If you have to start now, open nbsp;ChatGPT or Google Gemini and talk to them. It’s your only priority in the first weeks. Ask them for anything: to plan meals, to explain a concept, to help you draft an email. The goal is for you to lose your fear and understand what a “prompt” is (the instruction you give) in the most natural way possible: by conversing. Don’t overthink it; these language assistants are the most direct and intuitive gateway.
What you must avoid at all costs at the beginning are the tools that will frustrate you and won’t teach you the essentials:
- Local models like Stable Diffusion on your own PC. They require a good graphics card, knowing technical commands, and dealing with configuration errors. They will kill your motivation.
- APIs and complex automation platforms (like Zapier with AI). They’re for when you already have a specific problem to solve, not for learning the basics.
- Going straight to programming or studying advanced mathematics. It’s the fastest route to giving up. First play with the technology to see its potential; the theory will come later when you have a reason to learn it.
After mastering chat, your path should be this
Don’t skip steps. This progression is designed so that you build on what you already know, without roadblocks.
| Phase | Focus on this | Avoid this | Why |
|---|---|---|---|
| First Contact (First 2 weeks) | ChatGPT, Gemini, Copilot (chat version). Playing with simple prompts. | Local models, APIs, complex automation. | You need quick wins and to understand dialogue with AI without technical roadblocks. |
| Creative Expansion (Weeks 3-4) | Generative image AI: DALL-E 3 (in ChatGPT Plus) or Midjourney. | Base Stable Diffusion without an interface, generative video models. | You learn to be specific in your instructions (prompts) in a controlled environment with high-quality results. |
| Everyday Integration (Month 2) | AI in your tools: Copilot in Office/Google Workspace, Notion AI, Canva AI. | Programming your own workflows with tools like n8n or Zapier AI. | You see the practical value of AI in your daily life, without having to learn a new app from scratch. |
| Specialization (From month 3 onward) | Whatever interests you: data analysis, development, audio. | Nothing, you already have good judgment. | Now you can tackle complex tools with a solid foundation and a clear objective. |
Two uncomfortable truths you must accept from minute one
- AI lies and makes things up. It’s called “hallucination.” Never take an important piece of data it gives you for granted without verifying it, especially if it’s dates, names, or scientific studies. It’s a brilliant assistant but sometimes it confabulates.
- Not everything sold as “AI” actually is. It’s a buzzword. Many companies use it as marketing for products that are simple automations. Your compass should be what the tool actually allows you to do, not its label.
The only skill that really matters: writing prompts
Mastering AI is mastering the art of giving instructions. A vague prompt gives a vague result.
- Bad: “Make a study plan“.
- Acceptable: “Make a study plan to learn AI fundamentals in a month”.
- Good: “I’m a marketing professional with no technical background. Create a 4-week study plan to understand how generative AI works and apply it to my job. Include weekly objectives, resources in Spanish, and no more than 5 hours per week. Present it in a table”.
The difference lies in the context (who you are), the detail (what you want exactly), and the format (how you want it). Practice by adding layers of information to your requests.
If all this sounds like too much, ignore 90% of it and just remember the first line of this answer: start with ChatGPT or Gemini and talk to them. The rest is noise until you take that first step.