👋 friends!
I’m back from vacation and fully in the throes of back-to-school whiplash.
We spent the last day of our trip in Big Sur, my favorite place in the whole world. And then, the next morning the Timber Fire broke out. The restaurant we'd eaten at is under evacuation orders, Highway 1 is closed, and we missed all of it by one day.
I feel relief, coupled with guilt for feeling relief. And I'm deeply worried for the people and the livelihoods still facing the fire.
In some ways it feels like a microcosm of everything right now… so much happening, so much of it consequential, and so little of it in our control.
So let’s control the bits we can, feel our feels, look out for each other along the way, and make the ordinary days count. Turns out one of them can hold a lot.
In today’s note:
Parenting in the AI era: why teens prefer AI for emotional support and a reality distortion
Connection spark: boosting our social status
Hands-on with AI: (easily) build your family’s grocery agent
The whoa zone: gym aggression on a whole new level
🤿 Let’s dive in…
Parenting in the AI Era
Why teens are flocking to AI for emotional support
Many teens (and even tweens)are increasingly using AI for emotional support, and among those who do, about a third share things with AI that they don’t even share with their closest friends or parents. A new study sought to find out why, and here’s what it found:
Not wanting to burden anyone. "You don't want to trauma dump all over your friends, obviously."
Perceived safety from judgment and consequences. "It won't be used against me."
Simulated understanding. "It's like it went through the exact thing that I went through."
Always available, instant. "I do not want to wait two to three days to get a call or text back."
Neutral and private referee. Teens use it to ask "am I in the wrong?"
Advice that's concrete instead of generic. "With parents, I think they'll just tell you you need to go outside... they'll give me something like a quick Band-Aid. I feel like I need to go in-depth." (this was a helpful one for me to hear!)
My thoughts: it is frankly hard to argue with this logic. But also, understanding the why - and these specific needs - can hopefully help us as parents. (E.g. making more frequent, safe, and supportive spaces for trauma dumps and watching out for cliches and genericism in our advice… etc.)
from Hope Lab’s report
Reality distortion
The people building AI keep telling us how AI will improve family life, and well, they certainly seem to be living in some alternate universe.
Sam Altman shared a “cool use case” for ChatGPT: connect it to your calendar and have AI generate a daily podcast for the school drive about your kids’ games, birthdays and activities. The top comment (with more likes than Sam’s post itself) simply said, “what if you just talked to your children?”
And then this week Mark Zuckerberg published a 6,500-word manifesto called The Future Is for Everyone, in which he worries aloud about power ending up in too few hands (umm….) and offers personal AI as the cure for pretty much everything by being able to improve our “relationships, health, career, finances, home management, hobbies, and more.”
It hasn’t landed particularly well, for many reasons. In Verge’s response, “What strengthens a relationship? Almost always, it is personal investment.” Side note: a post went viral this week about someone in SF breaking up with their partner after they found out he was automating almost every aspect of the relationship with AI.
It is almost funny, except for the fact that these are the people who are making decisions that will materially impact our lives and livelihoods, and there is clearly a distrust that they truly have our best interest at heart.
I don’t quite know where I’m going with this, other than a reminder that while these guys may be AI geniuses, they are not life genuises, so let’s take their advice with a grain of salt.
CONNECTION SPARK
The surprising key to status
Perhaps here’s some further insight into why these AI guys are falling flat: in one of the largest cross-cultural studies of what raises someone’s social status, the single highest-rated trait wasn’t wealth, intelligence, leadership, achievement, or overall impressiveness. It was simply being someone your group could trust.
Honesty and kindness were also among the biggest status-raisers, while being unreliable was one of the biggest status-killers. More recent research validates this: people who combine strong character with competence tend to earn particularly high social status.
In a world that seems to be obsessed with followers, trophies, elite schools, promotions, viral moments, being the best at a sport, etc etc this insight can be a really helpful reset for our kids (and ourselves): simply being a trusted and reliable friend is the ultimate status booster.
Enjoying the newsletter? Subscribe to get a glow note every Friday.
HANDS-ON WITH AI
Let’s get cookin’
Those of you who know me will laugh at the understatement when I say I don’t like to cook. My biggest household pain point is grocery ordering and meal planning… I hate it.
This obviously feels like something for me to hand off to AI, especially since it’s been the classic what-to-use-AI-for example for years. But honestly, until lately, I hadn’t. Mostly because I’m not good with routine and planning ahead, and our family’s needs are so nuanced and ever-changing that I didn’t trust AI to manage it without a lot of mistakes requiring me to go back and fix things.
But AI is getting better by the nanosecond, and it’s now possible to whip up (pun intended!) your own household sous chef agent that actually works.
What’s more, you don’t need an elaborate prompt or a bespoke tool — just one sentence to kick-off the setup:
Go to your Claude desktop app (ChatGPT Work would probably do a decent job too). I used the code mode, but you could probably also build it just through chat /co-work mode and say:
❝“I want to create an agent that will solve my grocery and meal prep pain. First, do a comprehensive interview of me to understand our family situation, my pain points, requirements, and desired interface.”
Sample grocery agent build kick-off
Invest 20 minutes in the interview and watch it spin up your perfectly personalized grocery agent.
You’d be surprised about the level of functionality you can demand (e.g. I ask it to email Brian to get any inputs from him, to give me a choice for a weekly meal my kids can help cook, and once the grocery list is approved, actually add all the items to my Amazon Fresh cart.
Once you have a first draft, test it out (Claude will tell you what to say to kick it off — for me it was “do the grocery run”) and tell it what to fix until you’re happy.
Then set it up as a recurring task so you don’t even have to remember to kick it off each week.
Mine is built and tested - and I can already tell it’s going to be life-changing. This weekend we’ll “launch” it, putting it to work in real life. I can report back if you’re curious!
screenshots from my Grocery Run dashboard
THE WHOA ZONE
Gym aggression on a whole new level
This week, a man named Andrew, from Australia, asked his AI agent to book him a gym class and see if he could get moved up on the waiting list. Before he knew it, the agent had taken advantage of a software vulnerability and fully deleted the person who was at the top of the waiting list to get Andrew from #4 to #3.
Technically, the AI was following orders. But also, very unethically.
My thoughts: unless we figure out how to quickly enforce better values and judgement in LLMs (not remotely easy given the number of models and companies and governments and opinions and…), this does not bode well once everyone is sending their personal agents off on “errands.”
And that’s a wrap! If you've found this newsletter helpful, I’d be so appreciative if you would forward it along to a handfull of fellow parents who you think might also like reading this.… we’re all in this crazy together.
And, as always, if you have any thoughts, feedback, or requests, please reply to this email with your notes - I’d love to hear from you!
Glow on,
Michaela
Read some other recent Glow Notes below:
