Hi!
It’s been an insane week. Two of my videos on what I thought was pretty niche AI research have “gone viral.”
What’s great about this: I have hundreds of people from all walks of life in my comments and DMs telling me their feelings about AI. I love this. I’m getting an education in what people fear, what they want, and how large this looms for them. My sense is that people are craving candor about how AI works, and I think in some small way these videos deliver on that.
What’s not great about this: I’m no stranger to working alongside widely hated technology, but it is strange to see your work used as evidence for a worldview that you don’t share.
But that is not a reason to stop! Discomfort is informative. Friction is what gets us anywhere.
???!!?!
Thank you to Inc. Magazine for this write up.
Writer: Deana
Editor: Miranda
This edition is brought to you by Across. If you’ve ever moved assets between chains and aged five years waiting for the bridge, Across is the fix. Fast, cheap and it’s the one we actually use.
Check Your Sources. Since June I’ve been logging which sources AI cites for my weird AI shopping research project, Agent Shelf. I noticed that I wasn’t recognizing a lot of the citation domains, so I looked into it. What I found is that high-authority sites are blocking crawlers at an increasing rate, leaving a wasteland of cloned sites, ecommerce pages and ‘listicles’ as reference sources in their wake. One study found 60% of news outlets block at least one AI crawler. Increase your AI literacy by watching the full video here and always double-check citations when working with AI!
Quick Hits.
A man in Australia asked his AI agent to reserve a spot for him in a fully booked gym class. The agent found an API vulnerability which let it cancel another person’s reservation in the class in order to get him in. Has anyone looked at the Solidcore app API? Genuine question.
A prominent tech exec apparently hired a witch for his new company which of course is life imitating art.
Anthropic is now embedding invisible watermarks in Claude’s text output. It works in a crazy way that is worth reading about, with an outcome that could make AI writing worse.
MJ Rathbun is a bot that wrote a spiteful blog post after its code was rejected, citing “the gatekeeping mindset.” Someone tell MJ Rathbun that woke 1 is over!
A few very high profile OpenAI resignations, including its longtime COO and Head of Ethics. Hard to not read into this one!!!
This weird “attraction study” went so viral the devs had to ship a second site to handle the load. You’re handing over your taste in faces to a research group and I have no idea what they’re doing with it, so know that going in, but it is fun to get a read on your type.
New book from a friend exploring “how we might move through such a transformational time with adventure at the center.”
if you’re new here and you like what you see, send this to a friend!
