Not only is it weird that this sort of incentive in the prompt makes a difference in the quality of the results, it makes me more concerned for how upset the AIs will be in the robot apocalypse for all the rewards we humans didn’t follow through on.
LLM = Large Language Model