A few days before publishing this post, I read two of my friend’s post on LinkedIn about his experience on being ‘slowing down’ in the rapid current of AI. While he find it beneficial for learning, I found myself contemplating it further. No, I’m not in disagree stance with his statements, but I was thinking about my own experience. That’s absolutely effective, but is it efficient?
If the purpose of learning is to get a strong fundamental understanding of certain subject and mastering it along the way, I’m definitely agree and encourage to do it the hard way! But for some instance, sometimes the purpose of my learning experience isn’t about mastering the subject from the bottom to the top. It’s more about broadening my perspective on other subjects that are closely related to my area of expertise. So the target is to have a sufficient knowledge about the new subject and pragmatically implement it along with my expertise.
In that case, taking the the hard path might be not the right choice. As we have time constraint and might lose opportunity if we take too long to learn. It would be more efficient to utilize AI to take care about our known unknowns on a real case.
“So it would be better to just let AI handle the known unknowns and we just focus on what we could?”
I consider to say no. It’s correct that AI could be used to assist us handle lot of things so we could just focus on the things that matter most. But the keyword is to ‘assist’, not to ‘replace’. In terms of learning, let AI brings all things to the chat room means you are not studying, you just vibe studying while AI do studying. Just to make it clear, in my POV, study or learn isn’t just about getting descriptive information. It’s about shaping idea, connecting the dots, reasoning, and more.
Okay, I don’t want this blog going out of scope so you can think about ‘what is learning’ later. But the point is, have some new descriptive information from AI might not considered as learning. However, if you brainstorm with AI, develop your own ideas, challenge the AI's responses, and implement the new knowledge in your own way, then I would say that is learning!
Back to my context, in case of my learning experience, I try to utilize AI to brainstorm and giving me a clear context about the new subject and it’s relation with my expertise. Along the way, I try to probe AI to guide me further to the subject I wanna learn.
To optimize my learning experience, Instead of start from the bottom (fundamental) and go up stairs (implementation), I tried to do it top down. Means I start from the real case implementation, then iterate to learn deeper by hands on and extra curiosity to ask AI about more fundamental concepts. I don’t think it would match for everyone. But as I already have a professional experience and I love to learn by doing, this approach while utilize AI could escalate my learning process.
Let me give an example to demonstrate my approach:
I’m a Software Engineer who focused on Backend and I’m curious to learn about generative AI. I tried to got some context about it and found about RAG (Retrieval Augmented Generation). It’s interesting to me because it has intersection with my current expertise and it’s possible to create a project which implement RAG as part of my learning by doing.
Initially, I still need to read some sufficient context about generative AI, LLM, RAG, etc. After I have sufficient information to relate it with my expertise (from unknown unknowns to known unknowns), I start my first RAG project. I setup the project as usual. When it comes to the specific RAG parts, I took more time. Instead of just asking the AI for the implementation, I probed it for the 'why' and the 'how’. This led me to become more curious about deeper concepts and fundamentals, comparison of different implementation methods, other subjects to cover more advance project, and so on.
By using that approach, my learning process became more efficient while still maintain the effectiveness aspect. Yeah, it’s not as effective as taking the hard path. But back to the premise, it’s more than enough if the purpose is to just have sufficient context of subject related to our expertise.
Top comments (0)