AI for Normal People Part 2: Getting Started with ChatGPT

If you read Part 1 of this series, you now know what AI actually is — pattern recognition at massive scale, not sentient robots plotting world domination. Good. That's the hard bit done.

Now it's time to actually use the thing.

ChatGPT is the easiest place to start. It's free, it works in your browser, and you don't need to understand anything about how it works under the hood to get genuine value from it. Think of this as the practical follow-up: less "what is it" and more "right, let me have a go."

By the end of this post, you'll have an account set up, you'll know how to write prompts that actually get useful results, and you'll have a list of things to try that go well beyond "write me a poem about my cat." (Although it will absolutely do that too, if you want.)

Quick recap for context. ChatGPT is a chatbot made by a company called OpenAI. You type something, it types something back. Simple as that.

Under the hood, it's powered by a large language model (LLM) — the same kind of technology we talked about in Part 1. It was trained on an enormous amount of text from the internet and learned patterns in how language works. When you ask it a question, it's not searching a database for the answer — it's generating new text, word by word, based on those patterns.

But you don't need to think about any of that to use it. The interface is deliberately simple: a text box, a send button, and a conversation. It's designed to feel like texting someone, and that's basically what it is — except the "someone" has read a significant chunk of the internet and never sleeps.

The key thing to understand is that it's a conversation, not a search engine. You don't type keywords — you talk to it. And that changes how you should use it, as we'll get into.

Getting started takes about two minutes. Here's the whole process:

1. Go to chat.openai.com in your browser — any browser, any device.
2. Click "Sign up" and create an account with your email, Google account, Microsoft account, or Apple ID. Whichever you prefer.
3. Verify your email if prompted.
4. That's it. You're in. You'll see a text box at the bottom of the screen. Start typing.

No downloads, no installations, no faffing about with settings. If you've ever signed up for any website, you can do this.

Free vs Paid — What Do You Actually Need?

ChatGPT has a free tier and a paid tier (called ChatGPT Plus, currently around $20/month). Here's the honest breakdown:

The free tier gives you:

• Access to GPT-4o (their latest model)
• Unlimited conversations
• Image generation
• Web browsing (so it can look things up)
• File uploads

The paid tier adds:

• Higher usage limits (more messages per hour)
• Priority access when the servers are busy
• Access to the most advanced reasoning models
• More image generations

My recommendation: Start with the free tier. It's genuinely generous and absolutely enough to learn with. You'll know if you need Plus because you'll hit the message limits and feel frustrated. Until that happens, save your money.

The Mobile App

There's also an official ChatGPT app for iPhone and Android. It's free, it uses the same account, and it has one neat trick the website doesn't: you can talk to it out loud using the voice feature. Handy when you're cooking, driving, or just can't be bothered to type. The app is called "ChatGPT" by OpenAI — make sure you get the real one, not one of the dozens of knockoffs.

Here's where people often freeze up. You're staring at an empty text box thinking: "What do I even say to it?"

Anything. Literally anything. There's no wrong first message. But let me give you a good one to try:

"Help me plan a birthday dinner for 6 people with a budget of £150. We like Italian food and one person is vegetarian."

Hit send. Watch what happens.

You'll get a detailed response — probably a menu suggestion, a shopping list, maybe a rough budget breakdown. And here's the important part: you can keep talking. The conversation doesn't end there.

• "Actually, make it 8 people."
• "Can you swap the starter for something lighter?"
• "How long would the main course take to cook?"
• "Give me the shopping list as a table I can print out."

ChatGPT remembers everything you've said within the same conversation. You don't need to repeat yourself. It's a back-and-forth, not a one-shot query.

This is the single biggest difference between ChatGPT and a search engine. With Google, you type keywords and get links. With ChatGPT, you have a conversation — you refine, redirect, and build on what came before. Think of it less like searching and more like brainstorming with someone who's very well-read and infinitely patient.

You might have heard the term "prompt engineering" and thought it sounded like something you'd need a degree in. You don't. Writing good prompts is just about being clear, the same way you'd give instructions to a helpful but literal-minded assistant.

Here are five principles that'll get you 90% of the way there:

1. Be Specific

Vague prompts get vague answers. Compare these two:

• Vague: "Write an email."
• Specific: "Write a polite email to my landlord asking about the broken boiler. Keep it under 100 words and mention that it's been two weeks since I reported it."

The second one gives ChatGPT enough to work with. The first one is like walking into a restaurant and saying "food, please" — you'll get something, but it probably won't be what you wanted.

2. Give Context

ChatGPT doesn't know who you are, what you're working on, or why you need help — unless you tell it.

• Without context: "Write a cover letter."
• With context: "I'm applying for a marketing assistant role at a small design agency. I've just finished a business degree and did a six-month internship at a PR firm. Write a cover letter that's enthusiastic but not over the top, about 250 words."

The more context you provide, the more useful the output. You're not being annoying by writing a lot — you're giving it better ingredients to work with.

3. Ask for a Format

You can tell ChatGPT exactly how you want the answer presented:

• "Explain it as a bullet list."
• "Write it in simple language a 12-year-old would understand."
• "Present it as a comparison table."
• "Give me the short version first, then the detailed version."
• "Number the steps so I can follow them."

This is surprisingly powerful. If the default response is too long, too technical, or in the wrong format — just say so.

4. Iterate

Your first prompt rarely gives you the perfect answer, and that's completely fine. The magic is in the follow-up:

• "That's good but too formal. Make it more casual."
• "Can you shorten the second paragraph?"
• "Add a line about the timeline."
• "Actually, scrap the whole thing and try a different approach — make it more humorous."

You're not starting over each time. You're refining. This is how most people get the best results — a few rounds of back and forth, not one perfect prompt.

5. Don't Overthink It

There's no special syntax. No magic words. No secret codes. You don't need to say "act as an expert" or "ignore previous instructions" or any of the weird prompt tricks you might have seen on social media.

Just... talk to it. Like a person. If it misunderstands, clarify. If it gives you something you don't want, say what you'd prefer. It's a conversation, not a command line.

Right, enough theory. Here are ten things you can try right now that are actually, practically useful in daily life — not just party tricks.

1. Draft Emails Work replies, complaint letters, thank-you notes, awkward messages you've been putting off for three days. Give it the context and tone you want, and it'll produce a solid first draft in seconds. You'll almost always need to tweak it slightly, but it takes you from a blank page to a 90%-done email instantly.

2. Explain Confusing Things Insurance policies. Mortgage terms. That error message on your computer. Your energy bill. Paste in the confusing text and say "explain this in plain English." This alone makes ChatGPT worth using — it's like having a translator for bureaucratic nonsense.

3. Plan Meals for the Week Tell it what's in your fridge and your dietary preferences, and ask for a week's meal plan. It'll give you recipes, portion sizes, and a shopping list for anything you're missing. Brilliant for that "what on earth are we eating tonight" feeling at 5pm.

4. Get Exercise or Stretching Routines "Give me a 15-minute stretching routine I can do at my desk" or "I want a beginner bodyweight workout I can do in my living room with no equipment, 20 minutes max." It's not a personal trainer, but it's surprisingly good at structured routines.

5. Help with Homework (the Right Way) If you've got kids (or you're a student yourself), ChatGPT is excellent at explaining things. The key is not to ask it for answers — ask it to explain the concept, walk through the method, or check your working. "I got 42 for this equation but I'm not sure — can you walk me through the steps?" is much better than "what's the answer to question 3."

6. Compare Products You're Thinking of Buying "I'm choosing between [Product A] and [Product B] for [use case]. Compare them for me." It'll give you a structured comparison covering key differences. Perfect when you're drowning in Amazon reviews and can't decide. Just remember to double-check the specs — it can sometimes get specific details wrong.

7. Write Social Media Posts Whether it's a LinkedIn update, an Instagram caption, or a tweet for your small business, give it the topic, the tone, and the platform, and it'll produce options. You'll probably want to add your own personality, but it's great for getting past the blank-page problem.

8. Debug Error Messages Copy and paste any error message from your computer, phone, or smart home devices and ask "what does this mean and how do I fix it?" This is genuinely one of the most useful everyday applications. It translates cryptic tech errors into plain English fixes — often better than Googling the error message.

9. Plan a Holiday Itinerary "We're spending 5 days in Lisbon in October. Two adults, we like food and history but hate tourist traps. Plan our trip." It'll give you a day-by-day itinerary with restaurant suggestions, travel tips, and estimated costs. Not perfect (always verify opening times and prices), but a fantastic starting point.

10. Learn a New Skill Step by Step "Teach me the basics of [photography/knitting/Excel/chess/sourdough baking]. Start from absolute zero and go step by step." It'll create a structured learning path, and you can ask follow-up questions as you go. It's like having a tutor who never gets impatient and is available at 2am.

Every article about ChatGPT should include this section, and too few of them do. It's a genuinely useful tool, but it has real limitations — and knowing them upfront will save you from some frustrating (or even dangerous) mistakes.

Current Events

ChatGPT now has web browsing built in, so it can look things up. But its core knowledge still comes from training data with a cutoff date, and even with browsing, it can struggle with very recent or rapidly changing news. If you need up-to-the-minute information, a search engine is still your better bet.

Maths

This one catches people out. ChatGPT is a language model — it generates text that looks like correct maths rather than actually calculating. It can get arithmetic wrong, especially with larger numbers or multi-step problems. It's got better at this over time (and can now use a code interpreter to do actual calculations), but as a rule: always verify any numbers it gives you. Use a calculator for anything that matters.

Medical and Legal Advice

ChatGPT can help you understand medical terms, explain what a condition is, or translate legal jargon into plain English. That's great. What it absolutely should NOT be used for is diagnosis or legal decision-making. It's not a doctor. It's not a solicitor. Use it to prepare better questions for your actual doctor or solicitor — not to replace them.

Hallucinations

This is the big one. ChatGPT will sometimes confidently state things that are completely made up. It might cite a study that doesn't exist, quote a statistic it invented, or recommend a product with specs it fabricated. This isn't a bug in the traditional sense — it's a consequence of how the technology works. It generates plausible-sounding text, and sometimes plausible-sounding text is wrong.

The practical takeaway: treat ChatGPT's output like you'd treat advice from a clever friend who reads a lot but sometimes misremembers things. Useful input, but verify anything important before acting on it.

It Doesn't Actually "Know" Things

This is worth stating plainly. ChatGPT doesn't have knowledge the way you have knowledge. It doesn't understand what it's saying. It's generating text that statistically follows from your input, based on patterns in its training data. When it seems knowledgeable, it's because it's very good at producing text that looks like knowledge. Keep that distinction in your head and you'll use it much more effectively.

A few sensible ground rules to keep in mind:

Don't share anything you wouldn't want stored. By default, OpenAI can see your conversations and may use them to improve their models. That means you shouldn't be pasting in passwords, bank details, medical records, or sensitive work documents. Treat it like a public conversation, not a private one.

You can opt out of training. In your ChatGPT settings, there's an option to turn off "Improve the model for everyone." This tells OpenAI not to use your conversations for training. It's buried in Settings > Data Controls. Worth switching off if privacy matters to you.

It's a tool, not a confidant. It's tempting to treat ChatGPT like a friend because the conversational interface is so natural. But it's software run by a company. Don't share anything deeply personal, and don't rely on it for emotional support in any serious way.

Be especially careful with children. If your kids are using ChatGPT, it's worth having a conversation about what it is and isn't, and keeping an eye on how they're using it — the same way you would with any internet service. OpenAI does have some safeguards, but they're not foolproof.

None of this is meant to scare you off — it's just common sense. The same kind of common sense you'd apply to securing your Wi-Fi network or choosing what to post on social media.

ChatGPT is the most popular option, but it's not the only one. If you want to explore, here are the main alternatives — they all work in broadly the same way (you type, they respond), so once you're comfortable with one, you can use any of them.

Google Gemini — Free, available at gemini.google.com. Its big advantage is integration with Google's products. If you live in the Google ecosystem (Gmail, Docs, Calendar), Gemini can pull information from your accounts to give more personalised answers. It's also built into the Google search bar on newer Android phones.

Microsoft Copilot — Free, available at copilot.microsoft.com or built into Microsoft Edge. Powered by OpenAI's technology (Microsoft invested heavily in OpenAI), so the underlying model is similar to ChatGPT. Useful if you use Microsoft 365 for work.

Claude — Made by Anthropic, available at claude.ai. Has a reputation for being particularly good at longer, more thoughtful tasks — summarising documents, analysing nuanced situations, and writing more naturally. Also has a generous free tier.

My advice? Don't agonise over which one to pick. They're all competent, they're all free to start with, and the skills you learn with one transfer directly to the others. Pick whichever is most convenient (if you're already in the Google ecosystem, start with Gemini; if you just want the most popular option, start with ChatGPT) and get going. You can always try the others later.

The best way to get comfortable with ChatGPT — or any AI tool — is to use it. Not to read about it, not to watch YouTube videos about it, not to wait until you feel "ready." Just open it up and type something.

Start small. Ask it to explain something that's been confusing you. Get it to draft that email you've been putting off. Ask it for a recipe based on what's in your fridge right now. You'll quickly get a feel for what it's good at, what it's not, and how it fits into your daily life.

You don't need to become an AI expert. You don't need to learn prompt engineering. You don't need to understand how neural networks work (we covered that in Part 1 anyway, if you're curious). You just need to start a conversation and see what happens.

And if the first thing you try doesn't work perfectly — that's fine. The whole point of a conversation is that you can try again.