given recent events, re-sharing my thoughts on space datacenters :p
⚡️🌙@dystopiabreaker.xyz · 7mo agoenergy is more abundant (higher solar efficiency) in theory, it is operationally simpler (you launch swarms of modules precisely sized for the radiative heat envelope, that comm over optical links, space is big, no coordination problem)
We are starting a new, nonprofit alignment organization, ⊢ Sequent Research, bringing together researchers previously on UK AISI’s Alignment Team, Timaeus, and elsewhere to research how to align superintelligence. We are hiring! 🧵 sequent.org/launch
Don’t Worry — It Can’t Happen (also, the scientists who claim fission exists are just in the pocket of Big Science, and they’re literal nazis anyway, and also it’s just a stochastic reaction that peters out, and only physbros care about it, and it hasn’t ever happened before so it won’t)
Wow this farmer has a message for AI companies who want to buy his land
miss this discourse tbh. it was very funny in this specific way
⚡️🌙@dystopiabreaker.xyz · 10mo agoi’m completely serious when i say that much of the dismissive ai discourse on here fires the bigotry neuron
🔭 Saturn at Night Image Credit: NASA, JPL-Caltech, Space Science Institute, Mindaugas Macijauskas ap260601
i've been doing this increasingly in the past few months where I have time because of the urgency (esp security automation things), and now it looks like i'll finally be able to apply the amount of focus that i want to to this area now. very excited
⚡️🌙@dystopiabreaker.xyz · 6mo agoi want to work on interpretability, red teaming, and related security areas
one of my favorite cheeky papers, from 1999, is this one where the authors argue that sometimes it is better to simply wait and do nothing, because your astrophysical simulations will complete faster if you simply wait for the next epoch of compute arxiv.org/pdf/astro-ph...
found an old project from a laptop from high school where i had just learned about neural networks and was trying to implement an autonomous network security agent named 't-cell' by carving out a static set of decision classes from a static network manifest input lol. raw numpy, didnt get very far
how many layers of negative polarization are you on? you are like baby, watch this,
Anil Dash@anildash.com · 3mo agoThe AI of drugs. The same tech financiers have been saying for years that they want GLP-1s in the water supply.
everything is a bytecode-targetable virtual machine if you look at it sideways. JBIG2 image decompression? virtual machine. x86 mov instruction? virtual machine. truetype fonts? virtual machine. Magic: The Gathering? believe it or not, also virtual machine.
average MLeng: 'well, the adversary would have to know the details of how our defense works' the humble Kerckhoff axiom:
neural networks try to design towards “all structure” in their layers, whereas block ciphers must be “no structure at all” (any structure is a break, usually)
Everything MPU posts about data centers is complete garbage. They have zero respect for their audience. Literally no one here is losing power. This post is a complete lie. 🧵
More Perfect Union@moreperfectunion.bsky.social · 3mo agoNearly 50,000 people in the Lake Tahoe area have been told that their utility will stop providing power to them, because it's redirecting that power to data centers. /1
the vast majority of practical traditional software security problems are strongly 'on distribution'; detecting them at a high level is basically an n-classification problem, where n is actually quite small, and there is a _ton_ of shitty code out there.
i think i was mostly right about this but too unkind in the way i communicated it
⚡️🌙@dystopiabreaker.xyz · 3mo agothis is completely and utterly wrong. a close friend worked at twitch during the early days and showed me impromptu GPU farms for handling the codec load. Google has designed a custom ASIC just for youtube video codecs.