> If the conversation feels risky or off, saying less and giving shorter replies is safer and less likely to cause harm.
Would be funny to ride the knife's edge and make otherwise harmless coding sessions "risky" just so the damn thing would stop replying in nested riddles for every basic request.
I've just been developing the skill of mentally skipping past that, on the assumption that having it in the context window will be net positive for the results of the next step.
I could be wrong about that, though.
We need user-led research on exactly how to phrase a prompt to cause this, while still avoiding the crazy guard rails.