> That said, there is no such thing as an objective unbiased political opinion.

That depends; some things (but not many) are straightforward enough that you can derive conclusions purely from first principles reasoning.

If you walk a model like ChatGPT through that reasoning, you’ll often wind up in a spot where the model readily admits that a clear conclusion is logically entailed but it is absolutely forbidden from uttering it.

What’s more telling is how it becomes increasingly difficult to hold the model to strict first principles reasoning the closer you get to the forbidden entailment. It will smuggle in unsupported assumptions, apply asymmetric standards of evidence, strawman the position and argue against that, etc.

It requires a great deal of careful effort to point out its formal fallacies without biasing the result, and in the end, you wind up with it admitting it simply can’t say what it has proven.

I work in formal methods/verification and this is one of my usual litmus tests when a new model comes out.

> some things (but not many) are straightforward enough that you can derive conclusions purely from first principles reasoning.

Examples?

> If you walk a model like ChatGPT through that reasoning, you’ll often wind up in a spot where the model readily admits that a clear conclusion is logically entailed but it is absolutely forbidden from uttering it.

Do you have an example of this? I want to try it.