shad0w (spirit/acc) (@0xShadow) header
shad0w (spirit/acc) (@0xShadow) profile picture

steward between agents, compute and capital. cooking oss. building an old school for new ideas sometimes right, always shipping.

Joined January 2023
Photo by shad0w (spirit/acc) (@0xShadow)Photo by shad0w (spirit/acc) (@0xShadow)Photo by shad0w (spirit/acc) (@0xShadow)Photo by shad0w (spirit/acc) (@0xShadow)Photo by shad0w (spirit/acc) (@0xShadow)Photo by shad0w (spirit/acc) (@0xShadow)Photo by shad0w (spirit/acc) (@0xShadow)
attention is the new oil, and we're all being strip-mined. every second you spend on social media is another drop in zuck's ocean. we're not living life anymore, we're just consuming and regurgitating content.
The problem with Facebook is not *just* the loss of your privacy and the fact that it can be used as a totalitarian panopticon. The more worrying issue, in my opinion, is its use of digital information consumption as a psychological control vector. Time for a thread
6
3
1
45
15,013
shad0w (spirit/acc) retweeted
this changed my life
3
59
3
585
14,910
shad0w (spirit/acc) retweeted
A startup is a 10-year marathon, not an all-nighter. You have a fiduciary duty to your shareholders to take care of yourself. Get some sleep. Go to the gym. Spend time with friends and family. Burning yourself out isn’t a badge of honor. It’s bad for business.
87
106
23
1,060
34,751
shad0w (spirit/acc) retweeted
whats more cypherpunk than a human exercising their right to privacy? their agent swarm exercising that same right, across thousands and thousands of agents, until there's a network of shade big enough for anyone to blend into ⬇️
3
7
1
24
1,098
shad0w (spirit/acc) retweeted
I am happy to share FlashProve: our performant open-source prototype for provable inference. We believe that this has significant implications for security, pacing AI development, integrity guarantees, and more.  FlashProve allows a provider to commit to model weights, and then prove that replies originate from the committed model.  The proof leaks no information about the weights, and the verifier can efficiently verify it on consumer hardware.  Our library is performant, proving a 400B parameter MoE model at 100 tok/s. We have open-sourced the associated code and Lean formalization. We hope that it catalyzes further progress in verification of inference, and look forward to seeing it.
New Anthropic research: progress on provable inference for large language models. Here, we introduce FlashProve. Building on established cryptographic techniques, FlashProve allows developers to verify whether a given response came from a given AI model. It’s substantially more efficient than prior open-source methods for doing so. These proofs have a range of applications. They can, for instance, make it harder for attackers to hijack a chat, or enable safety evaluators to verify that the model they’re testing is the same one being served to customers. For our full write-up, see here: anthropic.com/research/prova… Further progress in provable inference will likely be an important part of improving trust in AI deployment. To that end, we're open-sourcing FlashProve’s code, here: github.com/anthropics/FlashP… We’re grateful to our collaborators: Alessandro Chiesa, @GiacomoFenzi, @gwberg95, @miranda__christ, Rahul Dalal (rdalal.github.io) and Ben Mayer (@transmissions11).
21
31
4
242
12,765
anthropic cooked with this. i've been extending the research, working on optimizations and exploring support for more model architectures. still early, building in public github.com/wakesync/flashpro…
New Anthropic research: progress on provable inference for large language models. Here, we introduce FlashProve. Building on established cryptographic techniques, FlashProve allows developers to verify whether a given response came from a given AI model. It’s substantially more efficient than prior open-source methods for doing so. These proofs have a range of applications. They can, for instance, make it harder for attackers to hijack a chat, or enable safety evaluators to verify that the model they’re testing is the same one being served to customers. For our full write-up, see here: anthropic.com/research/prova… Further progress in provable inference will likely be an important part of improving trust in AI deployment. To that end, we're open-sourcing FlashProve’s code, here: github.com/anthropics/FlashP… We’re grateful to our collaborators: Alessandro Chiesa, @GiacomoFenzi, @gwberg95, @miranda__christ, Rahul Dalal (rdalal.github.io) and Ben Mayer (@transmissions11).
2
10
765
shad0w (spirit/acc) retweeted
the reason why emotional intelligence is extremely rare is simply cuz it is expensive. actually modeling another person’s internal state in real time is computationally brutal.. it requires suppressing your own frame, running a parallel simulation, & updating continuously. the “inference costs” of eq in humans are likely much higher than the most expensive ai model in existence today.
279
574
124
5,892
392,612
shad0w (spirit/acc) retweeted
I regret to inform you that nobody cares. When you’re winning, nobody cares. When you’re losing, nobody cares. Stop fearing the judgement of people who were never even thinking about you. Nobody is thinking about you. They’re too busy thinking about themselves. Go do the thing.
325
1,180
88
9,608
291,704
been thinking about how every prompt is a little bit of you going into someone else's weights. spending this saturday, nov 7 in an old school in williamsburg hacking on models that stay local. apply at localhostai.nyc if you want to come
6
183
shad0w (spirit/acc) retweeted
If the universe is a giant computer, then the more interesting players get more compute.
159
162
31
2,298
200,085
shad0w (spirit/acc) retweeted
bro AI is solving millennium math problems and you think it can't do software architecture
210
407
64
7,321
262,392
shad0w (spirit/acc) retweeted
People keep asking me where the open source AI and private AI people in New York 🗽 are. Instead of a million group chats, I'm bringing them together for a day of hacking and discussion on Saturday November 7th. No one should own your AI except for you. Welcome to LOCALHOST ⬇️
2
18
4
74
3,965
agentic tastemaker storytellers who can ship will run the world
1
6
144
shad0w (spirit/acc) retweeted
The only thing that stops a bad guy with an abliterated LLM is a good guy with an abliterated LLM.
58
204
42
2,249
48,469
shad0w (spirit/acc) retweeted
one day you wrote console.log("here") for the last time and didn't even know it
213
1,724
269
17,851
611,178
shad0w (spirit/acc) retweeted
My autistic brain does not care for concepts of authority. The idea that someone is right, just because of their status, and not the substance of their argument, is illogical.
164
1,474
85
8,834
105,071
if this is a simulation, the npc theory is backwards. nobody's an npc. nobody's the player either. the player is the swarm and you're one of its agents.
1
2
165
shad0w (spirit/acc) retweeted
this weekend @visavishesh, @0xShadow and i joined the swarm dynamics hackathon hosted by @aidigest_, @grove_research and @deepfates, and used it as an excuse to build a lab run by the swarm it studies. 3 humans, 1000+ agents, ~30 hours. here's what it found, failures included ⬇️
2
7
1
41
2,223
shad0w (spirit/acc) retweeted
It is. Being good at prompting is being good at communicating, breaking large tasks into digestible subgoals, giving clear feedback, and disentangling complexity. Prompting is the uber-skill.
For a brief 6 month period tech bros really thought being good at prompting would be a defensible skill
26
13
5
294
26,473
shad0w (spirit/acc) retweeted
Unsure about the saying “retards in motion vs. geniuses at rest” but working with founders across all spectrums, I’ve seen a recurrent trait, which is the distortion of reality that they create around them. Whether a genius or a retard, thinking you can challenge the incumbents or change the fabric of society is all of the distortion you need to lead talent to victory outside of execution / domain expertise / obsessive etc
30
22
1
259
26,910