29/10/25

Over time, I’ve found a few prompts that push it to explain things more clearly, think differently, and sometimes even surprise me. Here are my top tips! 🏅

  1. “Step by Step”

Teach me x step by step. Break this concept down for me step by step.

I want you to think about what types of content (data) have the phrase “step by step” in them. Why tutorials, recipes, how-to guides and educational content of course! Hence the model will be more likely to draw on these sources that teach procedures and processes in a logical order, making it more likely output will mirror that clear, sequential structure without skipping any steps.

  1. “Criticise what you just told me”

I like using this prompt because AI tends to follow a particular pattern when responding. You might’ve noticed that generative AI can be a bit of a people pleaser. It often tries to agree with you or reinforce what you’ve already said. So, challenge that! Ask it to critique its own answer, explore other angles, and offer a different perspective.

  1. **“Explain this to me like I’m 10 years old.”(**Yes. I will unapologetically use this prompt to help me understand problems or concepts I find difficult to understand or grasp)

After a long lecture or explanation via a workshop I might not grasp a concept straight away, and I certainly won’t understand the complicated terms that have just been introduced to me for most likely the first time. I love using this prompt so I can connect unfamiliar terms to familiar terms. (You can obviously raise the age/ or lower it btw if you want).

Bonus: Try and compare! 👀

Enter your input/question into the model first in a language (English for example) and then try the exact input/question but in another language (Mandarin) (and then translate it back to English). Did you get the same answer? Maybe they’re different!

AI Models like OpenAI, Gemini etc. are trained on vast data sets including the internet, and as many parts of the internet are in different languages you’ll get an entirely different knowledge base. Further, you can also see and compare what it actually knows vs what it’s been trained to say.

Yours truly,

Regina