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
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.
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 ⬇️
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).
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).
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.
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.
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
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 ⬇️
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.