AI from scratch for non-technical people: a practical guide (2026)

What an AI chatbot is, what it can do for you day to day, how to start today without paying and which mistakes to avoid. No jargon, official sources.

By Cognenza editorial team · Updated 15 October 2026 · 5 min read · Sources cited

What the AI on your phone (really) is

When people talk about “using AI” today, they almost always mean a chatbot: an app where you type (or speak) in normal language and it replies with text, tables, images or even small interactive tools. Under the hood it runs on a language model (LLM), a program trained on huge amounts of text to predict which answer best fits what you ask.

What matters to you isn’t how it works inside, but three ideas:

• It understands normal language. There are no commands to learn.
• It doesn’t “know”, it predicts. That’s why it is sometimes wrong with total confidence.
• It gets better with context. The better you explain your situation, the better it answers.

The main assistants in 2026

On 7 October 2026 OpenAI began bringing its new generation of models, GPT-6, to free ChatGPT users too. That version can answer with charts, buttons and small tools inside the conversation itself. News moves fast: we track it on the AI radar.

Your first 10 useful tasks

1. Draft a tricky email (a complaint, a payment reminder, an apology).
2. Summarise a long text in 5 points.
3. Explain a complicated document “as if I were 12”.
4. Plan a shopping list or a weekly menu from what’s in your fridge.
5. Plan a trip with times and a rough budget.
6. Correct the spelling and tone of a text.
7. Translate a message while keeping a natural tone.
8. Brainstorm: names, gifts, activities for a birthday.
9. Practise a job interview by having it ask you questions.
10. Organise messy notes into a table.

How to ask so it gets it right

A good prompt (the request you write) almost always has these ingredients:

• Who you are or who it’s for: “I’m a primary school teacher…”
• What you need: “…I need a note for parents…”
• Key details: “…about Friday’s trip, leaving at 9:00…”
• Format: “…under 100 words and in a friendly tone.”

If the answer doesn’t convince you, don’t start over: ask for tweaks (“shorter”, “remove the emojis”, “make it more formal”). It’s a conversation.

What AI shouldn’t do for you

• Make decisions about your health, money or legal matters. Use it, at most, to understand terms or prepare questions for a professional. Google says so explicitly in the Gemini privacy notice: its answers are for information only.
• Give you unchecked facts. Figures, dates, quotes and laws: always verify them at the official source.
• Receive data that isn’t yours. No information about clients, patients or pupils. We explain this in what data never to paste.

Privacy: the setting worth checking on day one

Both ChatGPT and Gemini can use your conversations to improve their models, and both let you turn this off:

• ChatGPT: Settings → Data controls → turn off “Improve the model for everyone”. There’s also temporary chat, which doesn’t appear in your history and isn’t used for training (OpenAI may keep a copy for up to 30 days for safety).
• Gemini: in “Gemini Apps Activity” you can turn off “Keep activity”. Google warns that human reviewers may read some conversations if you leave it on.

And what does the law say?

In the European Union the AI Act (Regulation (EU) 2024/1689) applies. For you as a user, the most visible part is transparency: AI-generated or manipulated content that could pass as real (such as deepfakes) must be identified as such.

Educational content, prepared with the help of AI tools, with sources cited. It is not a substitute for professional advice. Tools change fast: if you spot something out of date, write to us at contacto@cognenza.com. How we work.