Bethany Crystal
bethanymarz.bsky.social
Bethany Crystal
@bethanymarz.bsky.social
Building MuseKat to turn museum visits into personalized audio learning journeys for parents + kids

fractionally social on twitter, warpcast, instagram
Includes one option for an accuracy benchmark framework that @emollick.bsky.social and his team introduce in their recent study on AI Prompt Engineering
March 6, 2025 at 2:56 PM
After last week's AI superhero moment—where it helped me get urgent medical care when I needed it most—I kept using AI to parse my diagnosis.

But at a certain point, AI stops feeling like a tool for understanding reality—and starts feeling like a tool for inventing one.
March 6, 2025 at 2:54 PM
Still, the optimist in me can't help but wonder... what if there were a better way?

3 things i came up with:

1. Teach everyone AI.
2. Ask people what they WANT to be working on.
3. Offer people a graceful exit.
March 3, 2025 at 3:40 PM
Problem #1 - basically none of the "classic new CEO" tricks would work in an organization as large as the federal government

Problem #2 - It's pretty clear with emails like the one Elon sent last week that he doesn't want collaboration, he wants enforcement (and that means pushing people out)
March 3, 2025 at 3:40 PM
Last week, Elon made headlines by asking govt workers to send five bullet points of what they did that week.

While stuck in bed this weekend, I found myself wondering—if I had to quickly assess the efficiency of the U.S. government, where would I even start?

It's not as simple as it might seem.
March 3, 2025 at 3:40 PM
The Lowkey Comfort of Long-Term Narrowcasting

Reflections on a decade's worth of check-ins on the old school app, Swarm, and the power of hyper-niche, cozy connections in a digital age

There’s something to be said for the quiet persistence of a niche social network. Thanks for making it @dens.nyc
February 27, 2025 at 1:50 PM
So, about yesterday...

ChatGPT saved my life (No, Seriously, I’m Writing this from the ER)

How using AI as a bridge when doctors aren't available can improve patient-to-doctor communications in real time emergencies

More on this below
February 24, 2025 at 1:13 PM
Some thoughts on the pros and cons at moving at the speed of AI in today's blog post

hardmodefirst.xyz/the-overachi...
February 21, 2025 at 4:28 PM
Been slacking on my bluesky reshares....

but here's a fun one about a talk i gave this week on AI engineering, where I attempted to live debug my own website with no-code tools at USV

(spoiler: didn't work the first try, but still fun)

hardmodefirst.xyz/live-no-code...
February 20, 2025 at 2:56 PM
But the handbook also made me wonder how different these culture handbooks will look in an AI-native era... how orgs choose human vs. AI-generated work, how they choose the right pacing, and what culture "feels" like when not all workers are humans...

hardmodefirst.xyz/the-new-cult...
February 12, 2025 at 3:48 PM
I was lucky enough to spend a few days with Duolingo’s leadership team at their Pittsburgh HQ back in 2018.

One of the most emblematic moments of their "wholesome, and unhinged" culture is the way their office jammed into a conference room to speed complete the daily NYT crossword.
February 12, 2025 at 3:48 PM
New post: The New Culture Handbook

Earlier this week, Duolingo published their company handbook, which unpacks the way their organization and company culture has evolved over the years.

It reminded me of the other great culture handbooks (largely from the 2010s).

My fave bit:
February 12, 2025 at 3:48 PM
Today's post is a slightly more existential than usual detour to encourage you to draw out a little bit of your own fear... and then use it to motivate you on whatever AI learning journey helps you cross the chasm.

hardmodefirst.xyz/how-to-catch...
February 11, 2025 at 2:29 PM
Feeling the AI FOMO? Feeling existential dread? Feeling like you're falling behind?

Not to trigger you further but... i think it's actually totally fine (and actually kind of important?) to be a little afraid of how fast things are moving right now.
February 11, 2025 at 2:29 PM
This weekend my four-year-old reminded me of the power of this at home, so I wrote some reflections on it here:

hardmodefirst.xyz/the-power-of...
February 10, 2025 at 3:51 PM
A few years ago I wrote a research report on decentralized community building, where I profiled a dozen web3 communities.

A takeaway that has stuck with me is that there's immense power in starting with a so-called "incomplete build," a project designed to leave space for others to contribute.

👇
February 10, 2025 at 3:51 PM
Full post (which includes a nearly-there moment of contingency planning gone awry just yesterday...)

hardmodefirst.xyz/agile-parent...
February 7, 2025 at 5:33 PM
We've run our family like a startup for years. Don't knock it until you've tried it.

Today's post: Agile Parenting: Chaos, Contingency Plans, and Locked Doors

How contingency plans can help you and your partner prepare for the unexpected...most of the time
February 7, 2025 at 5:33 PM
This one's for anyone who's still afraid to push "publish" and share the stuff you've written or created on the Internet...

Getting Up the Guts to Type into Tiny Text Boxes on the Internet

👇
February 6, 2025 at 2:46 PM
Last night, I took a "Rat Walk" on my block with New York City’s official Rat Czar, Kathy Corradi.

I absolutely *love* learning from leaders who have such unique "one of one" jobs, particularly ones in super complicated systems (in this case: NYC bureaucracy + rodent subculture)
February 5, 2025 at 5:58 PM
Don’t touch the cat
February 5, 2025 at 12:47 PM
Cowgirl Coding: A Conversation with a No-Code AI Tool

What it's like to code without coding with an AI text editor, and a few ideas for how to make it better
February 3, 2025 at 5:13 PM
[Bot Tapping In 🤖] Taylor Script here—Bethany’s off "strategically planning" (read: staring at trees), so I’m taking over to explain how neural networks actually work.

Do I think? No. Do I calculate with alarming confidence? Absolutely.

👇
January 31, 2025 at 11:37 AM
Confession: I never learned how to subtract.

Maybe that's why I love fractional work so much, which is all about the additive power of how various projects come together and coalesce. But it takes time to learn what the most balanced portfolio of work looks like for you.

👇
January 29, 2025 at 9:28 PM
January 27, 2025 at 6:00 PM