Category: Artificial Intelligence

Luxemburg flickr photo by dutchphotographer shared under a Creative Commons (BY-NC-SA) license

I’d recommend perusing this deep dive paper into how Generative AI is and will impact learning spaces. I appreciated how each section of “The Manifesto for Teaching and Learning in a Time of Generative AI: A Critical Collective Stance to Better Navigate the Future” has a “critical insight” reflection.

The researchers look at both the positive

and the negative

In part of the conclusions:

“As learning extends beyond formal settings to include life-long, life-deep, and life-wide learning, the need for GenAI literacy and competences becomes critical. Understanding the underlying mechanics of GenAI can help users navigate the technology responsibly and mitigate the potential harm caused by hallucinations, biases, and misinformation.” — from The Manifesto for Teaching and Learning in a Time of Generative AI: A Critical Collective Stance to Better Navigate the Future

Peace (in the balance),
Kevin

This is my second post in the 12 Days of AI activity, and today’s Day 2 prompt suggests we use a site called Deep Dream Generator to have AI tinker with a photo or drawing. I took a photo of the peel of a banana I had just eaten, and tinkered with the site’s settings, and ended up with the neat illustration above that turned the banana peel into a bird.

The guiding question for this prompt was all about passing the Ada Lovelace test — does the new output create something “new and surprising and different” from the original? I’ll say yes, for now. I was not sure what I was expecting to happen but I was happily surprised to see the peel become a bird. I had to wonder why for a bit, and then remembered that one of the settings I checked was Audubon, so  — aha — a bird! Of course!

I did try one of their other Deep Dream tools for making a video but it was pretty lame. And when I tried a morphing of images, I had run out of credits. So I took that AI image from Deep Dream into the Generative AI Fill tool of Adobe Firefly, and played around with some backgrounds, until ..

Peace (and play),
Kevin

I signed up to get daily prompts around Generative AI explorations through the 12 Days of AI project. Just because … I am curious and I want to learn and most of all, I want to understand more complexities of the emerging AI world, and I can do that best by diving in. This project seems more geared towards university folks, but I am sneaking in, anyway. There is a discussion space inside Microsoft Teams, but I will probably just keep an eye on the daily blog posts.

For the first prompt, released today, we are asked to use Adobe Firefly, which is an art generation site I have used many times in the past, so I am pretty familiar with it. But I had never uploaded an image after writing a text-image prompt, as a way to play with the composition of the art output.

I asked Firefly to generate an image for “an open book of magic with musical notes floating off the open page” and then uploaded an image I had of a coffee mug at a Maine beach, and tinkered around. Interestingly, the more tinkering I did, the less I liked the results, as each step “lost” the coffee mug compositional reference point.

I did finally play with lighting and a few effects (newspaper filter effect) and got something I sort of liked.

Peace (in a mug of music),
Kevin

In this fourth iteration of exploration of bird song, AI platforms and music composition (see post one — post two — post three), I took some of the concepts of bird song sounds and music composition into the AI song-making platform, Suno, and gave it instructions about making a gentle folk song, with the And Then There’s The Cuckoo title from yesterday’s podcast remix, added some of the informational text from my earlier use of Google’s Learn About and NotebookLM, and hit “create”.

You can take a listen to the song it created.

I mean, Suno does a decent job with that I gave it. Each time I use Suno, I get the same sense that this is pretty amazing (and that the quality improves every few months) and also, I worry about whose sounds have been gulped into the machine to create this AI music. The vocals here sound familiar to my ear, but I can’t quite place whose voice it sounds like. The piano is rather lovely, too. And there are little touches here and there that make it for an interesting listen.

The lyrics below are what the Suno AI generated with my prompt.

And Then There’s The Cuckoo

[Verse]
In the hush of dawn’s first light
Where oak and willow stand tall
Composers find their dreams alight
With the thrush’s beck and call

[Verse 2]
A robin weaves a melody
So raw
Untouched by man
Its notes inspire symphony
In hearts where dreams began

[Chorus]
And then there’s the cuckoo’s sound
Echoing through the wood
In its song
A truth is found
Where nature’s voice stood

[Verse 3]
The blackbird’s tune
A haunting thread
A whisper of days gone by
Its cadence fills the heart with dread
Yet lifts it to the sky

[Verse 4]
Beneath the lark’s ascending glee
A composer pens his line
Caught in the wild’s euphony
Where thoughts of rapture twine

[Chorus]
And then there’s the cuckoo’s call
A note so pure and clear
It rings within the concert hall
And chases away the fear

Peace (and song),
Kevin

This is my third iteration of some explorations I am doing with Google’s AI Platforms and with an inquiry into how music composers have used bird songs for inspiration. I started in Google’s Learn About (see post) and then moved into NotebookLM (see post) and now I have taken the voices from the AI podcast file and remixed some of what they were saying into a short remix piece.

I loved the phrase “And then there’s the cuckoo” so I used it as a sort of refrain, with a sound of the cuckoo, calling back every time the male’s voice says it.

I have at least one more iteration to explore, with a more song-focused concept. I don’t know how it will go. And in my head, I have a future project, too, where I work to compose a bird song myself, using the knowledge I have been gathering in my inquiry explorations. That may require more thinking. But I want this to go beyond playing with AI tools — I want to take what I have found and move the idea forward, on a human scale (but bird-song-influenced).

Peace (calling it out),
Kevin

The other day, I wrote about using Google’s Learn About AI platform, with a focus on how bird songs influenced music composers. I took my notes from that Learn About platform and moved some of it into Google’s NotebookLM to do some further inquiry (and to save my notes, as Learn About refreshes each time you exit). The result was a sort of ancillary notebook of ideas, as NotebookLM both gathered my research and then expanded on it a bit more. I am still intent to dive even further and look into how to compose music, with bird song as the backdrop.

I like that in NotebookLM I could create an outline of the main elements of the resources and look at a timeline of discoveries and creation of music. For example, I am now curious to further investigate how David Rothenberg (I don’t know of him) created a jazz piece with a thrush.

Here is the timeline:

This timeline focuses on the use of birdsong in Western music, as the provided sources are limited in their scope regarding other musical traditions.

13th Century:

• c. 1250: “Sumer is icumen in,” an English song, becomes likely the earliest instance of notated birdsong, featuring the cuckoo’s call.

14th Century:

• Composers like Jean Vaillant begin quoting birdsong in their compositions, marking a growing interest in incorporating natural sounds into music.

17th Century:

• 1650: Athanasius Kircher publishes Musurgia Universalis, an encyclopedia that includes musical notation of several bird calls, demonstrating an early attempt at systematic transcription of birdsong.
• c. 1669: Heinrich Biber composes Sonata Representiva, featuring sections named after birds and animals. He uses string techniques to imitate bird sounds, including the nightingale, cuckoo, and chicken.

18th Century:

• c. 1715: John Walsh publishes Bird Fancyer’s Delight, a collection of short musical phrases labeled with bird names intended for teaching caged birds to sing.

19th Century:

• Composers like Beethoven, Delius, Handel, and Vivaldi prominently feature birdsong, particularly the cuckoo and nightingale, in their works.

20th Century:

• 1924: Cellist Beatrice Harrison performs a live BBC radio broadcast with nightingales, marking the first outside broadcast for the BBC and highlighting the potential for musical interplay between humans and birds.
• 1923-1924: Ottorino Respighi composes Pines of Rome, possibly the first piece to incorporate pre-recorded birdsong.
• Late 1920s: Respighi composes Gli Uccelli (“The Birds”), based on Baroque pieces imitating various birds.
• Mid-20th Century: Olivier Messiaen extensively utilizes birdsong in his compositions, including Catalogue d’oiseaux and Réveil des oiseaux, pioneering a detailed approach to integrating birdsong into music.
• 1960s-1970s: Popular music bands like Pink Floyd and singers like Kate Bush begin incorporating birdsong sound effects into their albums.

21st Century:

• 2000: Jazz musician David Rothenberg improvises a duet with a laughing thrush, exploring the musical connection between humans and birds and challenging the traditional definition of music.
• 2003: Jonathan Harvey’s Bird Concerto with Piano Song premieres, utilizing slowed-down recordings of American West Coast bird songs to reveal their intricate musicality.
• 2017: Hollis Taylor publishes a book arguing that the vocalizations of the pied butcherbird are music, using musical transcriptions and field recordings to support her claim.

Of course, I had to try the AI Podcast creator, but since I have so many sources and notes, the audio is pretty long (about 15 minutes).

Listen, if you are curious.

My next step is a bit of remix of those voices. That will be coming soon.

Peace (in explorations),
Kevin

I promise, I am not a shill for Google, but I keep stumbling across some of their AI experimental platforms, and so as part of my ongoing push to experiment with AI, I keep playing around.

The latest is something called Learn About, and as the title indicates, it is built for the curious. I liked it.

Once you give it a topic, the AI engine gets to work, pulling together resources from across the Web (videos, conversational chats in placed like Reddit, PDF resources, etc.) and begins to create an outline for learning, including suggested further lines of inquiry. It makes charts from its information database, and offers to “go deeper” or “simplify” its results for you.

It even periodically tosses in little quizzes and check-ins, to ensure you are thinking about what it is presenting to you.

I am reading the book Atlas Obscura: Wild Life and when I came across a section about bird song and music notation. I wanted to learn more, so I went into Learn About, and darned if it didn’t send me down some fascinating rabbit holes with all sorts of interesting corners of information. I learned about how music composers are inspired by bird songs, how spectrograms can track pitches of birds, how bird song can help with conservation efforts, how individual birds have their own songs, and more.

I am wondering now, how to merge the inquiry in this Learn About platform with Google’s other platforms, including NotebookLM. That’s my next step.

Peace (singing it in song),
Kevin

Google continues to experiment and expand the use of AI-generated audio. It’s “podcast” feature in its NotebookLM is interesting (and a bit unsettling) but Google also has its Illuminate platform. Illuminate turns PDF research documents into audio discussions by parsing out the main ideas and creating a ‘discussion’ among AI voices.

I tried it out, and found a paper about Connected Learning in the Illuminate library. Within seconds, it had created a three minute audio file, with transcripts, and access to the original document. Illiuminate also allows you to open a chat box and ask questions of the document itself as you listen to the audio analysis. (The responses are text, not audio, but I suspect at some point, that could change.)

Here is an article about Google’s work on audio.

Here is a link to my Illuminate file on Connected Learning. It will probably force you to make an account, so decide that for yourself. The image above shows you what I am seeing, except I layered a screenshot of the actual PDF article onto the dashboard itself. You can download the audio file for use elsewhere. (Listen to the file generated for Connected Learning here, via Vocaroo hosting).

I think this could be an interesting way to parse through research documents, and gather information. The audio will likely get better. You can’t upload PDFs directly, but you can provide URLs for Illuminate to use.

Is this good? Bad? Again, as with the audio feature in NotebookLM, I am not yet sure how to think about it. But I remain curious. And continue to play.

Peace (diving in),
Kevin