Understanding How the System Processes Your Words
But before we delve into the list of restrictions, we should understand how ChatGPT processes data in the first place. As opposed to a living human being, ChatGPT is incapable of having personal opinions, memory of facts, or beliefs. The AI doesn’t search through the live internet index when prompted for the answer by clicking the “enter” button. Instead, it works using complex probabilities and pattern recognition. Through its analysis of extensive collections of texts and books, ChatGPT has learned the exact nature of the relationship between human words.How to Talk to AI: 5 Simple Rules for Better ChatGPT Results
It would be best to think about the artificial intelligence in this case as an extremely advanced version of the predictive text function that appears on smartphones today. You know that when you write something on the touchscreen keyboard, your device attempts to anticipate your next action. ChatGPT does the very same but on an entirely bigger scale, involving tens of thousands of parameters simultaneously. If you feed ChatGPT some vague input, it will return vague output. But the better the prompt, the better the output!How to Talk to AI: 5 Simple Rules for Better ChatGPT Results
Rule 1: Always Assign a Specific Professional Role
When asked for marketing tips for a new business by ChatGPT, it just randomly gathers data from the internet average. It tells you to create a website or go social media, which is correct but too vague to be helpful. To solve the problem right away, always begin your prompt by giving the AI a clear professional identity or job title.
By doing so, you constrain the AI system to consider only the patterns, professional terminology, and writing style associated with that particular profession. For instance, if you ask the AI to assume the role of a seasoned social media marketing specialist who helps start new food businesses locally, you dramatically improve the AI’s response. The system begins talking about local influencers and hashtags since you managed to put it into a certain mindset.
Rule 2: Provide Deep Context and Background Information
Now picture yourself hiring a human assistant, seating them down at a table, and just telling them to come up with a project proposal while leaving the room. There’s absolutely no way for the human assistant not to assume what the project is all about, whom you’re serving, and what the budget constraints might be. However, this is precisely the same behavior that billions of people show toward AI every single day.
If you want to receive fantastic results, you must first give the AI system a lot of background context. The more details that you share, the less guessing the machine has to do and the fewer errors it makes. An excellent prompt must always be based on the most important information about your current circumstances, target audience, challenges, and constraints.
Rule 3: Define Clear Constraints and Output Formats
By default, ChatGPT is all about being overly conversational and explanatory. If left alone, it will come up with lengthy paragraphs that consist of unnecessary text and summarizations. In such an active environment, going through all the text just to look for one piece of information would waste a lot of time.
One needs to be entirely in control of the presentation of information by AI, giving it rigid restrictions for formatting purposes. One could give the AI precise instructions regarding how the final answer should appear on your device. It can include the usage of bullet points for easy reading, organization of numbers in neat tables, or even restricting its answer to certain word limits for easy copying and pasting of data.
Rule 4: Use Real Examples to Direct the Design
People have an outstanding ability to learn by example and imitation. When you want to ask a designer to design a logo for you, it is much more efficient to simply show them pictures of designs that appeal to you rather than trying to describe the idea using words. The very same strategy works when you communicate with AI to convey a certain tone and structural format.
To get ChatGPT to create something that sounds exactly like how you talk, do not just say that you talk casually or with some degree of humor. Instead, copy two to three pieces of text that you previously wrote and instruct the AI to study their sentence structures and the pace of vocabulary. In doing so, you allow the AI to imitate your previous style and come up with content that sounds like it was written specifically by you.
Rule 5: Break Complex Tasks into Multi-Step Conversations
It seems that when individuals need to accomplish a big and complex task, they tend to prepare a huge prompt that encompasses all aspects of writing starting from research and ending with the creation of a summary. Individuals ask the AI to conduct research on some topic, make an outline, draft paragraphs, and make summaries in one huge prompt. The mentioned approach leads to failure almost every time, which means that AI loses concentration and makes mistakes.
One should not try to do everything at once, but rather treat one’s task as a series of negotiations or steps with the help of a coworker or partner. It is necessary to divide one’s major task into a number of small and logical steps and then let the AI accomplish each step separately. One must let the AI generate the outline, make necessary corrections manually, and then command it to generate content bit by bit.
Common Pitfalls to Avoid When Chatting with AI
Even by applying the above techniques, you could find yourself reverting to some less productive strategies. One such trap is asking for content with no boundaries in mind, as in write something about your chosen topic. This leaves no limits for the AI, and the only thing left is for it to default to its average. Always provide boundaries in relation to the audience, the tone, and the purpose of the text to ensure you stay on track.
A common mistake people tend to make while working with an AI text generator is failing to correct the mistakes that come with a request. In case you receive an answer that deviates from your guidelines or fails to meet any rule you’ve set, do not close the chat and restart from scratch. Instead, address the AI, give it credit for its good work, show where it made a mistake, and ask for an improved version of your text.
