I think it has never been easier to learn something new. Anything, anytime, as deep as you want. If you’ve been putting off a course or a certification, or you always wanted to learn something but never got around to it, AI can be the kind of support that just didn’t exist before.
But before you hand your study materials over to ChatGPT, let me tell you how I almost ended up producing content instead of actually learning. And why keeping things simple turned out to be the only way out.
The tireless study partner
I recently finished a postgraduate program. The last time I was at university was 12 years ago. Back then, studying meant going through forums, swapping handwritten notes with classmates, spending evenings Googling until I found an answer that made sense. It was slow and it was tiring. But it worked. All that effort was actually how the knowledge got into my head.
This time, I had AI. And it changed a lot. Just not always the way I expected.
The biggest difference was having a study companion with unlimited patience. There were no bad questions. No eye-rolling when I asked the same definition for the fifth time. I could go as deep as I wanted into any topic, at any hour, and get a clear explanation for exactly where I was stuck.
I could not have imagined this 12 years ago. It made understanding the bigger picture so much faster.
But I noticed something odd: the knowledge wasn’t sticking the way it used to.
The frictionless learning paradox
When you don’t have to fight for information (hunt it down, write it out by hand, organize it yourself), it passes through you more easily. The effort I used to complain about was actually what made things stick in my memory. Without it, I had to find new ways to remember: generated flashcards, endless quizzes, even custom podcast episodes from my study notes that I could listen to while commuting.
This approach, actively testing yourself instead of just re-reading, worked. But it also introduced a new problem.
Notes that never stay finished
Because regenerating materials was so easy, I could restructure my whole approach whenever I wanted. I worked through all 100 exam topics, then realized that some parts, like detailed code snippets, would never come up in an oral exam. So I reworked everything from the beginning.
With handwritten notes, I would have never done that. Nobody rewrites hundreds of pages in the middle of exam season. But with AI, I could. And that’s where things started to go sideways.
The webapp I didn’t need
Here’s where it gets embarrassing.
At some point I wanted to be able to review my notes on any of my devices, so I built a webapp from my study materials. It looked great. I spent hours tweaking the CSS, fixing layout bugs, adjusting colors. I felt really productive. I was “working on exam material,” after all.
Except I wasn’t learning anything. I was doing web development.
You know when students go through their notes with a highlighter instead of actually memorizing them? At least the highlighter forces you to read the sentence. Adjusting a background color teaches you nothing about your exam topics.
Just because you can build it doesn’t mean you should.
This was really bothering me so I looked it up, and apparently what I’d fallen into is called “productive procrastination” in psychology. You do something that feels like progress but you avoid the actual hard work. And AI makes this trap bigger than ever, because suddenly every format is one prompt away. Why not a quiz app? Why not a mind map? Why not another version of the notes, but better organized?
The spiral was real.
Back to basics
What got me out of it was deciding to strip everything down. I dropped the webapp and moved my notes into Obsidian, a plain-text note-taking app. Just text files and links between concepts. No design to worry about. No interface to perfect. Just the content.
It sounds too simple, but that was exactly the point. With nothing to distract me, I had to actually engage with the material. The app has a graph view that shows how your notes connect to each other, and it turned out this is much closer to how I actually needed to think about the topics: not as isolated lists, but as connected ideas.
Where AI still falls short
In the final stretch before exams, I tried something more ambitious. I used voice mode to practice for the oral exam: I’d say a definition out loud and ask the AI to evaluate my answer.
The idea is great. The reality is not there yet. The AI kept interrupting me mid-sentence, it misinterpreted statements as questions, and it fell back on its own knowledge instead of sticking to my notes. What I really wanted was for it to pick random topics from my material, ask follow-up questions, and give me an honest evaluation at the end. But that’s still too complex for current models.
It’s a promising direction though. I think it will get there.
Looking back, this experience taught me something I didn’t expect. The hard part of learning with AI is not getting the information. That’s the easy part now. The hard part is staying focused on actually learning it.
If you’ve been thinking about going back to school, picking up a new skill, or getting that certification: honestly, there has never been a better time. AI takes away so many of the things that made learning feel like a lonely struggle.
If there’s one thing I took away from all of this, it’s that the real question was never what AI can help me build. It was whether I was still learning while building it.