Asking AI for things without magic formulas
I tried the hidden tricks and the perfect prompt. What's left: a few principles that really change the result.
The ones that actually work aren't tricks: giving it context, showing it an example of what you want and correcting it when it gets things wrong. Many of…
The ones that actually work aren't tricks: giving it context, showing it an example of what you want and correcting it when it gets things wrong. Many of the "secret tricks" doing the rounds in videos are ordinary features with a flashy name, or magic phrases the model doesn't even need. If there's a secret, it's that ChatGPT answers better the less it has to guess.
ChatGPT doesn't have hidden keywords that unlock an "expert mode". It answers better when it understands what you need, who it's for and what the limits of the task are. Asking it to "act as an expert" adds little compared with explaining your situation in two lines. Something else that confuses people: tricks expire, because the model changes with every version and what worked a year ago may not work today. What is worth knowing are the options in the app itself, like giving it standing instructions about you so you don't have to repeat them in every chat.
Like almost everyone, I get the "tricks 90% of people ignore" videos too. In "Asking AI for things without magic formulas" I put them to the test on the same task, one after another, and I tell you what's still standing afterwards and which principles I use in my daily work. If you're looking for a list of tricks, fair warning: what you'll find is something more boring and a lot more useful.
I tried the hidden tricks and the perfect prompt. What's left: a few principles that really change the result.
One email a week with what was verified, what was refuted and what is still on the bench. No news roundups.