Introducing System One Models and Jev • Hacker News Board

Re: Introducing System One Models and Jev

#481

Post by sonink » Wed, Sep 16, 2026, 2:46 PM UTC

Spent a lot of time - but this makes zero sense to me. It can, maybe, return type safe outputs faster than larger llms - but there is little reason to believe that it will be more accurate. It does absolutely hallucinate - and seems to me that the claim is largely misleading.

You architect your systems with typesafe - because it is marginally faster, but inaccurate - to do what ? You can just wait for the next version of LLM's to get more accuracy at the same cost - or just use a faster model right now from a different provider.

Re: Introducing System One Models and Jev

#482

Post by djray » Wed, Sep 16, 2026, 2:47 PM UTC

bigglebear wrote: post #247

CompleteSkeptic wrote: post #141

Earlier quoted context omitted.

> Type safety is not factual correctness. I very much agree with this and want to hone in on where do actually disagree. Would you say a linear classifier hallucinates?

User input: "Hey, have your human support agent call me, tomorrow at 5pm." Model input: "Does the user want to speak to a human support agent?" Output: Yes. I imagine that your model would produce this, and I think it's fair to say this is a hallucination. A human would caveat it with: "Yes, but not right now.", your model is incapable of that. Yes is technically correct, but within the context of being in a live cha…

No, the model has answered correctly. Your question is poorly phrased (possibly deliberately).

Your question would correctly classify the user's input as requesting a human support agent, but at an indeterminate time.

If you wanted to determine whether the user wants to speak to a human support agent immediately, you would have to correctly qualify your question, e.g. "Does the user want to speak to a human support agent now?". You could have another question which is "Is the user requesting a call-back from a human support agent?". Or you could have a multiple choice query which would filter the conversation into one of a number of pre-written possibilities.

This is nothing to do with accuracy or hallucination. It's a different method of interacting with the model where you are relied upon to be precise.

Re: Introducing System One Models and Jev

#483

Post by gogoout » Wed, Sep 16, 2026, 2:50 PM UTC

Is this the reverse of LLM? Ie, "what's the capital of France?" LLM picks from "Paris" 99.9% / "London" 0.001%, LLM then with some randomness output you "Paris". For Jev, you ask it to give probability of a set of answers "what's the capital of France?" choose from answers (Paris/London), it then gives you (99.9%/0.001%)

Re: Introducing System One Models and Jev

#484

Post by sva_ » Wed, Sep 16, 2026, 2:53 PM UTC

Someone else noticed the base64 encoded block on the launch site[0]? Its the fast inverse square root algorithm q_rsqrt[1]. I guess its meant as a joke to put this algorithm that makes use of type punning on a site called typesafe.ai? Or maybe because of its efficiency?

0. https://typesafe.ai/

1. https://en.wikipedia.org/wiki/Fast_inverse_square_root

Re: Introducing System One Models and Jev

#485

Post by aatd86 » Wed, Sep 16, 2026, 2:55 PM UTC

gogoout wrote: post #483

Is this the reverse of LLM? Ie, "what's the capital of France?" LLM picks from "Paris" 99.9% / "London" 0.001%, LLM then with some randomness output you "Paris". For Jev, you ask it to give probability of a set of answers "what's the capital of France?" choose from answers (Paris/London), it then gives you (99.9%/0.001%)

99.90% ;D It basically gives you the softmax value instead of using it to generate the output token. (said loosely)

Re: Introducing System One Models and Jev

#486

Post by saldubai » Wed, Sep 16, 2026, 2:58 PM UTC

When we speak about good judgement in models and agents we are talking about humans skills such as critical thinking , judgement and decision making , emotional iq, mindfulness etc. We have been building since 2018 a structured good judgement data lake tied to ten core humans skills and sub skills levelled queaisn and answers against blooms taxonomy from a tagged community of experts , outliers , contrarians . Open ended situational questions and answers like how the real world operates in various sectors to capture tacit knowledge . We have a playground with a tiny slice of just 100 of our over 500k base good judgement scenarios that can then be synthesized across sectors and workflows . With just a tiny slice it outperforms Fable and Open Ai models. Exceptional human judgement outperforms consistently models…. Good judgement is just also rare in us humans like common sense ( no pun intended) . Check us out at lovelyhumans.ai. Holler if curious. Sallyann Dellacasa on LinkedIn .

Re: Introducing System One Models and Jev

#487

Post by jw1224 » Wed, Sep 16, 2026, 3:08 PM UTC

ursuscamp wrote: post #480

Is this fundamentally different from other text-based LLMs, or is it the same except with special reinforcement learning a safe guards around generating valid types? Surely it’s still generating some kind unstructured data internally? For example, what if I told it to generate a short story, but the short story is output as a JSON string?

An LLM takes (text in) -> (text out).

Jev takes (…questions in) -> (…probabilities out)

So Jev won’t write a story or emit arbitrary structured data. But if you ask it the right questions, it can make near-instant “decisions” against those questions, with accuracy and world knowledge on par with LLMs. The economic advantage is that it’s parallelizable and can give back up to 255 answers at once, in milliseconds.

Re: Introducing System One Models and Jev

#488

Post by yymir » Wed, Sep 16, 2026, 3:19 PM UTC

This could be killer for ingame AI for grand strategy games like Victoria 3 or EU5

Re: Introducing System One Models and Jev

#489

Post by brooksy » Wed, Sep 16, 2026, 3:24 PM UTC

gogoout wrote: post #483

Is this the reverse of LLM? Ie, "what's the capital of France?" LLM picks from "Paris" 99.9% / "London" 0.001%, LLM then with some randomness output you "Paris". For Jev, you ask it to give probability of a set of answers "what's the capital of France?" choose from answers (Paris/London), it then gives you (99.9%/0.001%)

Essentially, although LLM token probabilities tend to be miscalibrated (mostly bc of posttraining). Jev is meant to be particularly calibrated

Re: Introducing System One Models and Jev

#490

Post by brooksy » Wed, Sep 16, 2026, 3:29 PM UTC

sonink wrote: post #481

Spent a lot of time - but this makes zero sense to me. It can, maybe, return type safe outputs faster than larger llms - but there is little reason to believe that it will be more accurate. It does absolutely hallucinate - and seems to me that the claim is largely misleading. You architect your systems with typesafe - because it is marginally faster, but inaccurate - to do what ? You can just wait for the next versio…

For certain tasks, a model like Jev may be intrinsically more efficient than an LLM because it doesn't have to predict a token distribution and can instead focus solely on the probability of a single question/action