You opened ChatGPT once, typed “hi”, got a friendly answer back and closed the tab without knowing what to do next. If that was you, you are in the majority. The hard part of artificial intelligence is not the technology: it is that nobody explains the bare minimum before pushing an empty text box at you.<\/p>\n
This is lesson 1. No math, no code, and it assumes nothing. By the end you will know what an AI is, what that LLM acronym you see everywhere means, why these tools fail in such a specific way and how to write a request that works. It takes about 12 minutes to read, and at the end there are ready-made prompts for you to test.<\/p>\n
Artificial intelligence is a program that learns patterns from examples instead of following rules that somebody wrote by hand. That is the whole difference.<\/p>\n
An ordinary program is a recipe: “if the customer spent more than $100, give a 10% discount”. Someone decided that and wrote it down. An AI works the other way around: you show it ten thousand photos of dogs and ten thousand of cats, never explaining what an ear or a coat of fur is, and the program works out on its own what tells them apart. Nobody wrote the rule. It emerged from the examples.<\/p>\n
That is why AI gets right the things that are impossible to describe with rules (recognizing a face, understanding a clumsy sentence, suggesting the next song) and gets wrong the things any calculator handles. Keep this sentence in mind, because it explains almost everything that follows: an AI does not know, it estimates<\/strong>.<\/p>\n When you ask ChatGPT a question, it does not go to a database to fetch the answer. It predicts the most likely next word, then the next one, then the next, until the answer ends. It is statistics about language, running very fast.<\/p>\n That sounds disappointing and it is exactly what makes the tool useful. Predicting the next word well requires, in practice, having absorbed grammar, context, the structure of an argument and a good deal of knowledge about the world. But it also explains the most famous flaw: when the AI does not have the information, it does not go quiet. It generates the most plausible continuation, which is sometimes an invention wearing the face of truth. People call that hallucination, and it is not a bug: it is the same mechanism running with no good data to lean on.<\/p>\n LLM stands for Large Language Model<\/em>. It is the type of AI behind ChatGPT, Gemini, Claude and practically everything that talks to you in text.<\/p>\n It is worth pulling the three words apart, because each one says something practical:<\/p>\n Two concepts you will run into and should recognize. Token<\/strong> is the little chunk the text is cut into before it enters the model, something between a syllable and a word. Context window<\/strong> is how much text it can consider at once: anything that falls outside it is forgotten, which is why a very long conversation starts contradicting itself.<\/p>\n One detail that clears up a lot of frustration: asking for the same thing twice can give you different answers. There is deliberate randomness in the process. If you need consistency, say so in the request (“always answer in this structure”) instead of hoping it happens on its own.<\/p>\n The list below covers what most people really use in 2026. All of them have a free tier that is enough to get started.<\/p>\n Do not burn energy picking the best one. All three of the text tools handle 90% of what a beginner needs, and the prompt you write works in all three. Pick one, learn to ask well, and only then compare.<\/p>\n Here is the part that changes your result today. A bad request is short and vague. A good one carries four things.<\/p>\n Compare them. “Help me with my resume” gives back a generic list you had already pictured. Now the same request with the four elements, which is exactly what makes the Resume from Scratch (No Experience)<\/a> deliver a finished document instead of loose advice.<\/p>\n And the most underrated technique: the second message<\/strong>. Almost nobody uses it. The first answer is a draft. Say what did not work (“too formal”, “cut it in half”, “the third item does not apply to me”) and ask it to redo only that part. That is where the quality shows up, and it is free.<\/p>\n No honest course ends without this section.<\/p>\n An example of correct use in that gray zone: Understand Your Medicine Leaflet in Plain Language<\/a> is there so you can grasp what is written and turn up better prepared at the appointment. It does not replace your doctor, and the prompt says so.<\/p>\n Open ChatGPT, Gemini or Claude and go in this order. Every prompt below is already written; you just copy it, paste it and swap what is inside the brackets.<\/p>\n If you want to keep practicing afterwards, these work well for a beginner because the result is immediate and easy to judge: Private Tutor for Any Subject<\/a>, Chef of What Is in Your Fridge<\/a>, Gift Consultant<\/a>, Instagram Caption That Drives Engagement<\/a> and Back to School Without the Stress<\/a>.<\/p>\n If you want to watch AI settle a real decision with numbers in hand, take a look at Is This Civil Service Exam Worth It?<\/a>. And on the image side, two that show well what the technology does today: Restore a Torn and Creased Photo with AI<\/a> and Preview a Tattoo on Your Photo Before Getting It<\/a>.<\/p>\n Not to get started. ChatGPT, Gemini and Claude have free plans that are enough to learn on and to use day to day. The paid plan gives you more volume, faster answers and access to the better models. It makes sense once you already use it every day and hit the limit, not before.<\/p>\n AI is the whole field, which includes image recognition, recommendation and forecasting. An LLM is a specific type of AI, trained on text, and it is the one that talks to you. Every LLM is an AI; most of the AIs in the world are not LLMs.<\/p>\n It does not lie in the sense of wanting to deceive, because it has no intent. It generates the most likely continuation, and when good data is missing that continuation can be false while looking true. In practice the effect is the same, so check every number, name, date and quote.<\/p>\n It depends on the tool and on your account settings. When in doubt, assume that it does and do not paste sensitive data. In the privacy settings of the main tools there is an option to turn off the use of your conversations for training.<\/p>\n Role, your context, output format and constraint. Those four elements fix most bad results. And use the second message to correct course instead of rewriting everything from scratch.<\/p>\n No. Nothing in this lesson requires code. Knowing how to code opens doors for automation, but to use the everyday tools what counts is knowing how to write a clear request.<\/p>\n You already have the essentials: you know that AI estimates instead of knowing, you know what an LLM is, you know why it hallucinates and you know the four elements of a request that works. All that is missing is repetition.<\/p>\n The fastest way to get practice is to use ready-made prompts and modify them, because that way you see the structure before you have to invent it. Pick a topic in the Prompteiro catalog<\/a> or start with the text prompts<\/a>, which are the easiest ones on day one. In the next lesson we get into what separates a good prompt from an excellent one.<\/p>\n","protected":false},"excerpt":{"rendered":" What artificial intelligence is, what LLM means, how these tools work on the inside and how to use the free ones today. Lesson 1 of the course, from zero.<\/p>\n","protected":false},"author":1,"featured_media":39,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_i18n_grupo":1538,"footnotes":""},"categories":[64],"tags":[],"class_list":["post-38","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-courses"],"_links":{"self":[{"href":"https:\/\/prompteiro.com\/en\/wp-json\/wp\/v2\/posts\/38","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prompteiro.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prompteiro.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prompteiro.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prompteiro.com\/en\/wp-json\/wp\/v2\/comments?post=38"}],"version-history":[{"count":1,"href":"https:\/\/prompteiro.com\/en\/wp-json\/wp\/v2\/posts\/38\/revisions"}],"predecessor-version":[{"id":40,"href":"https:\/\/prompteiro.com\/en\/wp-json\/wp\/v2\/posts\/38\/revisions\/40"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/prompteiro.com\/en\/wp-json\/wp\/v2\/media\/39"}],"wp:attachment":[{"href":"https:\/\/prompteiro.com\/en\/wp-json\/wp\/v2\/media?parent=38"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prompteiro.com\/en\/wp-json\/wp\/v2\/categories?post=38"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prompteiro.com\/en\/wp-json\/wp\/v2\/tags?post=38"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}The golden rule: it predicts, it does not look things up<\/h2>\n
What LLM means<\/h2>\n
\n
How it works on the inside, in four steps<\/h2>\n
\n
Which artificial intelligences are the most popular<\/h2>\n
For writing and chatting<\/h3>\n
\n
For images<\/h3>\n
\n
How to use it: what separates a bad request from a good one<\/h2>\n
\n
When AI is not the right tool<\/h2>\n
\n
Your first exercise, in 10 minutes<\/h2>\n
\n
Quick questions<\/h2>\n
Do I have to pay to use artificial intelligence?<\/h3>\n
What is the difference between AI and LLM?<\/h3>\n
Can artificial intelligence lie?<\/h3>\n
Does AI use what I write to train?<\/h3>\n
How do I write a good prompt?<\/h3>\n
Do I need to know how to code?<\/h3>\n
What comes next<\/h2>\n