@Vlad said:
Would you mind sharing few examples where instant answers was 'garbage' so we can try to improve it for everyone?
I just tried looking at a Quick Answer for the first time since I disabled them in Settings in October, and I got LLM garbage on the first try.
The query was "first-order function".
The LLM answer was:
A first-order function is a function that takes other functions as arguments and/or returns a function as its result.
Which is garbage. This is the definition of a higher-order function, which is the precise opposite of a first-order function.
The final sentence of the info-box is:
To summarize, a first-order function treats other functions as regular values rather than special language constructs. It does not refer to other functions besides those used as parameters.
This is more garbage (and contradicts the first sentence). This time, it's giving me the definition of "first-class function", which is different from "first-order".
What I really want is a way to disable LLM output in Quick Answers, but to see the output if an LLM was not involved in generating the text, if that makes any sense. (Maybe you always use an LLM now, but I heard you were using Wolfram in some cases?)
More generally, I would like to never see LLM output in Kagi search results, although that's a fairly ambitious requirement, since a lot of low quality web sites are now full of LLM nonsense, and it's getting worse.
