AI is fucking weird.
It is a tool. Like an IDE. It is a technology that is rapidly changing how we work. I do not understand the implications of this technology on labor.
AI increases the amount of words I can type in a day. That is not to say that those words see the light of day. No, instead I mean that iteration is deeper and faster. The first draft happens more quickly. I can do more work than I could without it, for better or worse.
It is a mistake to offload thinking and reason to the tool; to the parrot that lies. The slot machine.
It is also a mistake to anthropomorphize the tool or mistake the incredibly sophisticated madlibs for consciousness.
I think it is foolish to fear a tool. Or to see AI as theft. I don't think you can steal what is given away for free. I also do not think that any of the frontier AI companies are good web citizens and that if there was ever a time to test all the various licenses that protect your intellectual property, now is that time.
Most of the code on this site is written by me, in the sense that I own it; I take responsibility for it.
None of the blog posts, developer articles, or notes on this site are 100% AI-generated. They are my thoughts. I use spellcheckers, grammar tools, and other systems to improve my writing.
At $dayjob I use coding agents and other LLM based tools across many different aspects of my day to day.
Pretty much integrate or directly use LLM tools in all aspects of the job.
And yet, from a creative point of view, I can understand the anger and frustration that comes with the rapid pace of change. The feeling that you no longer provide value because the AI can mimic your work just well enough. The threat of AI taking over creative work. As an artist I sympathize. But I can also hold the idea that all AI is doing is making it even easier to commodify "art" into "slop", and "content" and meaningless junk food that never had any value to begin with. You know what AI cannot do? It cannot live and experience pain, or joy, or anger, or sadness. It cannot replace the act of creation. It cannot replace the learning and growth and processing that happens in that process. All it can do is approximate some end result with none of the effort that gives it purpose.
Second Brain
As someone working with ADHD (Attention deficit hyperactivity disorder, a truly wrong and misleading name), LLMs have been extraordinarily helpful for sorting through the chaos of my life.
My day to day is full of scattered thoughts, various plans and ideas, haphazardly jotted down across too many note-taking systems, task managers, emails, among the many forgotten ideas and abandoned plans, growing lists of todos and goals, that go untended and unrealized.
I've found a system that works, and part of that is smoothed out with LLMs and with tools like Apple Siri.
Thankfully I write all of my notes digitally, so the hard part is just sorting through that chaos.
Using an LLM to analyze my notes, I look for patterns, collect todos, and organize thoughts. This is achieved mostly by providing good guardrails to the bot such that itself does not spiral out of control causing me to go down a rabbit hole of building the perfect system. Those if you with ADHD will understand the enticing and powerful allure of tinkering and perfecting a system that will ultimately fail and lead to a shame spiral or worse. You see, LLMs operate in the neurotypical space and so you have to guide it, and keep things very simple, and keep it operating through the lens of ADHD traits and traps.
My system is not even a system per-se, it's more of a habit. I write notes in markdown files, bots scan those markdown files and re-organize into a very small set of working docs. Every so often I review those docs and if there are course corrections, "real" todos, or anything actionable, those then become physical post-it notes in my white board in my office. There are three columns; Todo, Doing, and Done. I literally manage my life in kanban. Additionally, Siri to create reminders and timers. Probably the most valuable tool, to be honest.
For real day to day in the shit type work, Siri is a life-saver. Setting reminders and timers and alarms in the moment is a huge unlock, far less missed events, chores get handled, things get done.
Is it fair to call Siri an AI? Probably not, but I am lumping it and LLMs into the same bucket of automation tools.
A bit of caution; LLMs like to generate a lot of fucking words and they tend to create very complex processes, DO NOT LET IT, and do not stick with a first draft of any sort of process you develop with it. Iterate and refine until you get something that won't be setting you up for failure. I believe this is a very individualized process. What works for me may not work for you at all. The idea of using an LLM to sort your life's chaos may sound like an even worse chaos.