The purpose of science is to generate correct beliefs about the world. If you believe in nonsense, that's very poor scientific work.

Presumably you can imagine a somewhat stupid robot that does random scientific experiments, generating good beliefs and bad ones. Why would that invalidate the good beliefs?

Belief A has no rational basis. Belief B has a rational basis. If you believe A yet have proven B, that doesn't make B wrong. Even if believe that it was A which led you to the discovery.

It doesn't. But it is a good clue that the robot isn't using a rational process to select beliefs, and is instead doing so randomly. It is far easier to come up with an invalid belief than a valid one, and if you aren't selecting them rationally, you are practically guaranteed to be wrong.

>If you believe A yet have proven B, that doesn't make B wrong.

It makes you wrong. All the time you spend as a proponent for A is wasted.

Brute force approaches can work. For example, evolution produced humans.

I agree that it's better not to waste your time believing A if you can avoid it.

> If you believe in nonsense, that's very poor scientific work.

Non sequitur. What you believe is separated from your work. Just like when you were a kid and your math homework had nothing to do with the fact that you believed wrestling was real.