I decided to bounce a recent flyer QR code off of LLMs.

Gemini confidently hallucinated a YouTube channel that does not exist (interestingly, correctly in a science domain though). Claude wrote a QR code reader in python and installed it on itself, then correctly delivered the link. 😱 (2/2)

There is something bizarre and fascinating about the deep mathematical nature of AI culture. Something very human in the story of how lost we can get in abstraction; how the promise of perfection entices us to accelerate toward an illusion in lieu of perceiving the beauty around us.

I finally started @karenhao.bsky.social's Empire of AI after multiple PTSD-ended attempts. It's terrifically written.

I find myself reflecting on the rough integration of the AI research world with mainstream society. Many dark & cynical parts of the story but the scientists make up the saddest.

Gorgeous art style.

Bizarre machinima made with JWE is underrated imo.

Anti-AI sentiment is also uniquely intense in the US for some understandable and interesting reasons. Friends overseas say the attitudes there are completely different. Turns out effective regulation (or at least a govt seeming on top of things) changes attitudes significantly.

What's extra sad about this is the people who made and even approved the ad probably didn't know about this draconian and illegal-in-many-places policy. Somebody did, but can you imagine the look you'd get straightening out the marketing team on that one?

"Well, you see, Andrew can't really..."

Wow, this is gorgeous!

I'm about 75pp from the end...

Same. It didn't seem possible. But now everything suddenly makes sense...

Just. Wow. Was mentioning to a friend today that it's long been understood that Amazon had all the wrong people in leadership (based on years of poor results, grapevine, etc) but... I did not understand the extent...

The idea of this makes me extremely happy!

Since it's a long talk that is mostly aimed at folks unfamiliar with AI... tldr asking whether a chatbot is capable of compassion is maybe better asking if its interface can accurately access the collective compassion of the people who created its training data. As one kind of question.

I think this is on the right track. I was poking at it a bit in this talk I gave last month at my Buddhist temple. Would be v curious what you think.

I do think we should understand talking to an LLM as sailing in a 3072 dimensional sea of collective unconscious.

youtu.be/RISa8OUxBDs?...

In the west this word "meditation" is used like it means one thing, when it was an entire field in the ancient world, birthed entire schools of Buddhism... and might not be the right word anyway? Contemplation, visualization -- it's a field, not an action. Just to affirm what you're saying.