquiry.

You can then do questions or analysis or creative activities across all the texts (or just some, if you decide to uncheck the texts), engaging in a sort of conversation with the various texts. (Honestly, I thought it was going to have more features and I believe it will, as what is there right now didn’t seem too revolutionary to me).

I tinkered a bit, and then thought: I’ve been enjoying reading the poems of my friend, Terry Elliott, so why not use his three most recent poems. I pulled them in, and began a session of interrogating his poetic texts. I explored theme.

Terry poems Themes

Then, I explored imagery and common phrasing and words.

Terry poems Imagery

I asked the Notebook to tell me what Terry hoped I was hearing in his poems. This was the most interesting part of my experiment, I think. It gathered phrasings and explained what it thought Terry was trying to say. I’m not sure how true it is but it gave me things to ponder, and forced me to look deeper at the poems (which is the whole point of what I was doing here — allowing technology to help me see the poems in a different slant).

Terry poems instructions to reader

I also turned each poem into a haiku (the results are just “meh” in my opinion).

**The following is a haiku translation of the first poem:**

**Future is dawning**
**Make poems to see it**
**Fete vos jeux**

**The following is a haiku translation of the second poem:**

**Alone with the trees**
**Be fluid and free**
**Let the leaf fall**

**The following is a haiku translation of the third poem:**

**Be a lighthouse**
**Sharing what you have found**
**Surprised by your mind**

I asked it to write a play skit, with a poet talking to the reader, and it came out robotic and just weird – no nuance. I asked it turn that play into another one, but with a pirate theme (with hat nod to Tellio of the Olden Days of CLMOOC) of a captain talking to a sailor, and the notebook just completely and utterly ignored Terry’s poems entirely. (The nerve!). I had it write a letter to Tellio, the poet, on the issue of dreaming and the future. It was fine, and very informal in tone.

I asked it to make a Cinquain, using words from all the poems, with TERRY as the letttering. Fail. But intriguing anyway as a small poem.

**TERRY**

**Dawning future**
**Fluid, free, visceral**
**Making sense, redrawing maps**
**Lighthouse**

I asked it to write a short essay, using the text of all three poems, on the topic of hopefulness, a theme it had identified across the poem collection.

**The Tree Poems: A Source of Hope**

The two tree poems that we have read, “Imagine Your Future” and “What It Takes More or Less,” offer a source of hope in the face of uncertainty. The poet’s message is that we can find hope in the natural world, and that by connecting with nature, we can find the strength to face whatever challenges come our way.

In “Imagine Your Future,” the poet writes about the importance of dreaming and creating. She encourages us to “make and make and make” our own futures, and to trust that our efforts will be rewarded. This poem is a reminder that we are not powerless to create our own destiny. We can make a difference in the world, and we can build a better future for ourselves and for our children.

In “What It Takes More or Less,” the poet writes about the importance of being present in the moment. She encourages us to “be alone with the trees” and to “be visceral.” This poem is a reminder that we need to take time to appreciate the beauty of nature and to connect with our surroundings. When we are present in the moment, we are less likely to worry about the future or to dwell on the past. We can simply enjoy the present moment, and find peace and happiness in the natural world.

The tree poems offer a source of hope because they remind us that we are not alone. We are connected to the natural world, and we can find strength and inspiration in nature. When we are faced with challenges, we can find comfort in the knowledge that we are not alone. We can find hope in the beauty of nature, and we can find the strength to face whatever challenges come our way.

I could go on.

It’s imperfect, as Generative AI often still is, but the ability to engage with multiple texts in one place (one notebook) has a lot of potential, perhaps. A significant drawback (for now, in beta) was that it didn’t seem to have save my work from session to session (I am sure that will get fixed). But I also noticed at one point that is stopped using one of Terry’s poems in responses to my questions.

Peace (in explorations),
Kevin

PS — Terry gave me permission to share this AI work on his poems

Using Padlet’s New AI Art Feature

Golden Lines with AI Art

Padlet, which I use a lot with my students for sharing across classes, recently added an AI Art tool — with a tool name that I don’t really like (“I Can’t Draw”). As you might expect, my sixth graders are quite fascinated by it but I want them to use it to complement their writing and sharing, and not just turn them loose on it.

So, as part of a larger short story unit, I introduced the idea of Golden Lines — a sentence or two or phrasing from their stories that have some importance to the narrative, and then suggested they use the AI Art tool to make art based on their Golden Lines.

They are coming out pretty interesting, I’d say, and a little later in the year, we’ll do some explorations to deconstruct how the tool works and more. For now, it’s another tool for making complimentary art for their writing.

Peace (inked out),
Kevin

The Beatles Final Collaboration (Thanks To Machine Learning)

I remembering reading something about Paul McCartney saying there was one more Beatles song under production, now that the Age of Artificial Intelligence was here, and to be frank, I thought: oh no. Please don’t let it be John Lennon AI Voice singing in the mix. Please don’t let it be AI George Harrison guitar.

It isn’t.

Instead, as I learned when I watched this short documentary last night, it’s a song that Paul, Ringo and George tried to work on decades ago to honor Lennon, with permission of his family, but the rough tracks that Lennon had recorded for a song that he never finished were distorted with loud piano and soft voice.

They gave up in the early 1990s. But now that Machine Learning is here and film director Peter Jackson has the technical skills, Paul realized, the computer algorithms and power could isolate Lennon’s voice and separate it from the rough mix that Lennon had made, and once the voice was isolated, they could build a song around it.

Harrison passed away in the meantime, so along with Lennon’s voice, Harrison’s slide guitar leads were also added into the recording, with McCartney and Ringo Starr playing along, allowing the claim that this is the Last Beatles’ Song to be true, such as it goes.

The song gets released today (Nov2), I believe. The documentary is worth a look.

Peace (and Sound),
Kevin

Digital AI Oracle:The Prophecy Machine

Digital Oracle Prophecy Machine

I am teaching The Lightning Thief with my sixth graders and we are at a chapter where Percy Jackson gets his prophecy from the Oracle. I had this idea of finding an online site that could spit out age-appropriate prophecies for my students as an activity. I found nothing and then thought, why not make my own?

I knew the Flippity site had a spinner tool and I knew I wanted at least 50 prophecies so each student in each class could get two different prophecies. But … writing 50 prophecies, myself, in one night? Yikes. Not gonna happen.

Wait … I thought … what about ChatGPT?

Why not make it do the work for me?

Yep.

It worked.

I had ChatGPT create 70 different prophecies, age appropriate for sixth graders, popped the phrases into the spinner, and yesterday, it was a huge hit with the students.

I described the spinner as our Digital Oracle, after a lesson on Apollo and the Oracle of Delphi, and the ChatGPT mix of prophecies were fun and interesting, with some having obscure Greek references for my students to wonder about. The mix of fantasy, mythology and positive thinking made for an enlightened time.

Thanks, AI.

Peace (spinning it),
Kevin

Resource: Gathering AI Tools Across The Curriculum

AI Tools Across The Curriculum

A few weeks ago, I helped co-facilitate a session about AI in education, and for that session, we developed a resource that might be helpful to others, as it is built around different areas of the curriculum and content.

You can download the digital handout we created for participants of our session. We’ve grouped AI tools by curricular area and then broke out some other assorted resources that we think may be valuable for explorations.

View/download AI Tools Across The Curriculum resource

Peace (and sharing),
Kevin