Generative AI versus traditional software

For most of computing history, software worked one way: a person wrote down exact rules, and the computer followed them. A calculator adds numbers because someone coded the steps for addition. A tax form flags a mistake because someone wrote the rule that caught it. Give the same input twice and you get the same output twice, every time, because nothing is being invented. The machine is following instructions.

Generative AI is a different kind of software. Nobody wrote the rules for how to answer your question or draft your email. Instead the system was shown an enormous amount of examples - text, images, code - and it learned the patterns in them. When you ask it for something, it generates new content that fits those patterns. That is what the word "generative" means: it creates something new rather than looking up a preset answer.

The everyday difference

Think about the difference between a spellchecker and an assistant that writes your email.

• A spellchecker follows rules. "Teh" is not in the dictionary, so it is flagged. Predictable, and it only does the one thing it was built for.
• An email assistant was never given a rule for your specific email. You describe what you want, and it produces a fresh draft that never existed before. Ask twice and you may get two different drafts.

That flexibility is the whole point, and it is also why generative AI behaves so differently from the software you are used to. It does not run a fixed recipe. It produces a best guess at what you asked for, which is powerful and, as later lessons will show, something to keep an eye on.

Why this matters for you

You do not program generative AI, you ask it. The skill is not writing code, it is knowing what these tools can do, which one to reach for, and how to ask well. That is what this course is about. The next lessons map out the landscape before you touch anything hands-on.