I feel like humans, especially children, are in some respects similar to LLMs or current AI models. With an LLM, if you explicitly tell it not to do something,

I feel like humans, especially children, are in some respects similar to LLMs or current AI models. With an LLM, if you explicitly tell it not to do something, like keeping a secret hidden, it's notoriously eager to disobey the command and do just that. Similarly, if you tell generative AI like DALL-E to not include something, it will try super hard to include it nonetheless. It's as though once the concept is shown, they can't stop thinking about it. Maybe an "out of sight, out of mind" thing would do better than explicitly telling people not to do something?