TypeSafe AI：可靠决策靠代码而非决策模型
so we heard you like decision models
TypeSafe AI 回应“为何不把 Jev 称为决策模型”：面对复杂数据，模型本身无法保证可靠决策，可靠性需要代码，可靠决策来自在代码中组合可靠的判断。其补充说，可以把语义逻辑藏进字符串附加条件、嵌入向量和各种 ML 里，但数据分布总比你能想到的更离奇，因此要明确需要知道什么、并确认直接读取数据就足够。
查看原帖为什么我们 TypeSafe 不把 Jev 称为决策模型呢？ 如果你把 Jev 的问题视为在做出*决策*，我相信你会在错误的抽象层面上使用它，你的可靠性将会受损。 我在我们的网站上详细写了这一点 🧵👇 https://t.co/CfY5W84NqA
查看原文
Why don't we at TypeSafe call Jev a decision model? If you think of Jev's questions as making *decisions*, I believe you'll use it at the wrong layer of abstraction and your reliability will suffer. I wrote all about this on our site 🧵👇 https://t.co/CfY5W84NqA
正文
so we heard you like decision models
but how do you make reliable decisions against gnarly data with a model?
you don't: reliability requires code, and so reliable decisions are made by composing reliable judgments in code
Jev makes reliable judgments