"Artificial Intelligence is the Tool. Augmented Intelligence is the Goal."

 Hey everyone! 👋

So, you know how we're all hearing a lot about AI these days? It feels like it's everywhere, right? Well, I recently took a Google course about it, and guess what? It wasn't as scary or complicated as I thought! I wanted to share some of the cool basic stuff I learned, kind of like we're chatting about it over coffee or something.

Think of this as a super simplified intro to talking to AI, especially those really smart ones that can write stuff!

First Off, What Even IS AI?

Okay, so AI, or Artificial Intelligence, is basically just computer programs that can do tasks that usually require human smarts. Think about things like Google Maps figuring out the best way to get somewhere or how YouTube recommends videos you might like. That's AI at work!

A big part of AI today uses something called Machine Learning. This is where the computer program learns from data to make decisions or guesses. Like, imagine showing a computer tons of pictures of ripe and unripe apples. It learns what makes an apple ripe. Then, when you show it a new apple, it can predict if it's ripe or not. The better the pictures (data) you show it, the better it gets!

And What's This "Generative AI" Thing?

So, among all the different types of AI, one that's become super popular lately is called Generative AI. This is the cool one that can actually create new stuff. Not just analyze data, but generate brand new text, images, sounds, even code!

The most common type we interact with right now is called a Large Language Model (LLM). This is an AI that's really good at processing and generating text. Google's example is their Gemini model. You just type into it, and it gives you text back. You can ask it to help you write an email, brainstorm ideas, summarize articles, or just chat. Pretty neat, huh?

Okay, So How Do We "Talk" to It? That's "Prompting"!

The way you interact with these generative AI tools is by giving them instructions. This is called prompting. When you ask Gemini to write that email, you're giving it a prompt.

Now, just typing anything isn't always enough. HOW you phrase your instructions – your prompt – can totally change the quality of the answer you get. That's where the idea of "prompt engineering" comes in. It's basically learning the art of writing good prompts to get the results you want.

The Secret Sauce: A Simple Framework

The Google course shared a super helpful 5-step framework for designing prompts. It's really the base for everything else. Think of it as a recipe for a good prompt:

1. Task: Start with the core thing you want the AI to do. Like, "Suggest a gift for my friend's birthday".

2. Context: This is key! Give the AI background info and specific details. The more context, usually the better the result. You can add details like your friend's age, what they like (anime, maybe even which anime they like), and what kind of vibe you're going for. You can also tell the AI to act as a specific Persona (like an "anime expert") or specify the Format you want the output in (like a list, bullet points, or a table).

3. References: This is where you can show the AI examples of what you want. Sometimes it's hard to describe something, but showing an example makes it clear. LLMs are really good at seeing patterns and copying styles. So, you could show it examples of gifts your friend liked in the past or show it the style of writing you want it to use. The course calls examples "shots" – zero-shot (no examples), one-shot (one example), few-shot (a few examples). Providing more examples often leads to more specific results.

4. Evaluate: After the AI gives you an answer, check it. Is it accurate? Is it what you wanted?

5. Iterate: If the answer isn't quite right (and it often isn't the first time!), refine your prompt and try again. Prompting is almost never perfect on the first try; it's a back-and-forth process. The course's motto is "Always Be Iterating" (ABI).

You might start with just the Task, then add Context, then maybe add a Reference, and keep tweaking until you get what you need.

Stuck While Iterating? Try These Tricks!

Sometimes you've added context and references, but the AI is still not giving you exactly what you want. The course suggests a few methods for iterating when you hit a wall:

• Go back to the framework: Maybe you didn't add enough context or need to provide more examples.

• Break it down: If your prompt is really long or complex, try separating it into shorter, simpler sentences. It's like trying to explain something complicated to someone – breaking it into steps helps.

• Rephrase or use an analogy: Try asking for the same thing in a different way. Or, ask for something similar but framed differently. Like, instead of asking for a "marketing plan," you could ask it to "write a story about how this product fits into customers' lives" – it's an analogous task that might give more creative results.

• Add constraints: Sometimes having too many options makes the AI's job harder. Tell it exactly what you don't want or put specific limits. Like, for a road trip playlist, maybe say "only include songs about heartbreak" (if that's your vibe!) or "songs from a specific region". Adding constraints helps narrow the focus.

Beyond Just Text & Things to Be Aware Of

While we often type prompts, some AI tools can handle other types of input, like pictures, audio, video, or code. And they can give you outputs in those formats too. The basic framework still applies, but you might need to be more specific about the input or output format.

It's super important to know that AI isn't perfect. Two big things to watch out for are:

• Hallucinations: This is when the AI just makes stuff up that isn't true. It can be totally nonsensical (like "rabbits brainstorming the White House") or sound surprisingly believable but still be wrong. This is why you always need to check the facts.

• Biases: Because AI learns from human-created data, it can unfortunately pick up human biases. For example, it might assume doctors are men and nurses are women based on biased data. Developers try to reduce this, but it's something to be aware of in the output.

Because of these issues, the course strongly recommends a "human-in-the-loop" approach. This means you are always checking the output and have the final say. Don't just blindly use what the AI gives you, especially for anything important.

Why Practice This?

Being good at prompting can really help us out! It can save time on everyday tasks like writing emails, brainstorming ideas, or summarizing long documents. You can even use it to help with things like analyzing data or building presentations, but remember to be careful about putting sensitive information into the tools.

Getting better at prompting just takes practice and lots of iterating!

Anyway, that's a quick peek into the basics of talking to AI effectively, based on the Google course. It's all about being clear, providing context, using examples, and refining your prompts until you get the results you want. Hope this helps make AI feel a bit less mysterious! I encourage you guys to watch the videos below that can explain what I talked about above much better. 

Let me know if you have any questions or want to chat more about any of this! 😊

[THE AUTHOR] Ray is the Global U Class's (2025 Freshmen) Lead Servant and Developer of this Site, and community innovator of TeamKwail.com. He aims to create a system that would help not only the current class but all future global learner to collaborate and break through barriers.