Jev vs GPT-4.1 on a synthetic survey

runs Jev and GPT-4.1 as the same 300 synthetic respondents over 24,596 paired Twin-2K-500 cells under criteria fixed in advance, finding that asking a yes/no item as `Noul` rather than `Choice` moves the result more than the gap between the two models, at a thirty-fourth of the cost. Write-up: [jjd-lab.github.io](https://jjd-lab.github.io/jev-synthetic-survey/)

NOUL 调用长什么样

这是该提问类型的通用骨架，不是这个项目的真实代码。

from typesafe import TypeSafe

ts = TypeSafe()
result = ts.evaluate(
    state=tool_call,
    questions={
        "is_risky": {
            "type": "noul",
            "instructions": "Could this call delete or overwrite user data?",
        }
    },
)
if result["is_risky"] and result["is_risky_confidence"] > 0.6:
    escalate_to_human(tool_call)