Applying electromotive force to cloud computing. Office Hours: https://xitter.cf/t.co/pwq48UDhVm

Mt. Olympus
Joined June 2019
Igneous, our Python software for managing Neuroglancer 3D images, meshes, and skeletons, is now published open access in Frontiers in Neural Circuits! Aleksandar Zlateski @jalex_bae @tartavull @nkemnitz1 @jingpeng_wu @SebastianSeung frontiersin.org/articles/10.…
4
6
2
28
In developmental economics, a more advanced country, say America, might dump huge quantities of manufactured goods on an underdeveloped country. This is a method of promoting dependence and forestalling the development of indigenous capability.
1
1
54
Afterwards, the targeted underdeveloped country is called backwards and the developed country calls itself their saviors. Meanwhile, the raw materials of the underdeveloped country are extracted at a low cost and often sold back at a value added higher cost.
1
19
This imbalance acts like a vacuum sucking resources from the poor and gives them to the rich, increasing their global power over time. Anyway, I don't know why I brought this example up. There are no obvious parallels.
15
🌩️🌩️🌩️ Thunder Cloud Volume 🌩️🌩️🌩️ retweeted
Get up close with brain cells with the browser-based MICrONS Explorer: bit.ly/3Jh0fnC This OpenScience data portal contains connectivity and functional imaging data including #ElectronMicroscopy based reconstructions of cortical circuitry from the mouse visual cortex.
1
5
8
2,331
Quantity has a quality all it's own. Imagine a single AI had learned from the art world and could make amazing art, but at a rate similar to a human. This would be a marvelous and non-threatening curiosity. AI companies scale this industrially with the aim of replacing humans.
Listen, here’s the thing. A human learning from art is good. But the machine isn’t really “learning”. It’s scraping. It’s downloading and mimicking the work of thousands of artists and doesn’t pay those artists a penny. Does that seem right to you?
1
23
The function and intention, the vile maxim "all for ourselves and nothing for anyone else", the TESCREAL ideology, these are the crimes being committed against the species. Ideas that apply at the scale of a single person do not automatically apply to industrial processes.
5
🌩️🌩️🌩️ Thunder Cloud Volume 🌩️🌩️🌩️ retweeted
Unpopular opinion: AI companies are pushing society to accept an unacceptable level of risk for a non-essential technology.
394
1,358
124
8,560
366,439
When I first heard the words "software factory" I assumed it meant automation producing a lot of software using rented supercomputers. Now I understand "software factory" to mean large numbers of deskilled laborers working without thinking and with little negotiating power.
35
The shining promise of maximalist AI is a worse Plato's Republic where the philosopher-kings have no moral agency because they are not sentient. Also, there are no artists or intellectuals. A box of rocks is telling you what to do, and you have no thoughts.
Why are people so attracted to the idea of not having to think?
1
32
TESCREALists, nerds, and technocrats may think "but the decisions are better!" which is a questionable premise even assuming they are "better" (usually the decisions are better for the people making the decision). Despite the lip service, there is faint understanding of democracy
1
15
Rich people who write like this have invented an imaginary world where we live in socialism and not capitalism. This is because they are the leisure class and will in fact get their time back while demanding everyone else surrender their brain and their time.
If AI gave you back 2 hours a day, that's 730 hours a year. What would you do with them? Spend more time with your kids? Get healthy? Start the business you've been putting off? That's the conversation about the future of tech I want to have. What becomes possible when people get part of their lives back?
1
37
my guess is ai will keep improving until nvidia runs out of tricks AND us models become as efficient as chinese models. its a brute force algorithm on hardware that is just amazing.
This Post is from an account that no longer exists. Learn more
1
96
two things change the game: (a) a move away from silicon to a different computational substrate (b) an algorithmic breakthrough that is more efficient than transformers
1
8
actually I take back (a) because shuffling memory is more expensive than compute, so it would be some breakthrough in RAM bandwidth instead
6