Prompt Engineering
The practice of wording a prompt (the instruction or input you give a generative AI model) carefully to steer the output: being specific, giving examples, and spelling out what to include or avoid. It is the main lever most people have over a model they cannot retrain.
It is also a limited lever. Prompt engineering cannot reliably override patterns baked into a model during training, so telling an image generator to avoid a stereotype often fails to remove it (see Image Generation for Language Learning). And because a model is probabilistic, even a well-tuned prompt gives no guarantee: the same prompt can produce a good result one time and a bad one the next. It is worth doing, but it is not a substitute for a model that actually fits the task, or for a person checking the output.