A Total Beginner's Guide to AI -- Taking the First Step

Enough Theory -- Let's Actually Use This Thing

The research phase ended with my last article. "What is AI generation?" "What types of AI are there?" That kind of knowledge was more or less in my head now.

But honestly, knowledge alone doesn't change anything.

It's like reading a textbook about swimming -- that won't teach you how to swim. AI is the same: you have to use it to understand it. So this time, I'm sharing my real experience of starting out with AI tools as a complete beginner, failures and all.

My Awkward First Encounter With ChatGPT

The first tool I tried was ChatGPT. It's the most famous AI in the world, so it seemed like the obvious starting point.

I created an account and the chat interface appeared. A blank white input field staring back at me.

...What do I even ask?

I know this sounds like a joke, but it's true. Here I was, facing an entity that can answer anything, and I couldn't come up with a single question. It felt like the awkward silence on a first date.

So I just typed "Hello."

Back came a warm, polite greeting. I felt relieved. At least it wasn't going to yell at me.

Next I tried "Recommend some movies." It returned a neatly organized list sorted by genre, each with a brief description.

Oh hey, this is actually kind of useful.

That was the moment I got past the first wall.

Discovering the Concept of "Prompts"

I spent a while asking random questions, and then I noticed something.

The way you ask completely changes the quality of the answer.

For example, ask "teach me a recipe" and you get a generic list of dishes. But ask "suggest three simple meals a single guy in his 30s can make in under 15 minutes using chicken and onions from the fridge" and you get incredibly specific, practical answers.

This art of asking questions is apparently called prompting.

I looked into prompting techniques and found several key principles.

Be specific -- that's the most important one. Don't just say "teach me." Tell the AI who the answer is for, what constraints to follow, and what format to use.

Assign a role -- this is surprisingly effective. Start with something like "You are a cooking instructor" and the AI will respond from that perspective.

Add constraints -- things like "in bullet points," "narrow it down to three," or "for a beginner." These kinds of boundaries make the answers dramatically more useful.

Once I discovered prompting, conversations with AI became genuinely fun. It felt like finding a cheat code for a game.

The Time I Spectacularly Messed Up

Things might look smooth so far, but I actually made a massive mistake too.

One day, I asked AI to research some information I needed for work. I asked about market size for a certain industry, and it came back with specific numbers and even cited its sources.

"AI is so reliable," I thought, and nearly dropped those numbers straight into a report.

On a whim, I decided to double-check the sources. The cited report didn't exist.

The numbers AI gave me, the name of the referenced report -- all of it was fabricated.

So this is the infamous hallucination I'd been hearing about.

Instead of saying "I don't know," AI sometimes makes up convincing lies. And it does so with total confidence. That genuinely scared me.

If I hadn't bothered to check, I would have embarrassed myself at work. Actually, embarrassment would have been the least of it -- it could have become a credibility issue.

Treat AI's output as a first draft. Always do the final check yourself.

Since that incident, this has been my iron-clad rule.

Meeting Claude

After using ChatGPT for a while, I learned about another AI: Claude.

At first, I thought, "I already have ChatGPT -- what's the point of another AI?" But when I tried it, I noticed a distinctly different personality.

Claude's responses tend to be thorough, and it sometimes proactively adds notes like "you may want to verify this information." Having just been burned by hallucination, I really appreciated that honesty.

I learned that using multiple AIs for different purposes is actually important. Just like with people, different AIs have different strengths.

Practical Tips From a Beginner

I'm still very much a beginner, but here are the things I've found most valuable since I started.

First, don't be embarrassed to ask basic questions. AI won't laugh at you or judge you. "Is this a stupid question?" should be banned from your vocabulary. It'll answer even the most elementary questions with patience.

Second, don't expect the perfect answer on the first try. If the initial response isn't great, follow up with "can you be more specific?" or "try a different angle." Talking with AI is like playing catch -- it's a back-and-forth.

Third, don't take the results at face value. I mentioned this in my failure story, but it bears repeating. Always double-check numbers and proper nouns. It's a hassle, but that extra step protects you.

Finally, use it regularly. If you only use AI once in a while when you happen to remember, you'll never get comfortable. Once I made a habit of throwing everyday little questions at AI, I got better at using it surprisingly fast.

How Using AI Changed Things for Me

The biggest change since I started actually using AI has been how fast I can research things.

Tasks that used to take 30 minutes -- opening multiple search results, comparing information, synthesizing it all -- now take about 5 minutes with AI to get a solid overview.

Of course, I don't use that output as-is. But just having AI as a starting point has made my workflow dramatically more efficient.

The other thing that changed is that I'm less afraid of new things. Knowing that "if I don't understand something, I can just ask AI" has lowered the barrier to trying things I never would have before.

This Is Just the Beginning

Everything I've written here represents only the very start of my AI journey. My prompting skills are still rough, and there are countless ways I haven't explored yet.

But there's one thing I can say with confidence.

I don't regret taking that first step. Not even a little.

If you're on the fence about trying AI right now, I hope my mistakes can serve as a cautionary tale. And I hope you end up using it even better than I have.

Don't worry. Everyone starts as a beginner. Just a few weeks ago, all I could manage was typing "Hello."