SQL for Data Analytics · AI in this class
AI in this class
AI can write SQL. It can probably do most of the practice in this course correctly. So students sometimes ask: if AI can already do this, why do I need to learn it?
That's a fair question. Here's my answer.
I don't speak Spanish. If I need something written in Spanish, I can ask AI to write it. It will give me something. But I can't tell if it says what I wanted. I can't tell if something is missing or wrong. And I haven't learned any Spanish.
SQL is the same. When AI writes a query for you, you are the one who has to decide whether it's right. Is it pulling from the right tables? Is it counting what you meant it to count? Does the number it gives back make any sense? You can only answer those questions if you can read the query and you understand the data. If you can't, a wrong answer looks the same to you as a correct one.
AI is a really good tool for people who could do the work without it. If you could have written the query yourself, AI makes you faster. If you couldn't, you can't check what it gives you.
I'm not tremendously worried about whether you remember the exact syntax for every command. People look syntax up all the time, and that's fine.
What I care about is that you have a model in your head of how this works. You know what a table is. You know how tables connect to each other. You know what each part of a query does to the data.
So this is the standard I'm holding you to. With a reference in front of you, you should be able to build a query on your own. You should also be able to read a query somebody else wrote and tell what it does. That's what I'd expect of any engineer on my team. Being able to explain the code is non-negotiable.
You build that model by using the tools. The lessons explain the concepts, but the understanding comes from applying them. That means writing queries, getting them wrong, and fixing them. I'm not big on memorizing for its own sake, but you do have to learn the tools. They're how you work with the thing you're building a model of.
Learning something well is supposed to take effort. It's supposed to be confusing for a while. If it feels too easy, you're probably not learning much. That's the problem with having AI do the work. You can breeze through everything and still not have learned it, because the effort is what creates the learning. The first video in Unit 01 covers learning theory in more detail.
AI is allowed in this class, but you must use it appropriately.
Good uses. You're stuck, and you ask it to break the problem down or give you a first step. Your query isn't doing what you expected, and you ask it to explain why. You want a second explanation of something from a lesson. You want more practice, and you ask it to write you extra questions. In all of these you're still the one doing the work.
What doesn't work. Asking it for help before trying to solve an error yourself first. Or pasting in the question and turning in what comes back. You don't learn anything from it. The final is half your grade, and you take it without AI.
Getting unstuck. Don't spend more than five or ten minutes staring at a blank page and getting nowhere. At that point it's fine to ask AI for help. But once you're unstuck, try the question again on your own, or try a similar one.
The goal is to get to a place where you can do the questions yourself. If you get to the end of a unit and you still can't, that's the time to seek out more support. Office hours and the tutoring center are great resources for that. AI can be too.
You're expected to understand and be able to explain anything you turn in. If it looks like you're leaning on AI without understanding the material, I'll talk with you about it. That might be at the check-in or in a separate meeting. My full policy is under Academic Honesty in the syllabus.