[Question] What are some good examples of gaming that is hard to detect?

[SoerenMind]

For example, an RL agent that learns a policy that looks good to humans but isn’t. Adversarial examples that only fool a neural nets wouldn’t count.

[SoerenMind]

For example, an RL agent that learns a policy that looks good to humans but isn’t. Adversarial examples that only fool a neural nets wouldn’t count.

[habryka]

Could you clarify this a bit? I assume you are thinking about subsets of specification gaming that would not be obvious if they were happening?

If so, then I guess all the adversarial examples in image classification comes to mind, which fits specification gaming pretty well and required quite a large literature to understand.

[SoerenMind]

Thanks, updated.