The other day, I was talking to a friend who just arrived from Nigeria. He has been in France for a few months, still struggling a bit with the language, but he has a strong will to build his new life here. He asked me to help him understand the basics of a computer topic. After explaining the fundamentals to him, an idea crossed my mind: "Tell me, do you know about artificial intelligence?".
He looked at me with a really surprised expression. He simply said no. It was the very first time he had heard about it. Yet, he had his smartphone in hand and an internet connection. Nothing more, but everything was already there.
We then set up two distinct tools:NotebookLM, so he could learn effectively from reliable documents (like his own administrative files or official guides), andGemini, so he could ask questions to the world and conduct in-depth research. I explained to him that one is his "knowledge vault" and the other his "personal assistant".
Of course, we talked about the famous "hallucinations" — those strange moments when the machine invents facts with incredible confidence. It’s vital to keep a critical mind, even in front of a machine. As I was leaving, I felt something strong: I had just given him the keys to his autonomy.
Honestly, I wish everyone had this foundation. Because beyond the tool, it's really the way to use it that makes the difference.
Why does your AI often give you off-the-mark answers?
We often have the impression that AI is a magical genius capable of reading our thoughts, when in fact it is primarily a reasoning engine (we understand each other, purists!). If you throw a vague question at it, it will give you a generic answer. It's a bit like asking a passerby for "the way" without saying where you're going. You will end up walking, sure, but not necessarily towards your destination.
To get a meaningful result, you need to stop sending it dry commands. You need to build acontext framework. You know what? It's simpler than it seems. We can structure this around four pillars:
Expertise (The "Who"):What role should the AI take on? An employment advisor? A technical writer? Or perhaps a patient language tutor?
The Goal (The "Why"):What is the ultimate objective? Do you want to understand a difficult letter, write a standout cover letter, or translate a technical concept?
The Sources (The "Material"):This is where the distinction is major. If you want the AI to rely only on your documents to avoid errors, head toNotebookLM. If you need broad web research, now is the time to callGemini or ChatGPT or Claude.
The Method (The "How"):Specify the tone, length, and especially the step-by-step process. Don't ask it to do everything at once. Haste is the enemy of precision.
The secret weapon: the "Missing Information Check"
This is the specific step I showed my friend to truly master the tool. Before letting the AI write or analyze anything, just tell it this:
"Analyze my request. Before starting the work, list all the missing information or gray areas in my context. Ask me your questions so you can be perfectly precise."
This is what we call aiteration. Instead of rushing in and making mistakes, the AI will point out what you forgot to tell it. We pause for a moment, think together, and then we succeed much better. It’s a conversation, not a monologue.
Empowering the ability to learn
Seeing my friend manipulate these applications for the very first time, I understood one fundamental thing: AI is not here to do the work for us. It is here to make us more capable. Giving it context is giving it the necessary eyes to guide us through the fog.
It’s a matter of pure autonomy. Whether it's to integrate into a new country or to wrap up a complex professional file, the rule remains the same: the quality of the response will always depend on the clarity of your context.
We all have a superpower in our pocket, as long as we know how to direct it. By the way, just for the record, we testedGemini Livetogether and the effect was immediate. This is just the beginning of the cycle, and frankly, it looks promising!
