Writing

The Next Token Prediction Game – At their core, large language models (LLMs) are pattern-matching systems. They predict the next word in a sequence based on statistical patterns learned from billions of text examples. There's no explicit module for logic or mathematics; everything emerges from a single mechanism: analyze input, compute probabilities for possible next tokens, and select the most likely one.

Two wildly different AI takes appeared on my feed within minutes of each other. As someone working in computer vision and physics, I've seen LLMs fake competence brilliantly—but also watched them crumble when I actually know the domain. The real puzzle? How do you measure intelligence when you might need a more intelligent system to do the judging.

I built a Monte Carlo simulation tool because I was tired of making terrible decisions with my own money. Here's what I learned about uncertainty, assumptions, and asking better questions.

Trade debugging code for debugging strategy. After 5 years deep in the trenches at rabbitAI, I've finally accepted that I can't code my way out of every problem anymore... well, most days I accept it.