Steps in the right direction re: teen mental health

Big Tech is starting to listen: over the last 2 weeks, there have been some launches and announcements from Meta, Character.ai, and Open AI about teen mental health and safety that are encouraging, and hopefully a sign that this is a growing priority.’’

What was announced?

• Character.ai (a company that allows people to create and chat with AI companions) announced that starting November 25th, it would prevent anyone under 18 from having open-ended chats with the service. Teens will still be able to use the platform to generate videos and images through creative tools. For age verification, they will deploy in-house in-house age detection tech that analyzes user behavior, prompting verification requests when it suspects an underage user (similar to ChatGPT’s proposed approach).

• ChatGPT rolled out updates to GPT-5 that are designed to better recognize and respond to users experiencing mental health emergencies. The improved model scored at 92% compliant with desired behaviors in challenging mental health-related conversations as compared to 27% with the prior model. Note: also it was shared that over one million people each week have conversations with ChatGPT that include “explicit indicators of potential suicidal planning or intent.” Yikes.

• Meta announced new parental controls coming in 2026 that will give parents insights into the topics their teens are chatting about with AI characters and will allow them to turn off their teen’s access to chats with AI characters and/or block specific characters. That said, and of note, parents will not be able to block teens from chatting with Meta’s own AI assistant. 🙄).

AI and the new era of de-skilling

This week the Atlantic had a thought-provoking article they called “the Age of De-skilling.” In it, they argue that human “de-skilling”, as a result of AI, is happening - and is probably inevitable. And so, what to we do about this?

"De-skilling" isn’t novel - it has always accompanied new technology (writing meant we memorized less; calculators changed mental math). But AI is an entirely new beast and we must ensure we maintain our agency. AI is our collaborator and co-pilot, not our brain.

Our kids, with brains still developing, are growing up in the advent of AI and we can’t let them become dependent on AI before developing fundamental skills - a recent study found younger people who rely heavily on AI scored lower on critical thinking tests.

This leaves us with a not-small responsibility: it’s not about shielding our kids from AI; it's about arming them against dependency.

Trick-or-chat?

Tonight many/most of us will be running around the neighborhood after our kids, trying to make sure they’re being safe, not grabbing too much candy, and saying thank yous.

But this is also a rare time when we’ll have so many face-to-face interactions with our neighbors, and so why not make these interactions slightly more intentional and conversational and grow our friendships in our community while we’re out and about in person?

Moreover, costumes can be great conversation starters and/or opportunities for genuine compliments.

Tonight’s connection challenge: move beyond “trick-or-treat” and “thank you.”

• When a neighbor opens the door or as you meet other families in the street, try offering a genuine compliment and/or asking a question that shows genuine interest.

• After the interaction, replay their name along with something about them/the chat in your head to help improve recall for when you see them next.

  

Candy Overload

If there’s one thing that’s pretty certain, our homes are about to get an onslaught of candy tonight. There are some neat ways you can use AI to, believe it or not, turn the piles of candy into fun and creative learning opportunities. For example:

1. Candy data: create charts and graphs that show the candy breakdown in different ways.

   • Spread the candy out on the floor and snap a picture

   • Upload the photo to Gemini or ChatGPT and ask it something like “Analyze the attached photo of Halloween candy. Count and categorize each candy type, then create a series of colorful, fun graphs and charts to visualize the data in different ways for a kid-friendly audience.”

2. Candy science experiments: turn your treats into exciting science activities and learn while you play.

   • Upload the same or another photo of your candy haul and ask Gemini, Claude, or ChatGPT something like “Based on the candy shown in this photo, generate a list of fun, kid-friendly science experiments we can do using this candy. The experiments should be safe, easy, and visually delightful for young kids, with clear steps and playful explanations of the science behind each one.”

The gift of sight

Stanford researchers developed an electronic chip that, implanted behind the eye with smart glasses, restores vision to people with advanced macular degeneration. In clinical trials, 27 of 32 participants regained reading ability within a year. This may be the first prosthetic to restore form vision (not just light sensitivity), offering hope to those with irreversible central vision loss.

A record-breaking drone show

My son hates fireworks because of how loud they are. Maybe we should move to China. Check out this world-record-breaking drone show they recently put on there with 16,000 drones.