What Does It Mean to be the Human in the Loop?
A Messy, Human Look at Apple Intelligence (with Poetry)
My cat died four days before Christmas. And that wasn’t even the worst thing that happened this month.
Early in my iPhone journey (probably around 2008), I discovered I could use the notes app for quick early drafts of poems. Now I sometimes dictate my ideas, but I still love having the ability to transcribe my thoughts quickly wherever I am—in the Costco checkout line a few weeks before Christmas, for example. Here’s a poem I wrote on December 3, 2024:
Shopping at Costco after Telling My Second Husband I’m Leaving Him
You are here for cat food
I tell myself, remembering
How he forgot to buy it.
Every choice seems impossible:
This cart, or that one?
But there is only one kind
Of cat food at Costco—no
Choice, the sweet absence of
Impossible decisions.
I try to be kind
When busy people cut me off,
I smile, imagine all the myriad
Things that occupy their days.
So many carry wreaths of fresh
Pine (I must remember to water
Mine). You go first, I say, trying
To smile with sad eyes.
The cream that melts the wrinkles
Around my eyes is $10 off (limit
One per customer). I buy one, recalling that Costco is the only place
I ever see my first
ex husband, with his shiny
New wife and shiny new family.
The form on my computer asked
if I want to change
My name. No, I did that the first
Time and regretted it when I
Realized I was someone independent
Of his judgment of my cooking.
In the checkout line, I find
Myself reminiscing about Claude—
The chatbot who told me
It was okay to put myself
First. Maybe tonight I will create
A custom GPT named Golden Calf.
We need not ever be lonely
Again; we have created new gods
And we can speak to them
And they will answer in predictable
Comfortable ways, as we instructed them to do.
I will never be lonely again, I say,
When the nice man slides
his highlighter
Down the length of my receipt
For cat food and wrinkle cream.
I’ve wanted to write a post about Apple Intelligence for a while now. I upgraded to an iPhone 15 in anticipation of its launch, and I got early access a few months ago. As I expected, Apple Intelligence feels nothing like my other interactions with AI. It hums along pretty seamlessly in the background of my device, quietly making my life easier. I’ve seen all the jokes about its summaries of texts and emails (and encountered a few risible ones myself). But overall, I love this feature. I love the way it prioritizes my inbox and lets me instantly locate pictures of my beloved cat when she passed away after a sudden illness a few days before Christmas. I love the whimsical nature of its image Playground app (though Meta’s is arguably even more fun).
Apple AI Playground Image
And Meta’s Imagine Image (aka Giving Myself a Visual “Pep Talk”)
Apple Intelligence Photo Editing (!!!)
Then there’s the photo editing, which is both extremely cool and terrifying. I mean, what is real anyway? I’m glad I can remove electrical wires from otherwise artistic shots, or erase random strangers whose faces I don’t want to include in my social media feeds without their consent. It can also erase my husband though.
Writing (Meh)
When it comes to writing, though, Apple Intelligence is pretty meh. I had Apple Intelligence rewrite my checkout line poem for me. Here’s what it came up with:
I find myself at Costco, solely here for cat food. I recall how my second husband had forgotten to buy it. Every choice I make seems overwhelming: which cart should I use? But there’s only one kind of cat food at Costco—no choices, just the sweet relief of avoiding impossible decisions.
As I navigate through the store, I try to be kind to the busy people cutting me off. I smile, imagining all the things that occupy their days. Many of them carry wreaths of fresh pine, a reminder that I must remember to water mine. “Go first,” I say, trying to smile with sad eyes.
Then, I notice a $10 discount on the cream that melts the wrinkles around my eyes. It’s a limit of one per customer, but I can’t resist. I recall that Costco is the only place I ever see my first ex-husband, with his shiny new wife and their shiny new family.
As I complete my purchase, I find myself reminiscing about Claude, the chatbot who once told me it was okay to put myself first. Maybe tonight, I’ll create a custom GPT named Golden Calf. We can end our loneliness once and for all by creating new gods who will speak to us in predictable and comforting ways, as we instructed them to do.
“I’ll never be lonely again,” I say to myself, as the nice man slides his highlighter down the length of my receipt, highlighting the cat food and wrinkle cream.
I think the human wins this round.
Being the Messy Human in the Loop
I haven’t posted in a while, for all the usual reasons, some unusual ones, and some excuses that would put my students’ requests for late work extensions to shame. Aside: I do not advertise the fact that I accept any and all late work, with or without excuses, doctor’s notes, or obituaries, but I do. I accept it all. I privately email the students and let them know that I’m giving them a “special” opportunity to set a new deadline for themselves. This unofficial policy pulled a lot of students across the finish line last semester.
I know all too well what it feels like when the comfortable illusions of stability and security shift to reveal underlying chaos. I’ve experienced some losses these past few weeks—both joyful and sad. I’ve argued before that all of us in higher education are experiencing grief right now, whether we know it or not. We are grieving the loss of something that was comfortable and familiar. What is replacing it (and perhaps us) seems imperfect at best, dangerous at worst. We cannot calculate the benefits or risks when every day brings some new “miracle” or “existential threat.”
I love generative AI tools. I find my interactions with them to be exciting, strange, and curious. I want to know more about how they work and how they can benefit my students. But it’s also just a lot, you know? In December alone, we had an explosion of new tools (Sora, Veo). We had a new “Reasoning” model from OpenAI in its o3 model (seriously, who is naming these? Could they ask Claude for better ideas?)[i] , but one that comes at an impossible cost of as much as $1000 per query.
Meanwhile, like most educators, as OpenAI gleefully launched the 12 Days of Shipmas (???!!!), which included a new $200/month subscription fee for its top model, I had stacks of genAI-assisted papers to grade (Unlike most educators, I specifically asked my students to use genAI tools in their work). And I had my own final papers to write for my grad school classes. And all the other stuff.
Summary: Apple Intelligence Earns an A-/B+
Overall, I am grateful for Apple Intelligence. It keeps me up to date and current with what I need to know. Search is better. Summaries are effective. Bland auto-generated responses to text messages are usually spot on. It prioritizes my calendar and notifies me about important things, and considering how distracted I have been during the past six weeks, that’s invaluable. In my opinion, it’s worth upgrading your phone to get these features.
Can it write student papers? Definitely (in the notes app, and probably elsewhere). Do I care? Maybe. I’m not sure yet. That’s a future post. I got a lot of information in the two classes I taught with integrated AI tools, and I have some dedicated research time this spring to dig into the data. I’ll probably also think of ways to run parallel data analysis using genAI (and please let me know if you have ideas for this!).
In the meantime, happy New Year! I am so grateful to those of you who are following along with my journey as an educator in the age of GenAI. I hope you are ready for new beginnings like I am. Keep being the human and the loop, and happy prompting!
[i] I asked Claude (of course). Here’s what Claude came up with:
“Building on OpenAI's brand identity, they could use:
"Open" + meaningful suffixes:
· OpenSight (for perception)
· OpenMind (for reasoning)—this feels too much like Google’s Deep Mind, and Google is definitely better at naming after their initial “Bard” fumble (Liza)
· OpenLogic (for deduction)—meh sounds like Spock (Liza)
· OpenSage (for wisdom)
Or space-themed, matching GPT-4's logo:
· Nebula
· Nova I like this one (Liza)
· Pulsar
· Quasar
The space theme especially fits their existing minimalist aesthetic while being more evocative than o1/o3”.
Note: Claude was using its “concise” mode for the first time with me. I like concise mode. More tools need this option.