Why this blog exists
We are entering a phase where AI can generate words, code, and opinions faster than any human can verify them. The bottleneck is no longer production — it is judgment.
I keep this blog as one place where my thinking is slower, explicit, and accountable. Not optimized for an algorithm. Not optimized for engagement. Optimized for clarity.
The core motivation: stay human in an AI-amplified world
AI has become a daily collaborator. It accelerates writing, debugging, and exploration — but it also raises a harder challenge: modern AI architectures are evolving quickly, and systems can now produce code and content beyond what we fully understand, or feel comfortable trusting by default.
That means the job has changed. We need to learn continuously, evaluate outputs rigorously, and actively guide and guardrail AI behavior instead of passively consuming it.
The risks still show up in familiar ways:
- Deskilling — letting AI do too much hard thinking until our own mental models weaken.
- Sycophancy — models agreeing too easily, turning bias into false confidence.
So this blog is a counterweight. Writing in public forces me to structure arguments, state assumptions, and make claims I can revisit later.
My working principle with AI
I treat AI as a sparring partner, not an oracle.
- I ask it to challenge assumptions.
- I ask for failure modes, not just polished answers.
- I define the thesis before opening the chat.
- I keep human responsibility for direction, quality bars, and safety constraints.
In short: AI accelerates execution; humans own direction.
Why a personal blog, not social media
Social feeds are built for velocity and reaction. A personal blog is better for durable thought.
A blog gives me:
- Sovereignty: I own the content, source, and archive.
- Continuity: ideas can evolve over months, not vanish in 24 hours.
- Signal over noise: no recommendation engine deciding what matters.
I want a digital space that behaves more like a notebook than a casino.
The infrastructure is part of the philosophy
This site is intentionally simple and transparent:
- Framework: Next.js 15 (App Router)
- Styling: Tailwind CSS v4.1
- Hosting: GitHub Pages (static export)
- Content: Markdown + frontmatter
No tracking scripts. No engagement gimmicks. Just files, version control, and ideas.
I also enjoy adding a bit of internet weirdness — like the Oneko cat on the homepage — but even that has a purpose: to keep the web personal, not purely optimized.
What you can expect here
Most posts will focus on:
- Human–AI collaboration in real engineering workflows
- Agentic systems, reliability, and safety trade-offs
- How to avoid cognitive outsourcing while still using AI effectively
- Practical tools and experiments from my own workflow, including what worked, what failed, what to delegate, what to stay skeptical about, and the practical dos, hows, and avoidable mistakes
I also use this blog to explore personal growth in AI-assisted learning: how to learn faster with better retention, and which education theories can evolve for the next generation.
I care less about hype cycles and more about one question:
How do we use AI to extend human capability without eroding human agency?
Final note
This blog is my long-term memory and my intellectual test bench.
If AI is the new engine, this is my steering wheel.
Welcome — and thanks for reading.