What boiled egg pudding taught me about AI
Add Axios as your preferred source to
see more of our stories on Google.

Illustration: Aïda Amer/Axios. Stock: Getty Images
I won't lie: The mix of food in my blender looked really gross. It was literal AI slop.
- I was in the middle of another AI experiment, one of many that have given me a handful of light bulb moments with the technology that I was profoundly skeptical about only a year ago.
Why it matters: To understand AI's transformational power, I had to try it, to noodle and experiment, to see if it could become useful — or even essential.
- Once I did, doubt turned into cautious belief — or dare I say, faith.
Zoom in: What we'll call the parable of the boiled egg pudding began with a "what if."
- Could Muse, Meta's new viral assistant, take the hundreds of high-protein recipe videos I had saved on Instagram Reels (I have a scrolling problem — don't judge!) and turn them into a tidy recipe book, with nutrition facts included?
Sure enough, it took Muse a few seconds to produce "My Recipe Book: 250 recipes from my Instagram saved collection / Collected over time, cooked with love."
- I left the title untouched — a reminder that AI still can't do everything, even if the instant PDF was impressive.
- As an always-aspiring healthy eater, I had apparently saved 250 videos with plans of making them some day, but had never revisited a single one.
🍳 The intrigue: Reels is a great medium for finding recipes, not so much for following them. You have to watch the video over and over to log the ingredients and scroll back and forth to make sure you've heard the steps right. The book made the videos instantly usable.
- Back to my blender: Once combined, the ingredients — six boiled eggs, cacao powder, maple syrup and some milk — turned into a pretty smooth-looking chocolate pudding. It actually tasted pretty good. I mean, not like creme brûlée or anything, but a decent solution to my never-ending late-night sweet tooth.
Zoom out: The surprise of that moment, and the accompanying musings on how long it'll be before we are all living in the movie Her, reminded me of my foundational conversion moment as an AI user nearly a year ago.
- I had enjoyed the chatbot era, was an early power user of Google's Notebook.lm and spent many a night working on Studio Ghibli-esque portraits of my family. I liked it. But the singularity? Give me a break, I would say to myself.
Then my Aunt Betty died. A fiery, accomplished psychologist with a penchant for quicksilver jokes and asides, Betty passed away after a long battle with multiple sclerosis at the age of 93.
- I flew to New York to help her wife Marilyn (my aunt and my mother's sister) manage the aftermath.
As many people are after a family death, we were both overwhelmed. Marilyn was shell-shocked. We needed to notify friends, organize a service and deal with all the end-of-life to-dos.
- Marilyn, a writer who had recently self-published a memoir about her pioneering, 55-year relationship with Betty, didn't feel up to the task of writing an obituary, but she also didn't want to send out a note about her wife's death without it.
I had an idea: What if I uploaded Marilyn's memoir into Notebook.lm — an AI productivity tool that synthesizes only data you compile into a notebook — and asked it to use the information to write an obituary?
- I provided a few instructions, or prompts, in AI lingo: Write an obituary of about 1,200 words that highlights Betty's work as an activist, her love for Marilyn, and her tendency for blunt and hilarious quips and one-liners.
In a few seconds, the "thinking machine" produced a lovely obituary that felt very much like it was in Marilyn's voice.
- We then worked through it together over about an hour, with her suggesting revisions, additions and finishing touches. We read it aloud together, laughing and crying. It was magical.
After that, I decided I would give AI a chance. As a writer, I couldn't dispute its utility, and I was finally convinced about its transformative power.
Reality check: When I say I am a believer, I don't mean booster. My journalistic fine-tuning is as punk rock as ever, and I intend to keep giving all the companies an equal-opportunity hard time as they shape the future, sometimes far more carelessly than any of us would wish.
- But I won't waste a lot of time pretending it's no big deal. Give me a break.
