AI Basics: Intro
What actually is AI? (In Plain English)
At its core, AI—especially the kind we use most today—is powered by something called machine learning. This means a computer model is trained on enormous amounts of data (like text, images, code, and more) so it can learn to spot patterns and make predictions.
Tools like ChatGPT are built on what’s called a Large Language Model (LLM). These models don’t “understand” language the way humans do. Instead, they predict what’s most likely to come next based on the patterns they’ve seen in real-world writing.
Think of it like this:
If you see the phrase: “Once upon a…”, what word comes next?
You probably thought: “time.”
Not because you reasoned it out—but because you’ve seen that phrase so many times in stories. That’s exactly how an LLM works: it’s seen that phrase thousands (or millions) of times in its training data, so it predicts “time” as the most likely next word.
Now scale that up.
LLMs don’t just predict the next word—they can generate entire sentences, paragraphs, or pages based on what’s most likely to come next, using patterns they’ve learned from vast amounts of data.
But here’s the catch: they’re just guessing.
For example:
If I start with “I went to the bank…”, what comes next?
Is it:
- “…to withdraw some cash”?
- “…to sit by the river”?
Both make sense. Without more context, the model might choose the wrong one—and confidently continue in the wrong direction.
That’s how AI can go off course.
AI Is Brilliant—But It’s Not Perfect
LLMs are powerful, but they’re not mind readers. They’re making informed guesses, not understanding meaning the way humans do.
That’s why:
- A vague or poorly written prompt can lead to confused or irrelevant answers
- The AI might “hallucinate” facts that sound right but aren’t
- You sometimes get completely wrong or misleading code, summaries, or responses
A recent BBC article even reported how some companies now exist solely to clean up the damage from bad AI output—like broken websites, inaccurate legal advice, or even privacy issues.
But when used wisely, AI can:
- Save time
- Improve quality
- Spark new ideas
- Automate tedious work
This site exists to help you use AI intentionally, effectively, and safely—so you get results, not regrets.
More Help With This Section: Video
📄 View Video Transcript
Hello and welcome!
When most people these days discuss AI, they’re really talking about machine learning models. That’s when computers study vast piles of data like text, pictures, code, and audio to spot patterns and predict what’s likely to come next.
Think of it like learning to cook. After reading thousands of cake recipes, you notice that flour, sugar, and eggs are almost always used. If one’s missing, you can guess what should be there. Machine learning does the same thing—on a massive scale.
A specific type of machine learning model is a large language model, like the one behind ChatGPT. These specialize in text generation, but they don’t actually understand language like humans. Instead, they work like a supercharged autocomplete.
For example, if I say “Once upon a…”, your brain probably completes that with “time.” Not because you reasoned it out, but because you’ve seen it many times. That’s what the AI is doing too—predicting the most likely next word based on patterns it’s seen in training data.
Now imagine doing that prediction one word at a time, really fast—that’s how it generates whole paragraphs that sound human. But here’s the catch: it’s still just guessing. It doesn’t know what words mean. It’s predicting, not understanding.
Here’s another example. If I say, “I went to the bank…”, the next words could be “to withdraw some cash” or “to sit by the river.” Without more context, the AI might pick the wrong one and continue confidently in the wrong direction. That’s called a hallucination—when AI generates info that sounds right but is false.
For example, if you ask “What’s the capital of Australia?”, it might say “Sydney” instead of “Canberra” because Sydney appears more often in popular writing. It’s not lying—it’s just predicting based on frequency, not accuracy.
So what can you do? The key is being specific. Vague prompts lead to vague answers. Clear, well-structured prompts with context lead to better results. And always double-check what AI gives you—whether it’s facts, summaries, or code.
Think of AI as a super fast assistant. It can help a lot, but you’re still the editor.
Used well, AI can save time, boost creativity, and improve your work. It’s like having a creative partner who never sleeps—but still needs clear instructions.
Used wisely, it’s an amazing tool—and we’re here to help you explore it.
Is AI Free To Use?
While some AI tools offer free access, most are not truly “free.” That’s because:
- These tools are expensive to build and run.
- They rely on enormous datasets, powerful servers, and constant updates.
- The companies behind them are commercial businesses that need to earn a return.
You’ll often encounter AI in two main ways:
- Directly (like paying for ChatGPT Plus)
- Indirectly, when you use an app (like a meeting summarizer or writing tool) that uses an LLM behind the scenes via an API
Here’s the important bit:
API access is a paid service. Every time you interact with an AI-powered app, that app is sending your request to an LLM and being charged for it. That cost adds up quickly—so to stay afloat, these apps have to charge their users too.
That’s why:
- Most tools have paid plans
- Many offer free tiers, which let you try out the tool with basic features or usage limits
- For some light users or early experimentation, the free tier might be enough
- But for frequent use or more powerful features, you’ll likely need to upgrade
Throughout this site, we’ll highlight when tools are free, freemium, or fully paid—so you know what to expect.
What You’ll Find on This Site
Action With AI is a practical guide to using AI in the real world.
It’s a growing collection of:
✅ Step-by-step guides to real-world AI tasks
🧠 Explanations of how and why things work
🧾 Copy-and-paste prompts that actually get results
📸 Screenshots, examples, and real use cases
📺 Links to useful YouTube videos where relevant
The goal is not to hype AI or make unrealistic promises. Instead, it’s to help real people do real work more efficiently and creatively—without getting lost or overwhelmed.
A Note of Caution
Everything on this site is based on my personal experience using AI tools in day-to-day work. I’m not an AI engineer or a legal expert—just someone who’s spent a lot of time figuring out what works and what doesn’t.
This site is intended to educate and inspire, but:
- Use your own judgment when applying anything from these pages
- Always test and review AI-generated outputs carefully
- I can’t accept liability for decisions you make based on what you read here
Think of this site as a guide, not a guarantee.
Join the Conversation
Tried something that worked really well? Hit a wall with a certain tool? Have a better way of prompting?
I’d love to hear from you.
Every guide includes a comment section where you can share your results, suggest improvements, or ask questions.
AI is evolving fast—and we’re all figuring it out as we go. Let’s help each other do it better.