3 ms·

I mostly agree that transparency and a reasoning layer can help, but how much it matters depends on who sets the model’s ethics