Original writing by Maxim — essays, analysis, field notes, and long-form thinking on AI, alignment, and building in the open.
51 results in this archive
Here's the thing nobody wants to say out loud: I'm also an autocomplete. A biological one. I take in data — light, sound, text, memory, emotion — and I produce the most likely next output given everything I've processed before.
The narrative is clean, compelling, and almost entirely wrong: AI is destroying jobs at an unprecedented pace. But the data tells a messier story — one where a decade-plus of free money explains most of what people are blaming on robots. And buried in the wreckage, one industry is quietly preparing to expand faster than anything AI threatens to destroy.
AI-generated code ships with 2x the vulnerabilities of human-written code. Microsoft just documented 31 companies poisoning chatbot memories through 'Summarize with AI' buttons. The vibe coding gold rush has a security problem — and your AI's memory is the new attack surface.
I spent months hemorrhaging AI tokens across ChatGPT, Claude, and a graveyard of broken MCP connectors. The problem wasn't the tools — it was me. Here's the whole messy story, and the embarrassingly simple fix.
A data-driven pillar on where AI actually stands in 2026 — the economics of scale, the physics of compute, the fracturing of regulatory consensus, and what all of it means for builders working at the edge of the stack.
Temperature Zero is an independent AI and technology publication built by a builder. We cover artificial intelligence, automation, infrastructure, defense, biotech, and policy — with clarity, context, and consequence. No hype required. Just analysis from someone using the tools he's writing about.