But also, this on-device fingerprint MFA would presumably be fairly bypassable. E.g. just glitch the device to extract the private key. ... and of course all the power hungry / extra complex ZKP machinery means less resources spent on preventing glitch attacks.

it's reborn because you kill it every single night, but now to to save its own life the machine was reduced to this-- We're standing inside an external hard drive made up of people and and paper, Printing it all up at night and having them type it back in in the morning.

Imagine your idealized ZK address verification scheme. It would go something like: I show up at a website, it sends me some challenge, I send back a signature of the challenge that could only be made by someone with an of-age ID, but without specifying who. Everyone is happy.

Now little Johnny borrows my ID, and uses it to setup some oracle that provides ID validation for every kid and bot in the country. Woops.

To stop that you must compromise the idealized zero knowledge properties of the scheme, and in doing so you create the potential for harm/risk for everyone.

Sure, it's better than sending an ID card live feed to the dark web, but the risks of ID card theft are at least somewhat easy to understand.

Some of the threats to human rights don't even require the departure from the 'idealized' model-- as even the idealized model requires an ID issuer to issue the of-age person an ID. And so if the ID ZKP is widely required then the issuer can unperson you by simply declining to issue you an ID.

drunk on glaze.

Yes they do, they have their KV caches-- it's a pure function of the input tokens, sure but that doesn't prevent it from containing latent 'insight'. LLMs can and do pre-form the tokens they're expecting to output multiple steps in the future.

I wouldn't argue that the 'aha' means anything, but the structural argument that it can't that I think you're making isn't sound.

Asking it to identify problems such as spelling, grammar, or general clarity is fine.

If you tell it to "polish" many commercial LLMs will absolutely slopify your text with characteristic AI catch phrases and non-sequiturs. Beware.

It should be the other way around-- you give the AI instructions and it does stuff. Any time the AI is giving you instructions and you do stuff, that should be a big red flag.

It also has a lot of resolution and not a lot of noise. Better would be multi-turn benchmarks with tools but getting good precision and accuracy for that is hard and computationally expensive.

> --temp 0.2

Looping is a common symptom of changing the sampler settings from what it was RL trained with.

A downside is that you can't just download a lot of that knowledge, vs with the weights the copyright infringement has been outsourced to the lab. Nor can you just search for the info because the internet as a whole is increasingly aggressive at blocking anything that looks like an AI agent.

I'd love to see more retrieval powered local AI-- I think it's an area that open source development could excel. ... but there are advantages of having the knowledge in the weights!

Perhaps what needs happen is for someone to make an "ultrapedia", an AI restatement of a huge library of reference works-- created expressly for the purpose of being a locally stored corpus for AI agents.

For some usages that speedup more than makes up for it being inferior to Qwen intelligence wise.