While generative artificial intelligence (AI) is increasingly used for a whole bunch of tasks, many users do not know how to get the most out of this technology. Know that Chatgpt, but it is also valid for Claude, Grok, Gemini or the Cat, are much more efficient if prompt, that is to say textual requests, are perfectly refined. And there are precisely several tips to get there.
Give a role to AI
Asking the chatbot to take on a specific role is one of the most effective ways to get a well -calibrated response. For example, by typing ” Pretend you were a university professor with 30 years of experience in sociology », You will receive a sharp, structured and in -depth analysis.
You can, on the contrary, ask him to communicate as if he were a school teacher, so that his explanations are the most simplified possible. Concretely, The more you define the expected tone and level of expertise, the more the AI will accurately adapt to it.
Illustrate your expectations with concrete examples
The AI works by mimicry: the more you show it what you expect, the closer it gets. Giving an example, even simple, helps orient style, tone or level of detail.
For example, if you ask “ Writes a professional email “, You can specify the desired tone as” POLI MORE DIRECT ». Same bell sound for a LinkedIn post, an item summary, a grip or a message: Providing a model helps AI to better understand your intention.
Contextualize your requests
It is certainly one of the most common errors: Too many users ask a raw question without specifying what they really expect. Yet, It is by explaining why you ask this question that AI can really adjust. Is it for a duty, an article, an oral presentation, or by simple curiosity? What do you already know about the subject? Do you want a synthetic, technical or popularized response?
All this information allows the model to adapt its level of detail, its tone, or the structure of its response. The more transparent you are transparent on your intentions and constraints, the more relevant, targeted and useful the answer.
Assign a personality at AI
One of the levers not to be overlooked is the possibility of specifying the tone or the personality you expect from the AI. Because yes, she can be a pedagogue, synthetic, biting or even funny … provided you asked her. Tell him, for example, that you want a hard-hitting, ironic or inspiring style. In another context, you can tell him to adopt a benevolent posture, with concrete examples and simple reformulations.
You can also specify that you want something direct, corroborated by statistics. The AI then adjusts its way of expressing themselves, as a human would do according to its audience. Enough to obtain more suitable and engaging answers.
Ask for a step by step reasoning
When a question is complex or has several components, it is better to ask the AI to detail its reasoning. This not only makes it possible to grasp the why of the how of the answer, but also to identify any errors or approximations. Because do not forget: the models are not perfect and can tend to hallucinate.
To solve a mathematics problem, write an argument or analyze a text, ask him to explain his point in point by point to follow his intellectual journey. Note that now, many chatbots offer an option with advanced reasoning that appears automatically. For some, this option is, however, paying.
Specify the desired response format
To get a really useful answer, it is better to clearly indicate the expected structure. This ranges from the flea list to the comparative table via the summary in a few lines or even an ultra-short version of the Tweet type. This can save you time, avoid reformulations and potentially integrate the answer directly into your support.
Clearly define what should be included, or excluded
To refine the answer, it is relevant to indicate what AI must take into account, but also what it must ignore. Mention a specific period, a type of source or a specific angle allows you to frame the content from the start. Conversely, dismissing certain elements, such as social networks, too technical examples or data prior to a certain date, avoids unnecessary digressions.
This method prevents you from filtering the superfluous information yourself: the clearer your criteria, the more AI align with your objectives.
Advance in stages, as in a real conversation
Interacting with an AI is not limited to a single answer. To go further, it is sometimes better to have a progressive approach: start with a general demand, then refine over the exchanges. This dynamic, close to a dialogue, makes it possible to adjust the formulation, to explore new angles or to dig a particular point.
Clearly indicate to whom the answer is addressed
AI adjusts its language, level of detail and tone according to the target audience. It is therefore better to specify from the start that the content is intended: high school students, experts, beginners, etc..
This indication makes it possible to avoid too technical explanations or, on the contrary, too simplistic. It directs the choice of words, the length of the sentences, the examples used. By adapting the response to the right interlocutor, we gain in clarity and impact.
📍 To not miss any Citron press, follow us on Google News and Whatsapp.
