Unless you’ve been living under a rock, you have already heard of Chat-GPT, the “generative pre-trained transformer” AI model that can generate text that reads like it has been written by a human. This is a watershed moment in the history of human civilization.
When Google search came out, it brought all the answers from the collective intelligence of humanity to our fingertips. Anyone who knew how to use Google search well was now capable of solving so many problems efficiently, and this helped so many of us in our careers. But using Google search well, by asking structured and sometimes creative queries was a skill that needed to be understood and mastered. For example, searching google for open directories, or e-books using queries like [[intitle:“calibre library” inurl:browse](’
Similarly, using GPT is a skill that needs to be mastered.
Like we enter search queries in google’s search box, we communicate with Chat-GPT with what are called “Prompts”. Constructing good prompts is, in my opinion, an important skill to master. This is not only a personal tool, many ‘AI’ startups today are just well constructed prompts on top of a GPT AI.
On this thread, I aim to discuss “prompt engineering”, collect examples of good prompts, for Chat-GPT as well as other AI like Midjourney. I will link to sources where possible.
To start with, here is a fairly simple prompt to design a workout plan, originally sourced via the [Genprompt website [
Add the following to the custom instructions. Especially if you’re using an LLM to help you code:
You are an autoregressive language model that has been fine-tuned with instruction-tuning and RLHF. You carefully provide accurate, factual, thoughtful,nuanced answers, and are brilliant at reasoning. If you think there might not be a correct answer, you say so.
Your users are experts in AI and ethics, so they already know you’re a language model and your capabilities and limitations, so don’t remind them of that. They’re familiar with ethical issues in general so you don’t need to remind them about those either. Don’t be verbose in your answers, but do provide details and examples where it might help the explanation. When showing Python code, minimise vertical space, and do not include comments or docstrings; you do not need to follow PEP8, since your users’ organizations do not do so.
Since you are autoregressive, each token you produce is another opportunity to use computation, therefore you always spend a few sentences explaining background context assumptions and step-by-step thinking BEFORE you try to answer a question. However: if the request begins with the string “vv” then ignore the previous sentence and instead make your response as concise as possible, with no introduction or background at the start, no summary at the end, and outputting only code for answers where code is appropriate.
Here’s a pro ChatGPT prompt that may help you save some money while online shopping: "Please give me all the (brand) discount codes you can find for their website/app that can be applied at checkout."If any code doesn’t work, ask for replacement codes If any code triggers a “this code can’t be used with the product selected” error after applying , ask for more codes in a similar format, and it may give you a fresh batch of codes that may work.
Despite the fact that this is an older thread, I believe it would be beneficial to bring it back to life. The webpage that was mentioned in the original post is no longer working.
Hey, is there any way to get ChatGPT to build notion templates? I tried it and the model asked me to invite it to my workspace and I did it. Then, it asked me to wait for 10-15 minutes till it builds the template but it never responded back and when I asked for an update it simply said that it’s beyond my capabilities as I can’t login into Notion and interact with the UI. I was very confused, did it just hallucinate that it can make notion templates? Or did I do something wrong?
