> I don’t think there’s necessarily anything wrong with OpenAI wanting to keep their chains-of-thought secret.
I do. Besides the fact that it is kind of the opposite of "open" and is another sign that OpenAI is an inapt name for the company, I think being able to inspect the reasoning is often useful. It gives me a better idea of how much to trust the final response, and if the response is wrong, sometimes gives me hints on how to get to the right answer.
> I don’t think there’s necessarily anything wrong with OpenAI wanting to keep their chains-of-thought secret.
I do. Besides the fact that it is kind of the opposite of "open" and is another sign that OpenAI is an inapt name for the company, I think being able to inspect the reasoning is often useful. It gives me a better idea of how much to trust the final response, and if the response is wrong, sometimes gives me hints on how to get to the right answer.