My Descent into AI Madness (and Why I Love It)

As you might've guessed from my previous posts—and the fact that I'm writing yet another one about AI—I'm deep in the rabbit hole. From summarizing grocery lists to debugging my emotional state, it's taken over.

Here's the different phases I went through, from the first time hearing about this fancy new 'ChatGPT' thing, to how I'm currently using AI.

Phase 1: New kid on the block

I heard on the news, and on hackernews about this fancy new 'ChatGPT' thing that can produce text that looks like it was written by a human. Ok, fun, let's try it out. It sounded way too American, overly eager to please—clearly not human. Never mind, let's forget that this thing exists.

Phase 2: Let's use it to summarize these documents as I'm too lazy

I went to this conference and wrote some notes on the different talks, now I have to give a presentation to my team about it. Let's use ChatGPT to summarize my notes and draw conclusions. Oh wow, it did it well!

I started replacing Google. Could it help me pick games? Plan trips? Write better? Turns out: yes!

Phase 3: Self hosting a brain

Oh but now I'm using ChatGPT so much that I'm hitting the limits of the free plan. No way am I paying the monthly subscription for that. I'm a DevOps engineer, I can self host this stuff! And if it breaks, I'll call it 'feature experimentation'. Let's install Ollama + OpenWebUI, now I have my own self-hosted ChatGPT. Let's use this for everything! I'm a genius!

Phase 4: There's more than one!

Hmm, my 32GB memory isn't enough to fully load the Deepseek R1 model... No way in hell am I buying a bunch of Nvidia graphic cards so I can run them, that would cost thousands! Better buy a bunch of API credits for the OpenAI API service, that's cheaper!

But oh wait, there's other AI's out there! There's Google Gemini, Anthropic Claude, Microsoft Copilot, Deepseek, Mistral—it's a crowded party.

Phase 5: Vibecoding and letting AI control your life aka 'Nirvana'

I'm fully plugged into the matrix. I use AI all day long, from the moment I wake up: Gemini, summarize my emails and tell me what's on my calendar today. When I first make my coffee: "Hey ChatGPT, what's the best way to brew coffee?", to when I start working: GPT4.1 or GPT4o generating thousands of lines of code, things that would normally take me days to complete can now be done in hours, or mere minutes.

Researching a new topic: ingest books and articles in AnythingLLM that I'm self hosting. Using Perplexica to deep research things on the internet. Using Google Gemini on my phone to integrate with mail/calendar/messages. Using ChatGPT to brainstorm ideas and whip up working prototype projects in minutes instead of hackatons.

I've become one with the Matrix, I am Neo.

I don't just use AI—I co-author reality with it.

Disclaimer: The first draft of this post was written by me, a human. I then fed it to ChatGPT and asked it to be my editor.

Anyway, I’ll let AI proofread this. I trust it. Mostly.