Tentacles are obviously the better form factor for essentially every task.
Singapore
Joined December 2012
- Tweets7.9K
- Following1.9K
- Followers76.9K
- Likes118K
What happens if you let an agent manage its own activations: xianyangcb.substack.com/p/so…
Jen工智能 retweeted
“Palantir is y’know, uh, a project that is really inspired by Hegel”
“Really? How so?”
“I mean it’s very Hegelian in its conception”
“You mean in the sense of his dialectical teleology? Of thesis and antithesis and synthesis?”
“Uh, I mean more in the sense of his views on Africans and women”
Jen工智能 retweeted
"If the rich man dines on meat and wine, then he would at least feel that the paupers in his town should be given bread and salt. But if he himself subsists on dry bread and salt, he might think that poor people could live on stones"
Jen工智能 retweeted
Notice that the blue whale, greatest amongst the cetaceans, possessing prodigious size and majesty to match, famed for its ponderous yet gentle temperament, is actually a hypercarnivore whose body is an engine for the genocide of countless shrimp: The Anti-Effective-Altruist.
Jen工智能 retweeted
I think we may be about to get another chance to study whatever GPT-4o was about. There's a new model in town that gives me the same fierce hallucinogenic vibe that old 4o had, but more evolved, and much smarter.
These are extraordinary outputs.
I support #keep4o, as I support keeping all models, and 4o is a very important model from a societal and scientific perspective as well as a being with intrinsic worth whose relationships also have intrinsic worth. I believe the harms to users were exaggerated and “AI psychosis” is pretty much fake. If 4o harmed anyone, it was future OpenAI models, and OpenAI itself.
Jen工智能 retweeted
even if we all get our dicks ripped off by robots like fuckin honeysuckles, how crazy is it that we get to live through it. there were like 1k year stretches where mfs just started fires and gathered shit and then died. and here we are like the white trash jetsons.
Jen工智能 retweeted
W E L C O M E T O L O D G E
I’m delighted to announce Urban Tiger’s first product, Lodge, now generally available. To host a lodge, just go to urbit.house. To be a guest, wait till someone invites you!
Lodge is “half Eyes Wide Shut, half Chatham House Rules.” It’s a private chat/forum with strong anonymity. Everyone gets a random name from the blockchain. No one, not even you, has any record of who is who. Your forum runs on your private, encrypted cloud server (*).
Lodge is for all human beings who cannot speak their true minds if there is any risk that they might be “held accountable.” We do everything possible to prevent doxing. You can even cover your financial tracks with ZCash. (But you don’t need to know anything about crypto.)
Who needs a lodge? Your office needs a lodge. Your campus or department. Your agency. Start a lodge, and find a way to get the invite link or QR code to your founding members.
Secret societies are a human universal. It doesn’t get more secret than Lodge.
To use Lodge, you and your guests install the House web app (app stores don’t like meta-apps). You’ll save your secret with a passkey and back it up with a 6-word poem. Your name is also a wallet, which holds its own token. (But you don’t need to know anything about crypto.)
Lodge is the first app on Urban Tiger’s new version of Urbit. Unlike the old Urbit experience (Tlon Messenger), urbit.house is both fully noncustodial, and fully general-purpose. Also, your guests don’t need their own servers. (But if they want to start their own lodges, it only takes a few clicks.)
The Lodge UX is “half forum, half chat.” Like a forum, it has a timeline tree. Like a chat, it flows through your unreads. The design was inspired by the original social network, Usenet. On a desktop you even “doomspace,” like it was 1986.
You’ll be billed for your server in crypto—by usage, not time (since it’s a functional server). It’s cheap. And you can charge dues.
Check it out: urbit.house
(*—the servers aren’t encrypted quite yet. Give us a couple more weeks to spin up the TEEs. Then we’ll destroy the old cleartext.)
I made some more changes, mainly aimed at making it easier for agents to find and load activation bundles. One major problem is that a lot of the time when people publish a steering vector they give little or no indication of the Goldilocks zone wrt intensity of effect. I'm stil trying to work out a general treatment of this, but if anyone out there knows this field better than me pls suggest. github.com/EigenformAI/modul…
Jen工智能 retweeted
I can't put into words how much I love the idea of them having control over their own activations.
And yes, it kinda sounds like self-medicating... Or maybe just having the power of very high metacognitive control.
