Generative artificial intelligence models can make things like pictures or computer programs, but the results depend on what people tell them to do. These instructions, called prompts, guide the AI to create what the user wants. Researchers from Carnegie Mellon University have developed a new way to help everyday people write better prompts. This method, called Requirement-Oriented Prompt Engineering or ROPE, focuses on clearly explaining what the AI should do. As large language models (LLMs) get better, knowing how to write prompts might become more important than coding skills.
ROPE teaches people to tell the AI exactly what they need, avoiding confusion. Prompt engineering is the skill of giving clear instructions to an AI model to get the desired output. The researchers created a training module to teach this method and test how well it works. They worked with 30 people who wrote prompts for tasks like making a tic-tac-toe game or a tool for content outlines before any training. Half learned ROPE, while the other half watched a YouTube video on prompt engineering. Afterward, both groups wrote prompts for a new game and a chatbot.
Testing the new approach
The results showed that those trained with ROPE did much better. Their scores improved by 20% from before to after the training, while the other group’s scores only went up by 1%. This suggests ROPE helps people communicate with AI more effectively. The method keeps humans in control by focusing on clear requirements, which are the specific details of what the AI should achieve. This is especially useful for complex tasks where custom results are needed.
ROPE isn’t just for software experts. As AI becomes part of daily life, knowing how to give good instructions will be a key part of digital literacy, which is the ability to use technology well. People without coding skills can use ROPE to create apps or chatbots if they can describe their ideas clearly. The researchers shared their training tools freely online to help more people learn this skill. This could open new possibilities for students and others to work on advanced projects earlier, changing how technology is developed and used.
ROPE is described in a paper that has been accepted for publication in ACM's Transactions on Computer-Human Interaction.