A prompt is the text instruction you give an AI model. It is the single biggest factor in the quality of what you get back.

To use AI well, start with one core unit: the token. A token is a small chunk of text, often a word or part of a word, that the model reads and writes. Models do their work in tokens, and every token of your input and every token of the answer takes computation.

That computation is called inference: the work the model does to turn your prompt into a response. Inference runs on servers that draw electricity, and many data centers use water to cool that hardware. More tokens means more inference, which means more energy and water.

One honest note. You may have seen exact figures for the water or energy used per AI query. Most of those numbers are unreliable or out of date, so we will not lean on them. The mechanism is what matters: fewer wasted tokens means less wasted compute.