Agentic AI: Productivity or Laziness

So, I have to confess; I only managed to get this page up using an agentic AI (Claude Code). For those unfamiliar, agentic AI refers to AI systems that can work independently on complex tasks, making decisions and taking actions rather than just responding to prompts. It's not because I don't know how to do it, but the fact that I've started working on a new web-site at least a dozen times. Every time I've been faced with so many small tasks and side quests. Every small thing like color, design, how to serve it etc. have made the journey feel too long to complete.

When I use "AI" I generally mean generative AI as in LLM and diffusion based image models. I dont think we have AI yet, as in AGI (Artificial general intelligence), and will not have it in a long time.

Laziness and Brain Rot

I don't like AI generated content. I will never use AI generated content as if it is my own. I've started to notice newspapers use AI to write articles. I've received tickets at work that clearly have been written by ChatGPT. I see AI-generated images pop up everywhere. YouTube is offering up travel guides with AI text, with AI-curated (and sometimes generated) images. I don't like it. It feels off. Why should I read something if someone did not care enough to write the content themselves, why should I consume it?

Large language models have never visited a country like Malaysia. It's never felt the humidity of Kuala Lumpur or bargained in a night market. It doesn't know the smell of durian or the sound of call to prayer echoing through the streets. Yet somehow it's writing travel guides about places it will never experience. That's the problem with AI-generated content - it lacks the authenticity that comes from lived experience.

Productivity

Using an agentic AI feels like cheating. The first implementation of this site was "written" in a couple of hours. I described what I wanted, I asked questions and I delegated most of the small side quests to the agent. I only made a handful of decisions. The ones I felt were important. The look and feel. The architecture. It was a great experience, and I was able to come further in the process than I ever had. I'm even writing this post the same evening. I never came this far before.

Finding the Line

So, instead of looking at AI as brain rot or cheating, I think we should look at it as an enabler. Let the AI take care of all the things that are in the way of your creativity. The question isn't whether to use AI, but where to draw the line.

I'll let AI generate build scripts and handle configuration files. I won't let it write my thoughts or pretend to have experiences it's never had. I'll use it to prototype interfaces quickly, but I'll make the design decisions myself. It can suggest code patterns, but I'll decide what problems are worth solving.

The key is maintaining authorship over what matters. If the content, the vision, the creative decisions are yours, then using AI for the grunt work isn't laziness - it's just good delegation. The danger comes when we start delegating the thinking itself.