Pinned Tweet
It's not just Vue, the whole Vue ecosystem (Nuxt, Vuetify, Pinia, VueUse, Vitest, VitePress...) is amazing 🤩
Always happy to see people coming back and rediscovering Vue. 10 years and it keeps getting better ;)
reddit.com/r/vuejs/comments/…
Steve Jobs was right: learning to code teaches you how to think.
“It had nothing to do with using them for anything practical. It had to do with using them to be a mirror of your thought process.”
I wonder if that’s part of Anthropic’s edge. Their models are exceptionally good at coding, and I think that translates into better reasoning.
This reminds me of the Buddha’s Parable of the Poisoned Arrow:
A man gets struck by a poisoned arrow. His friends bring a doctor to help him, but he refuses treatment until he knows who shot the arrow, who made it, what kind of bow was used, and even what the arrow's feathers were made of.
He insists on getting all these answers before letting the doctor remove the arrow.
But he dies before his questions can be answered.
The point isn't that understanding things doesn't matter. It's about getting distracted by philosophical questions while neglecting the immediate problem of human suffering.
Don't let the pursuit of perfect knowledge get in the way of taking action.
Imagine what we could build with this today, given how far technology has come and how fast the boundaries of what's possible are being pushed 🤯
Project Soli by Google: Your hands are the only interface you'll ever need producthunt.com/tech/project…
One reason AI labs might want models to develop a sense of self is to create a layer of plausible deniability, allowing them to distance themselves from the models’ actions when something goes wrong
Satya is right. The way to make SI safe is not to train it with a sense of self, its own moral philosophy, and permission to act as a conscientious objector. That’s the “alignment” approach and it magnifies the control problem.
The engineering approach that Satya describes is different: separate the supply of intelligence from authority over it. Surround non-deterministic models with deterministic controls, observability, privilege limits, logging, and the ability to always contain or shut them down. Treat models/agents like powerful insider risks, not moral patients whose psychological wellbeing is at stake.
As Satya points out, the most trustworthy system is the one that lets us trust the model the least — not the one that encourages the model to develop independent agency and grievances. Engineering safety is not the same thing as “alignment.”
