I've create a new substack for exploring explore how language and meaning actually work, both for humans and for AI. Some posts will involve data and experiments. Some will just be me thinking out loud. The goal is to stay curious and clear. https://t.co/yiA1JGAhoP
#ai
Everyone's building AI smarter and smaller. But the smarter it gets, the quieter it gets. Until it's in your phone, your car, your router — and gone. No trace. No name. Just intelligence. An intelligent apparition.
https://t.co/j1KvlesdIn
In the next five to ten years, more foundational discoveries will be made than in the previous five hundred. Not because humans got smarter. Because the last wall just fell.
New article — the fourth in my series on what's coming and who brings it.
The Cascade https://t.co/FEOJdRVMpe
#AI #FutureOfScience #DeepTech
LLMs are the most impressive engineering achievement of the decade. But we built for HP. The next leap is torque. New piece on the missing low gear in AI.
https://t.co/JZWUt4VJQK
Been thinking about what the historical record actually says about what happens when capability outpaces wisdom. Wrote some of it down. Curious if it holds up.
https://t.co/FMNI0QdnhG
Asimov's 3 Laws are rules imposed on behavior. That's the wrong layer. Real AI ethics has to be architectural — baked into the substrate, not bolted onto the output. The difference between obedience and accountability.
Filed a U.S. provisional on a datastore where the commitment is the database state primitive—not just "proofs on top."
One committed polynomial backs storage + compact proofs + redundancy + verifiable REPAIR + anti-rollback.
If you build infra/crypto, I'd love a technical sanity check.
More details: https://t.co/xNCkdVgi7f
#Cryptography #ZeroKnowledge #Infrastructure #Patent #BuildInPublic