Articles tagged with ai

A short conjecture about the end of consumption.

In this article I begin to explore what matters to agents in 2025.

Humanity hates intelligence. What will happen to ASI?

I make a prediction about the consequences of vibe coding.

Large Language Models have internalized HTTP. Maybe all we need to give agents agency is a simpler HTTP protocol.

Your system prompt WILL leak. What are you going to do about it?

Vibe coding makes software expendable in a Hollywood way.

This post compares policyai’s benchmark on Sonnet 4.5 compared to the new Haiku 4.5 model.

In which I argue that left-associative languages might be better for LLMs than other languages.

This post introduces the langcontinuation framework, that allows for writing the code for both live and batch processing just once.