2026 and the rise of AI

2026 has been a really crazy year! I started the year writing most of my own code, and using AI only to ask some syntax-specific questions or to gain a better understanding of the API or SDK I was trying to use. Anthropic and Claude Code really changed everything. For the first time, I was seriously impressed with what AI was able to do. I started out just sending Claude off to do my monkey-task work, and what would have taken me hours to write was now done in minutes. I have coded less and less, and produced more than I ever have this year.

I have seen developers blindly follow what AI tools like Claude Code tell them to do, causing security issues and production outages all because they didn’t stop to think about what the agent was telling them to do. There are professional examples of this, GitHub being one of the prime examples.

In an all-hands meeting at the Arbinger Institute, Chip Huth made a comment that has stuck with me: “I don’t want to outsource my critical thinking to AI.” I have been reflecting on this statement for months now, and on how it fits with what I want to use AI for. Critical thinking has gotten me to where I am now in life, and I don’t want that skill to atrophy. So as I continue to experiment with AI, that thought will keep ringing in the back of my head.