In the last post I walked through CROFTC, a six-part checklist for prompts that don’t make the AI guess: Context, Role, Objective, Form, Tone, Conditions. This post is the practical follow-up: 50 ready-to-use ChatGPT prompt templates, one for each situation people actually run into, from blog titles to cold emails to planning your week. I call them recipes because they work like one: the steps are written, and you bring the ingredients. They work the same way in Claude, Gemini, or any other chat AI. One credit before we start. The 50 situations are borrowed from the list in Tomasz READ MORE
Category: Prompt Engineering
How to Write a Prompt That Works: Context, Role, Objective, Form, Tone, Conditions
Ask ChatGPT to “write an email to my landlord about the heating,” and you’ll get something polite, tidy, and useless. It won’t mention when the heat went out, what you’ve already tried, or what you want to happen next. It can’t. You never said. That’s the whole problem with most AI prompts. The AI isn’t lazy, and it isn’t dumb. It’s guessing. Every detail you leave out is a blank it has to fill, and it fills each blank with the most average choice it can find. Average in, average out. This post walks through a six-part checklist that removes READ MORE
Two important aspects of AI
In this post, I am going to briefly talk about two important aspects of AI – tokens, hallucinations. What are tokens? You communicate with GPT with natural language such as English. How does ChatGPT understand your question? Sentences, words, and characters are transformed into many individual tokens. Here is the example sentence: Chat GPT processes text by splitting it into tokens. Each token will have a unique ID. What happens after the tokenization? It tried to predict what would be best token after the current one. And this could potentially lead to a fundamental problem – hallucination What are hallucinations? READ MORE
Principles of Prompt Engineering – Evaluate Quality
So far, you have given direction, specified format and provided examples ChatGPT can follow. But are you going to take the first result as is without any confirmation? How do you know if your prompt is actually proper and optimized? You need next principle – evaluate quality. How do you assess the quality? You need to run the same prompt multiple times and check each response, evaluate and modify the query accordingly. If you consistently get satisfying responses after running the same prompt 10 times for example, then it is probably good to go. As you can see in the READ MORE
Principles of Prompt Engineering – Provide Examples
So far we have taken a look at two principles – Give Direction, Specify Format – which are relatively simple. For direction, you just need to say something like ‘in style of Steve Jobs’. For format, you can say a table, (or comma separated list, JSON). Providing examples is essentially providing test cases that ChatGPT can test its results and follow. Provide diverse set of test cases where the task was done correctly. Providing example involves more time and effort because you have to research and pick or come up with the best example. Again, a simple and naive prompt READ MORE
Principles of Prompt Engineering – Specify Format
Continuing from the last post, the next principle is “specify format” to get a reliable result. Specifying format is really defining what rules to follow, and what response structure is required Again, using the same simple prompt to generate 10 product names for a backpack for men Instead, you can specify which format the GPT should follow for the response The result from the prompt should be consistent because it could be used as the input to another task. A consistent result is key.
Principles of Prompt Engineering – Give Direction
It is needless to say AI is changing the world. Some people say this is the 4th industrial revolution. However, it would be more accurate to say the 1st AI revolution, because this will be a complete game-changer. It is very important to know how to use the AI. It will give you the answer based on the quality of the question. Prompt engineering is the newly arising field to communicate with AI to extract reliable, maximum result. I am not an expert prompt engineer and this post is more like a learning note for me. Need to give clear READ MORE