Following last week’s onslaught of cutting-edge AI model launches, this week the topic has shifted to AI safety - and has exploded across the Internet and politics globally.

And while the sentiments that sparked the discussion are shocking and scary, the fact that we are finally having this conversation publicly and seriously is, IMHO, a good and essential thing.

The tipping point was Tuesday evening when an Anthropic researcher, Jacob Coxon (who had also been at OpenAI) quit and announced his reason on X: “The people building AI earnestly believe that it could kill us all by the end of the decade.” and “Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives.”

That post now has over 168 MILLION views… !!!, in large part because then other top AI researches started chiming in to validate his sentiments.

I have so many thoughts. But I’ll focus on three here:

1. Um… really?? The end of human civilization in the next few years?!

I’m sure you’re wondering, is this remotely credible or possible?? And as you might imagine, there are strong opinions across the spectrum. My personal position is that very smart people who have been working on this for many years are very worried. And so yes, I think AI safety is an urgent international societal issue that has not gotten the attention and coordination it needs. That said, I’m also not about to start running to the hills or buying into bunkers.

I AM grateful that this guy finally managed to get the topic of AI safety (and esp the risks of recursive self-improvement and superintelligence) out into the mainstream and top-of-mind. I found Derek Thompson’s write-up yesterday to be very good if you’d like to start to understand this better.

Bernie Sanders is going to be addressing the US senate next week (16th) with an AI regulation bill that proposes to ban Superintelligence (AI that is smarter than humans). One thing we can do right now: send a note to our congressperson if we’d like to express support for smart AI safety regulation.

2. This may reach our kids, and we should be ready.

With a post that viral, there’s a good chance this “news” will trickle down to our kids. We should be ready for them to ask us about this.

I went looking for expert guidance written for this specific moment and wasn’t able to find anything (as of yesterday evening). But here are a few tips from experts for talking to your kids about big societal worries, and as experts start to weigh in on this particular issue, I can circle back as well.

tl;dr: Confirm what they’ve heard, validate their feelings and put the prediction in perspective; point to all the grown-ups and experts working on it; take it off their plate.

3. WHERE are the women and mothers in this debate??

I won’t belabor this, but it has become painfully clear how few women (and moms) are leading and influencing within these AI frontier labs. I’m on a mission to find and support more who are.

This week, I’m focusing on the person who so often ends up last on our list: ourselves. Here are two ideas I love for those moments when we’re feeling a little depleted (or a lot).

What’s in your soul box?

Vivek Murthy shared an idea that I am really into: creating a “soul box.” Essentially, a collection of things you return to when you need a lift - a poem or quote, a special song, a photo, a old concert ticket stub, something sentimental or hilarious your kid wrote - anything meaningful, really.

It could be an actual box, a folder on your phone, or a note with a few links. Then when you need a lift, spend a few minutes with it. (Bonus: this can also be something your kids can do!).

What would you put in yours? Even choosing the first three feels like a lovely exercise.

Go outside. Yes, again.

“Spend time in nature” is advice we’ve all heard before.

But a major research review in Nature Human Behaviour, drawing on more than 3,800 studies and 10 million people!! really hammers home the benefits. Researchers found that nature-based activities helped reduce anxiety and depressive symptoms and improve relaxation. There’s still more to learn about which approaches work best, but the evidence gives this very familiar advice serious weight.

Maybe this week, give yourself a little more time outside with no objective other than to relax and be present. Sit near the water. Wander somewhere woodsy. Listen to the birds.

You’re allowed to spend time on something simply because it makes you feel better. 😘

While I know that detailed prompts with lots of context and clear goals gives us much better outputs, I also get soooo tired of putting so much effort into this each and every time.

A hack that really works, especially for bigger projects/tasks: simply ask AI to interview you about the thing you want it to do for you first.

Here’s a fun experiment to try that was shared by one of my favorite AI influencers, AllieKMiller:

Pro tip: you can use the “interview me” approach for any big project or task you want to kick-off with AI as a replacement for a long-winded mega-prompt. e.g. “I want a kick-ass family dashboard. Interview me to understand everything you need to know about what I envision before you begin.” It’s really great.