How to Use ChatGPT – A Beginner's Guide

Lately, there’s been a lot of talk about ChatGPT – it’s always just a click away, always ready to help and provide convincing explanations about the world around us. Some use it for time-killing chats, others for solving tough geometry problems, and some for analyzing massive volumes of information – and so on.

But do we actually know how to work with ChatGPT effectively, and how much we can trust it?

If we’re not familiar with communicating with AI language models, we might not get the result we expected or needed, or we might even be misled by the information provided.

Here are a few tips for beginners on how to work with ChatGPT.

But before we jump into a conversation with it, let’s get to know it a little first (remember what our parents used to say: “Don’t talk to strangers”?)

Does ChatGPT have a name?

We asked our chatbot what its name is, and it chose “Alek,” because “Alek” could stand for *Artificial Linguistic Expert Companion – but that sounds too much like the title of a scientific paper, so let’s just stick with Alek.” That’s what we’ll call it throughout this post.

What are AI models like ChatGPT?

AI models like Alek are large language models. They are trained on vast amounts of textual data to understand, summarize, translate, and generate human-like language – and even write code.

How do language models work?

• These models use a mechanism called self-attention to analyze the context of words in sentences. This helps them understand the relationships between words, phrases, and ideas in long texts.
• They are pretrained on massive text corpora (internet, books, articles), where the model learns statistical relationships between words.
• Then, they are fine-tuned for specific tasks or improved using human feedback.
• When we ask Alek a question or give it a prompt, it starts "thinking" about which word is most likely to come next. It doesn’t "understand" like a human, but it has learned how words typically connect. Word by word, Alek builds an answer that sounds meaningful and logical.

Limitations of AI models

• They don’t truly understand the world; they operate through probabilistic patterns in text.
• They can generate incorrect or made-up information (so-called “hallucinations”).
• They may reflect biases from their training data.
• They require significant computing power to train and use.

Now that we’ve met Alek, let’s see how to “talk” to him properly:

 

7 Tips for Working with AI

 

1. Clarity and context

When “talking” to Alek, phrase tasks as if explaining to a 10–12-year-old. Provide context, describe the task in detail, and specify the desired result. The clearer the question, the better the answer.

Example:

❌ “Tell me something about work.”

✅ “What are the pros and cons of remote work in the IT sector?”

AI doesn’t read minds – context is key to getting helpful results.

2. Avoid misleading phrasing

AI works through probabilities, not understanding. When we ask a question, the AI doesn’t grasp its meaning like a human. Instead, it analyzes which words typically follow others across billions of texts. It’s a statistical model, not a logical brain.

For example:

If we say: “How do I raise a unicorn in my apartment?”

We won’t get: “Wait... unicorns don’t exist.”

We’ll more likely get: “Make sure there’s enough space for its horn and feed it moon grass regularly.”

So — if a request sounds valid, Alek will develop it, even if it's not logically sound.

Solution: Be critical of the assumptions in your question – if the premise is false, the answer may be pure fantasy.

3. Fact-check

Alek can sound very convincing, but sometimes he's like an overconfident teenager who didn’t read the textbook. If the info is important (e.g., medical, legal, financial), always verify it with a trusted source.

As the saying goes: “Trust, but verify.”

4. Ask for multiple perspectives

AI can present information from various angles, compare, and highlight pros and cons – if we ask it to.

You can try:

• “Show the advantages and disadvantages...”
• “What do different experts think about this topic?”

This way, you avoid one-sided answers and get a more balanced view.

5. Don’t expect a “human” answer

Talking to Alek might feel natural, but he doesn’t have feelings, opinions, or life experience. So don’t expect emotional understanding or moral judgment unless it’s simulated based on the data provided.

6. Don’t share sensitive information

AI may be a good conversationalist, but it’s not your diary or legal advisor. Never input sensitive information like:

• Personal data
• Financial details
• Passwords or confidential business secrets

(Think of it like writing to a very smart participant in a public forum – stay cautious!)

7. Share your goal, not just a question

Alek works better when he knows what you’re trying to achieve – then you become co-creators instead of passive users.

Try saying:

✅ “I want to write a motivation letter – help me with structure and ideas” 

instead of:

❌ “Write me a motivation letter.”

In Conclusion

ChatGPT isn’t a wizard – it’s a tool that understands us only as well as we communicate with it. With a bit of context, clarity, and human critical thinking, you’ll get more than just an automated answer – you’ll get a meaningful exchange. (Still, always keep a critical eye and double-check!)

AI won’t replace us – but it’s already here to help, inspire, and sometimes even make us laugh.

Top tip:
Next time you sit down to “chat” with Alek, ask for a unicorn care guide. Just in case. Better safe than sorry.

 

See more on ChatGPT:

AI in HR - In which areas should artificial intelligence NOT be used?

How to Optimize Your External Recruitment Budget?