Using AI as My MLA-C01 Tutor: What It Did in Four Weeks, and What It Couldn't
Part two of my notes from passing AWS MLA-C01 on September 26, 2026: having AI check the exam specs, write study notes in Traditional Chinese, quiz me five questions at a time with hints, and turn 271 past questions into Chinese audio, plus the things I still had to do myself. A personal study log.
Contents
Learning without thought is labor lost; thought without learning is perilous.
—— The Analects, “Wei Zheng” (Spring and Autumn period); translation mine
I passed AWS MLA-C01 on September 26, 2026, and for most of the four weeks before that I studied alongside an AI. It did what a tutor does: checked the exam specs, wrote study notes in Traditional Chinese, quizzed me five questions at a time with hints, and turned 271 past exam questions into Chinese audio I could play on my commute. Three things it couldn’t do for me: think through a question first, redo my mistakes days later, and sit a full 140-minute timed mock. The setup only works if you give it one rule: quote the official source, never answer from memory.
Day one, it almost let me relax too early
My last post covered the study route. This one is about the tools, and it starts on day one. The first thing I asked the AI to do was copy the exam specs from the official pages, word for word, with no answering from memory.
It found that the English version would retire on September 28, 2026, and that candidates registered in a localized language can switch to English during the exam. All good. Then it added that the Simplified Chinese version “has no deadline,” and I’ll admit I felt relieved.
I took one more step and had an AI from a different company pick apart the first one’s work. It pushed back right away: AWS just hadn’t published an end date, and “unknown” doesn’t mean “none.” That pulled me back. I didn’t ease off, and I finished before the end of September as planned. Looking back, that one sentence may have been the most valuable thing in the whole four weeks.
It started writing my study notes
There’s no shortage of study material online, but nearly all of it is in English and scattered, and it was slow going for me. So I had the AI write study notes in Traditional Chinese, weighted by what the past questions actually test.
First it grouped the keywords across all 271 questions to see which topics come up most. Training came first, then inference cost and data processing, and the notes were sized to match. Every lesson had the same shape: one takeaway, a summary table, a memory hook with common traps, one diagram, and a few self-check questions.
For diagrams I gave it one rule: only draw things that actually have a shape. “Three ways to roll out a new model version” takes a paragraph in words, but three curves make it obvious at a glance. Draw something with no shape and you’ve just added one more thing to decode.
It became my quiz master
I drilled by asking the AI for questions right in the chat. It translated past exam questions into Traditional Chinese and gave me five at a time, answers only after I’d finished. For questions I’d never seen, it explained the concept in plain words and gave a hint instead of the answer. For multiple-response questions it told me how many to pick, so I wouldn’t lose points on the count.
The most useful part was how it broke down my mistakes. It explained why the right answer was right and where my pick went wrong. I used to piece that together from other people’s discussions, which took ages. Now a question took a few minutes, and the time I saved went into another round.
It turned my walks into study time
Later I wanted to review on my commute too, so I had the AI add Chinese audio to all 271 questions, with a play button on each one.
That was harder than I expected. It failed many times along the way, each time for a different reason, and I nearly gave up more than once. But once it worked, it changed things. Walks and train rides became listening time, and a lot of concepts only clicked on the third listen, somewhere on the street.
The three things it couldn’t do for me
After all that, I want to be honest about what it couldn’t replace. First, you have to think through each question yourself. Reading its explanations often made me feel I understood, and that feeling was an illusion. The answer only stuck after I picked one myself and got it wrong. Second, you have to redo your mistakes days later. The ones you still miss are the ones you really don’t know, and nobody can do that for you. Third, you have to sit through the full timed mock. Only after 140 minutes in the chair do you find out where your pacing falls apart.
If you want to study with AI too
I suspect the worry for a lot of people is whether AI gets things wrong. It does. The dangerous part is when it’s wrong and sounds completely sure, like that “no deadline” line on day one.
My fix has two layers. First, restrict it to quoting the source. Then get a second AI to look for holes. When both agree, I believe it. Where they argue is usually exactly where I need to go back and read the original.
One thing to take away
When you use AI to study or look something up, give it one rule first: quote the official source, never answer from memory. Then have a second AI pick holes in it. You can try it today. Take a question you’re currently looking into, ask two different AIs, note where they disagree, and check the original to see who’s right.