Big rocket fly soon?
This is a good bit, but it's unironically true atm- you can flip those gpt detectors from "AI generated" to "definitely human" with a single typo most of the time
Yeah, AI gives you an API to human systems, and they're not used to standing up to those kinds of attacks. It's super fun, but, yknow, chaos
daily reminder to carry an IFAK- bleeding out from a potentially treatable extremity wound is cringe, m'kay
not sure if nostr likes YouTube links, but this by PrepMedic is a real good build guide: https://www.youtube.com/watch?v=x_--5fuM1Dc
โAt least once every human should have to run for his life, to teach him that milk does not come from supermarkets, that safety does not come from policemen, that 'news' is not something that happens to other people. He might learn how his ancestors lived and that he himself is no different--in the crunch his life depends on his agility, alertness, and personal resourcefulness.โ
โ Robert A. Heinlein
Okay, awesome, this worked well!
Thought lifted from Craig Douglas: wrestling and BJJ are the delivery system. Submissions are one of several possible payloads.
Okay, awesome, thanks! That makes a lot more sense.
Spending time on the mats has a way of making you both confident and aware of your vulnerability at the same time. I know what I can do and what I can handle, but also know the 60 year old brown belt could maim me for life if he wanted to
Okay, good, there's key-convertr, at least,
ty! had forgotten to set that up yesterday, done now
yeah- the word sovereign gets overused a bit, but the way I think of it, your computing devices are increasingly part of your brain, your exocortex - you share so much information with them that they're a part of you
having that part of you working for someone else, not having sovereignty over it, is anincreasingly dangeorus and unpleasant
okay, now trying out the iris.to web client
I've got to say, at this point seeing a box with "paste your private key here" on a website is, uh, a little disconcerting
but I guess that's how this works?
Really cool stuff! Also interested to see how this meshes with robotics- I know some groups are using LLMs for task planning now, but I don't know if they've got end to end execution really working yet
Jet fuel for the Great Weirding :)
Probably no apocalyptic foom, but we're just getting started integrating LLMs into feedback loops/agent models. It turns out that language token prediction is a really good trick for dealing with fuzzy inputs/world models, far beyond just generating text. Text alone is pretty disruptive to work, but agents that can be delegated tasks will be more so, I think.
Trying out nostr and bluesky in the same weekend has been interesting - different feature sets, different communities, the same impulse to exit
Haha, will do. Still getting set up on nostr today!
Euler-Lagrange, I think- there are a few fundamental secrets of the universe, and least action really feels like one of them
so how does this work