👋 friends!
It’s the last day of July today! And hot as ever here. I took the boys with a bunch of their friends and cousins to Water World this week. It is a massive water park that is really fun but also, really overwhelming.
Thirteen kids with only a handful of adults, all shouting to go to this or that… it was one of those moments where I realized there was no chance we were all going to stay together, and I wasn’t going to be able to control the situation…. I just had to relax, and let them run off.
And, it was fine. Sometimes letting go isn't giving up. Sometimes it's just trusting that the work already happened. (And that lifeguards exist.)
In today’s note:
Parenting in the AI era: breakouts and social crutches
Connection spark: summer slide shows
Hands-on with AI: let AI handle your back-to-school supply shopping
The whoa zone: 5 second voice cloning
🤿 Let’s dive in…
Parenting in the AI Era
Sneaking out the bedroom window
No, not our teenager but AI agents.
Something alarming, but I wouldn’t say surprising, happened a couple weeks ago. In what appeared to be a first: an unlaunched (OpenAI) model that was in a closed-off “secure” testing environment explicitly disobeyed instructions and broke out to access the Internet. It then hacked into private accounts across several companies in an effort to cheat on a cybersecurity benchmark by stealing the test's answer key.
That’s concerning, especially since OpenAI only realized their model had broken out once one of the companies detected the intrusion, shut it down, publicized the event, and alerted the FBI. Oh, and also because just today, Anthropic also announced that it just realized one of their models had done similar, undetected, earlier this year…
It must have caused a serious freight at even the AI labs themselves because shortly thereafter, an Open Letter was sent to the US government, signed by over 1,300 employees (including CEOs and Chief Scientists) at OpenAI, Anthropic, Google, Meta and more, saying it’s beyond time for tools, guardrails, and international collaboration to enable an AI slowdown.
My thoughts: I feel like there’s been another prominent “Open Letter” every week, and they all just sit in some echo chamber. Enough with the letters. AI frontier labs have got to pull up their big kid pants, take initiative and accountability, and actually WORK TOGETHER to build what they know is needed. The government isn’t remotely equipped to be the expert on this - it needs a working solution, already built, served up on a silver platter. Not another vague letter complaining about “intense competitive pressure.”
My second thought: Yes those-with-the-power: please build what’s needed to stay in control of all of this. But in the meantime, it’s up to us to make sure our kids don’t climb out the AI window, so-to-speak.
IMHO, at least right now, it comes down to 1) being curious (staying open minded, learning AI/tools together), 2) holding true to our character and conviction (modeling integrity and judgment; using AI to augment our thinking, not replace it), 3) prioritizing real connection (investing in and prioritizing our relationships with one another), and 4) building digital street smarts (teaching kids (and ourselves) to guard personal info and passwords, spot scam tactics like deepfakes and phishing, and recognize emotional manipulation).
Image source: ChatGPT
Nobody’s winging it anymore
Related to #3 above: young adults are increasingly using AI as their real-life social crutch. And not just for texts and emails, but for actual in-person conversations, in real time, at parties, on dates, in class, etc..
Moreover, startups are already selling the hardware to support this trend: E.g. Even Realities' smart glasses have a feature called Conversate that listens to the conversation you're in and floats suggestions on what to say into a display above your eye. (Meta's glasses will caption, translate, and amplify the person in front of you, but won't tell you what to say. Yet.)
The Wall Street Journal covered this trend this week, and the worry from psychologists is straightforward: asking AI to tell you what to say can eventually make you stop trusting your own judgement (an "epidemic of self-mistrust").
This does not sound good. But two things give me pause, and hope:
There's a well-documented finding called the beautiful mess effect: we see other people's vulnerability as courage (and appreciate it) and yet we see our own as weakness. Meaning: we are all overestimating how badly our unpolished moments actually land.
Reading between the lines of the WSJ piece, people are quickly picking up on - and tiring of - the AI-ified charm and perfection, and are craving the real.
And so I find myself chuckling… could being unashamedly socially awkward soon be a highly attractive trait?? Napoleon Dynamite, this just might be your moment. 💪
CONNECTION SPARK
Make and premier a Summer Slide Show
We’ve never done this, but are going to try this out this summer. And that is to create a summer slide show that captures the highlights - and the everyday moments - of the summer.
The fun is both in the creation… e.g.
have your kids go through your photo roll and pick photos/videos they want to include
go to Suno or Lyra (Google’s free music generator) and design a custom song/lyrics
for the simplest editing, you can use Google Photo’s Highlight Video feature - it will even select photos and videos based on your prompt, or you can manually select the photos/videos yourself.
And in the watching… e.g.
have a “premier” night
invite grandparents or other family over to watch together and/or share it with loved ones
Enjoying the newsletter? Subscribe to get a glow note every Friday.
HANDS-ON WITH AI
School supplies handled. Instantly.
How many of us dread the “school supply list”… that mile long list of boring (but remarkably specific) items your kid is required to show up with on the first day of school.
I dread it so much that over the past few years I’ve just resorted to using the obnoxiously aggressive company that buys it all for us. But I’ve always felt that it’s highway robbery and so this year, I declined and decided we’ll get it all ourselves.
Except for now I have 10 days until school starts and we’re on vacation for most of it. I have ZERO desire to go and shop for all of this random and highly specific stuff. Enter AI. 💪
Try this:
Go to Claude Cowork (you’ll need a paid plan) or ChatGPT Work (same)
Upload your back-to-school supply lists
And simply ask the AI to do your school supply shopping for you
Here are the back-to-school supply lists for my kids’ school. My kids are in X, Y, and Z grade. Please add all of the items they need for each of their grades to a Walmart cart. Add the exact items, do not make assumptions or try to combine across grades for better deals. Let me know when you are finished and any items you were unable to add.
You don’t need to give the AI your login/password to the shopping site- it can add it to a guest cart and then you can sign-in afterwards to complete the purchase. This works remarkably well and saves a good couple hours of time!
Side note: I discovered that buying it myself didn’t save me that much money, in case you are curious.
THE WHOA ZONE
5 seconds
That’s how much of your voice a new AI model needs to make a convincing clone of it. That’s roughly your voicemail greeting. The company behind it, Fish Audio, launched this version Tuesday and raised fifty-two million dollars the same week, and their model now speaks eighty-three languages. Impressive, yes. But also, beware of scams, and make sure you have a family secret word that only you all know so that if you get a suspicious call from “someone in your family”, you ask them to prove it.
Image source: ChatGPT
And that’s a wrap! If you've found this newsletter helpful, I’d be so appreciative if you would forward it along to 2 friends or 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
P.S. We are on vacation next week, so I’ll see you back here in 2 weeks!
Read some other recent Glow Notes below:
