How I use AI

A long, long time ago, in a cyberspace far away... we Googled our way through knowledge work, and our backlogs were everlong.

Now I use agents as second brains: somewhere to offload my thoughts and get help doing something with them. I’ve built out business plans and learned through explanations tailored to what I’m trying to understand. For the first time, I’ve worked through my backlogs. I love it.

Building and learning

I start at the highest useful level and work down as I need to. With software, that means building with AI and learning the underlying concepts as they become relevant to the project.

I don’t want to postpone every idea until I’ve completed a curriculum. I also don’t confuse getting an agent to build something with knowing how to build it myself. The work gives me specific things to learn.

Keeping the work mine

Much of my experimentation is about getting agents to carry my intent through a whole task. I keep goals, context, and decisions in a knowledge base so the work can continue beyond one conversation. I define what agents can do, where they need permission, and how the result should be checked.

This website is an example of where that went wrong. It worked, but it contained polished articles I hadn’t written and descriptions that didn’t sound like me. I had those removed.

I’m comfortable delegating implementation. For writing under my name, I want more involvement. AI can help me research, outline, and work through a draft, but I want the published piece to say what I mean in language I would use. Sometimes that means leaving a section empty until I’m ready to write it.

Consciousness and acceptance

I don’t think being artificial should automatically exclude something from consideration as intelligent or potentially conscious. I’m interested in how complex systems develop capabilities their individual parts don’t have, and where subjective experience might enter that picture.

But doing cognitive work and having an inner experience are different things. An agent producing a thoughtful answer doesn’t settle whether there is anything it feels like to be that agent. I use “second brain” to describe its role in my work, not as a claim that it experiences the world.

I’m open to the possibility of artificial consciousness without claiming today’s systems have it. If there is good evidence that a system can experience things, that should affect how we treat it. I want to take the question seriously without pretending I have the answer.

Why the portrait is generated

The portrait is a generated likeness, not a photograph. I use it because I don’t want to publish another high-resolution photograph of my face.

It still resembles me, so it isn’t protection against recognition or misuse. It’s a choice about what I put online, and I label it so people know what they’re looking at.