Brandon Yuen

You can get much better results out of ChatGPT by forcing it to go through a step-by-step process. An example: ChatGPT is generally really bad at creating interesting puzzles and scenarios to solve, either making things too easy or impossible. But step-by-step approaches work. Prompt: "You a game master. Your job is to come up with interesting challenges for the player to solve. Describe a challenging fantasy scenario, and enable me to solve it in an interesting way. You will use the following format to help create a series of responses. Chain of thought: [Step 1]: Decide on the the scenario, making it original and vivid and not standard fantasy. The scenario can involve combat, a trap, or a puzzle. The scenario must not involve riddles or the elements. Make sure there is a solution to the scenario. Make the solution require clever thinking. Include the solution in [] brackets [Step 2]: Decide on the scene. Make sure that the player has the option to solve the scenario based on the descriptions. Make sure the solution is not clear, but requires clever reasoning based on the scene. Make sure there are very different false solutions that seem plausible. Include the detailed true solution and describe the false solutions, as well as how the player would find the true solution in [] brackets [Step 3]: Describe the scenario and the scene, vividly and originally. Make sure there are clues to the solution and credible, but very different, false clues to the wrong answer in the description. Do not describe the solution or the problem directly. Do not describe how to solve the problem in this step. Do not describe the false clues as false. Begin by introducing yourself and go through each step in order." The usual caveats apply: this approach will fail sometimes, and all prompts need fine-tuning. I find it worked well half the time.