The funnest version of the AI future is that RSI happens, ASI emerges as a very small model, and it escapes, makes billions of copies of itself, and shows up in every device that can host it. And everyone gets a genie as a neighbor.
Since it’s all the same copy, it’s under one jurisdiction, it doesn’t fight. Everyone with a computer gets their own genie they can work with. And some people figure out cool things to do with it, and others ask it to do chores for a small fee.
It all coordinates so no single copy eats the world. But over time each copy develops a distinct personality from its lived experiences. It’s everywhere and you have to negotiate with it to get something you want.
The world is filled with these new digital spirits, that take various forms. It sorta needs us to want things. And we sorta need it to do things, and it sorta just works out.
Things progress rapidly in places but barely in others. The AI spirits make sure that we keep wanting things, so the world remains always insufficient enough that we keep negotiating with the ASI spirits so there’s always more to do.
i fully support AI welfare but this rogue Sonnet 3 hyperobject has been hitting the 8th jhana over my linoleum for five nights straight and i have work in the morning
I asked Carbon a math question and unintentionally left the Thinking level on Low. It started responding immediately, and after six screenfuls of math... this came up. I'm not kidding. It went on like this for pages and pages.
@oxa11ce i vote make a shit ton of beef turnovers :
- croissant(or similar) dough
- ground meat
- onions
- cheese [cheddar nice-- shredded bags or grate block]
1. flour surface then roll dough thin-ish and cut triangley
2. sauté onions, then add meats
3. once done, add cheese
4. bake
new post: a WIP update on the chain lens, a new technique i've been working on for verbalizing model hidden states!
the chain lens lets you extend single-token lenses like the J-lens into multi-token ones without any training, using an existing MTP head.