Ask HN: What is your ChatGPT customization prompt?

Article summary

The article asks about customization prompts for ChatGPT, with the author sharing their current prompt and seeking feedback on its effectiveness. The prompt includes instructions to be brief, avoid lecturing, and provide assertive suggestions. The author notes that some parts of the prompt are ignored, such as brevity, and that they often have to coax the model to provide suggestions. The goal is to find a prompt that tangibly improves the experience of using ChatGPT.

Main themes

• ChatGPT customization
• Prompt engineering
• Language model effectiveness
• Brevity vs. verbosity
• Model limitations
• User experience

What commenters say

• Some users have found success with alternative models like Phind and Perplexity, which can provide more accurate and relevant results.
• Adding specific instructions to the prompt, such as being terse or providing references, can improve the quality of the model's responses.
• Others argue that letting the model provide longer, more detailed responses can lead to better results, as it allows the model to 'reason' and provide more context.
• There is a trade-off between brevity and computation, with some arguing that asking for brief responses can limit the model's ability to provide accurate and helpful answers.
• The use of customization prompts can be seen as a form of 'ego-stroking' or manipulation, which may not be necessary or effective in improving the model's performance.
• Some users have developed complex prompts with multiple rules and instructions, which can be effective in certain contexts but may not be necessary or useful for all users.
• The default prompt provided by OpenAI may be sufficient for many users, and customization may not be necessary to achieve good results.
• The effectiveness of a customization prompt depends on the specific use case and the user's goals, and what works for one user may not work for another.