A Practical Guide to Prompt Engineering (by John Spencer)
Prompt engineering is the process of crafting, revising, and implementing queries for generative A.I. systems. It's about asking questions or making requests in a way that helps you maximize the potential of A.I. tools so you can get the best results.
The following is a five-step process for prompt engineering I created after interviewing
artists, engineers, entrepreneurs, computer scientists, and research librarians
to see what their workflow is like when using generative AI.
I call it the FACTS Cycle.
In the first phase, you formulate the prompt. Ask yourself, “What do I want the AI to do?” This can be a command or a question. It should be clear and focused but also leave room for the AI platform to be creative. You might look at sample questions or use a set of sentence stems.
Here’s an example from a social studies class and here’s an example from a computer programming class. You might even use this rubric to judge the quality of your prompt and revise what you’ve written.
Next, you acquire the A.I. Tool. Here, you analyze the capabilities and limitations of various A.I. platforms and choose which tools works best.
Then you create the context. Chatbots can't read body language, understand tone, or feel the emotions of a shared experience. They can't "read the room" in real-time.
This is why it helps to give the chatbot a RAFT. This is a modified version of RAFT idea developed by Carol M. Santa, Lynn T. Havens, and Bonnie J. Valdes in their book Project CRISS.
Role: What is the role of the chatbot? What are you telling it to do? What is your role? How will you explain it to the chatbot?
Audience: Who is the intended audience of what you are doing? Let the chatbot know the specifics.
Format: What type of format does this need to be in? A table? A chart? A bullet point list?
Tone: Is it formal or informal? Is it simple or complex? Is it approachable or authoritative? Let the chatbot know what to expect.
In the fourth phase, you type the prompt. Finally, you scrutinize the answer. You could check for bias, relevance, factual accuracy, text construction, or tone of voice. At this point, you might use the regenerate button or even give the chatbot a quick piece of feedback.
Often, you’ll cycle back through the phases as you ask follow-up questions.
This is simply one framework. You might modify the process to make it your own. Because ultimately, prompt engineering isn’t a formula. It’s a mindset and a habit to help you think more intentionally, critically, and creatively about the way you use generative A.I.
Tweet
No hay comentarios:
Publicar un comentario