[Self-Introduction] The story of how I, who couldn't become a prompt engineer, started an experiment to make AI my underling

Hello, nice to meet you.

Born and raised in the Showa era.
My hobby is bench pressing.

I am a permanent low-level employee at my company.
At home, I get scolded for "always staring at my smartphone,"

a man who is, by my own admission and others', completely hopeless,
I go by Mada-o.

First of all, I should clarify that I am not good with computers.
I type by poking the keyboard with my right index finger, and just hearing about programming, English, or difficult commands gives me a mental muscle ache.

Even so, I am currently experimenting with reducing the tedious groundwork for my job and side hustles by leaving difficult settings and detailed tasks to experts and AI.

It's not that AI does everything for me on its own. Humans handle the fine-tuning and final checks, but I have been able to let go of the "tasks that make you tired before you even start," such as gathering information, organizing, brainstorming, and drafting.

I gave up on doing difficult settings myself, and instead, I have an expert who breaks down tasks into "things to leave to AI" and "decisions for humans to make" set up an environment that works on my behalf.

I just ask my AI underling in Japanese, "Do this for me." I will be revealing the process of this experiment here without holding anything back.

The past where I challenged online side hustles with muscle and failed miserably

Until a little while ago, I was fired up by anxiety about the future, thinking, "I'm going to make money with an online side hustle." I heard that "this is the era of ChatGPT," so I tried logging in with my clumsy hands.

However, I hit a major wall there. Difficult prompts (commands) that look like spells created by smart people!
If you don't enter these correctly, the AI won't work as you expect.
I got errors over and over, and my brain's memory completely short-circuited.

"Is my muscle worth absolutely nothing in the world of AI?"
I can't count the number of nights I spent in despair while lifting dumbbells.
Yes, I could not become a prompt engineer who gives orders to AI.

"I should just be on the side that throws work at AI"

That was when I encountered the latest technology called autonomous AI agents.

With the help of an expert, I had an environment prepared where AI could organize themes and trends based on publicly available information on my behalf.

I just ask in Japanese, "Check what themes are being read in the AI world right now and summarize them in a notebook." After that, the AI organizes the information in the background and delivers it to me in a form that is easy for me to judge.

At this moment, I realized something.
What we should do is not become experts in AI. It is to separate the tasks we leave to AI from the decisions we make as humans. As long as you don't get that wrong, even if you are bad at computers, AI will properly become your underling.

What I will be posting on this account

I named this account "Showa Mada-o's AI Underling Experiment Lab."
I will not provide any difficult technical explanations of AI.

Instead, I will deliver the following types of experimental results in a gritty way.

• Verification of how much I was able to focus on my decision-making tasks by leaving tedious information gathering and organization to my AI underling

• A realistic story of the failures a Showa-era guy like me stumbled upon when introducing AI into my work

• Tips for sorting out 'which tasks to delegate to AI and which tasks humans should hold onto'

Let's stop being exhausted every day by being used by AI, and instead move to the side where you use AI as a subordinate and focus on the work you really should be doing.

On my official LINE account, I am giving away a free 'Checklist for Tasks to Delegate to AI vs. Decisions Humans Should Hold Onto' that I actually use.

If you don't know which parts of your work you should delegate to AI, please start by using this checklist to organize your own work.

First, I will use my own work as a test subject.

What I delegated to my AI subordinate.
Where I failed.
What I ultimately decided for myself.

I will keep a record of that here.

Today's sorting

Things to delegate to AI:
Information gathering, brainstorming, structuring, and drafting outlines.

Things for humans to hold onto:
What to publish under my own name.
How much to delegate to others and where to make decisions myself.

Mada-o's failure:
Trying to become an AI expert and diving into difficult prompts from the start.
As a result, my brain felt like it had muscle aches before I even started.

What to try next:
I will break down my work one by one and find the groundwork that can be delegated to my AI subordinate.