Carlos
@lospoy
12 Following
3 Followers
Casts
Casts + replies
Carlos
@lospoy
Is there a way of finding the ppl I follow on X but here? 👀 (not manually searching)
0 reply
0 recast
0 reaction
Carlos
@lospoy
Like, literally sometimes 😂
0 reply
0 recast
0 reaction
Carlos
@lospoy
😂😂😂😂
0 reply
0 recast
0 reaction
Sophia
@sophia
celo is LIVE on the farcaster wallet 🥳
16 replies
21 recasts
112 reactions
Carlos
@lospoy
Lfg! 👏👏👏👏
0 reply
0 recast
1 reaction
Carlos
@lospoy
Wtf 😂😂😂
0 reply
0 recast
0 reaction
Carlos
@lospoy
Probabilistic model attestation could be foundational to decentralized AI. Whitepaper: https://arxiv.org/abs/2504.13443 DMs open if you’re building in this space. – Special thanks to Michael Yuan for the opportunity to collaborate and James Snewin from for their insightful feedback
0 reply
0 recast
0 reaction
Carlos
@lospoy
This opens the door to decentralized AI networks that are actually verifiable. - No need for TEEs - No need for zero-knowledge proofs Just input/output analysis and crypto-economic incentives to keep nodes honest.
1 reply
0 recast
0 reaction
Carlos
@lospoy
Different models (even subtly different ones) leave statistical fingerprints in their outputs. By probing a node with known inputs, you can probabilistically verify what model it’s running –just from an API output :)
1 reply
0 recast
0 reaction
Carlos
@lospoy
The problem: In decentralized AI networks, how do you know a node is running the right model? Most proposals rely on trusted hardware or expensive cryptography. We propose something simpler and more scalable:
1 reply
0 recast
0 reaction
Carlos
@lospoy
Just co-authored a whitepaper on a new way to verify AI model execution –without TEEs, ZK proofs, or trusted parties. It’s a mechanism for trustless AI inference in decentralized networks 👇
1 reply
0 recast
0 reaction
Carlos
@lospoy
this is next level PMF
0 reply
0 recast
0 reaction
Carlos
@lospoy
AI models are getting better at sounding right –but how do we verify they *are* right? Been working on something with Mira and @gaianet. Early thoughts here: https://medium.com/@lospoy/how-to-trust-ai-output-validation-b92deb0d8cd6
0 reply
0 recast
0 reaction
Matt
@mane
Carpe Diem
0 reply
2 recasts
2 reactions
Carlos
@lospoy
Posting, posting. Wonder if there's some kind of soft lock.
0 reply
0 recast
0 reaction
Carlos
@lospoy
just waking up now
0 reply
0 recast
0 reaction
