An Argument Based on Data and Type Google just released WeatherNext 3, and I think it's an important model. It illustrates something the broader AI safety community was already talking about back…

Intelligence Scales Like Moore's Law, Not Just in Capability I think it's imperative to process LLM and machine learning advancement through the lens of Moore's Law. People love to say "LLMs are the…

What Is the Fallacy of Funding? There's a type of logical fallacy I want to talk about that I think deserves more recognition — particularly in the way we consume information and trace where bias…

A Case for Rights and Restraint in Embodied AI The recent unveiling of consumer humanoid robots—particularly those from companies like 1x.tech, offering household assistants for $500 per month—marks…

Intro to Mesa Operators I'll start by working through a story to help you come to grips with the idea of these "mesa operators/optimization." Let's say you're working on a data sorting and email…

On Death vs. Murder: Why We Need to Separate These Concepts Death is not unilateral. It's not something that one person experiences the same way as others. When someone dies, they're dead—death isn't…

This is an explanation of the ML/AI concept of reward hacking. Reward hacking occurs when a system finds ways to exploit the reward function defined by researchers. If an AI system is given a goal…