When playing LLMs, it is tempting to only share and discuss the prompts that cause the model to return something interesting that demonstrates the models reasoning capabilities or the knowledge it has memorized.

For me, sometimes those prompts come after a lot of iteration. Less often, they are the first thing I write after the general idea pops into my head.

Most of the lessons in prompt design, and more generally systems design, come from the iterative process. The failures are just as important as the successes. They are the things that didn’t work, and the things that didn’t work are the things that we can learn from.