"vibecode" journaling for responsible AI use

I garauntee I’m not using AI as optimally as I should be.

AI was just kind of thrown at us in the blink of an eye and in such a blur that I don’t think very many people had the chance to develop an intentional approach towards it. I know I didn’t, so I bet others haven’t as well. Well no more! Today I’m deciding to take a little bit of intentionality towards AI with the creation of a vibecoding (I hate this word but its weirdly become the vernacular now so cest la vie) journal.

The idea is that I want to stop using it like google search and spitballing whatever haphazard thought I may have at it and use it truly as a tool; not a VOID for throwing brain junk into. AI companies would probably prefer I continue to throw whatever at it because that leads to better and more honest profiling of their customers but we can do better that just be the perfect data harvest for these companies :)

What advantage do I hope to get from doing this? Is it worth the time sunk into it?

To answer the second question first, it takes no time at all because as I would later find out: You can ask these AI tools to remember information about you, including prompts that essentially kinda act like functions , so if I create an instruction “every time I write: journal this - do XYZ” , it will do it. As for the first question: I hope to just waste less time with AI tools, and to succumb less to brain rot/WALL-E syndrome where machines are doing all my learning/thinking. Part of slowing down is to not lose the very skills you spent time acquiring.

Here’s the template I’m starting with right now (in my vimwiki, if u read yesterdays post):

Would this make better sense as a google search? *

What is my thinking contribution to this task and why just this rather than doing the whole thing? 

Analysis of AI Performance (The "It" Errors):
    Example: The AI optimized for conciseness and modern Python practices (pathlib). 
    It didn't make a "mistake," but it failed to read my implicit need for simplicity.

Analysis of My Prompt (The "Me" Errors):
    Example: My error. I didn't specify my skill level. 
    I forgot my own rule! I assumed it would give me a simple os.listdir() and os.rename() loop. 
    It gave me an advanced (though "better") solution.

Action/prompt improvement for next time: 

This first question is in the context that I shouldn’t be wasting AI tokens I’m paying for, especially on pro models on google search esque questions. The second question is to ensure I’m not prematurely cognitively offloading work before it really needed AI and there is a concrete reason for why the alternative is much worse.

My goal overtime is to build a shorthand journaling approach once I get a feel for the types of typical errors/patterns I notice in my use of AI and essentially can just flag what happened.

This is just food for thought, I don’t even know if I’ll stick with this, but seems like a nifty experiment. The other neat part is , once you have these records in your notes , you can essentially pipe & append (bash commands) them into the same file if you wanted too and ask AI to notice any meta-patterns from that master file, while obviously take what it says with a grain of salt but if it can spot something you can’t that you can subsequently confirm - that’s a win in my books.