IRS

The Flaw in Defensive Tech Regulation We spend too much time worrying about what we want to stop. Society obsesses over potential catastrophes, algorithmic bias, and job losses. While these fears merit attention, this purely defensive posture leaves us directionless. We know what we want to avoid, but we have no idea what we want to build. True progress requires a proactive vision. The Two Pillars of Public Machine Learning Solving public problems with machine learning requires clearing two massive hurdles. First, you need substantial funding. Even today, running complex models demands expensive compute resources. Second, the problem must be legible to the machine. Governments struggle here because they rarely organize their vast data archives into clean training sets. Without structured data, even the most expensive system is useless. Why Your Taxes Should Be Automated Consider the IRS. The agency already holds your income data and co-writes the tax code. There is no structural reason why a public large language model cannot file your taxes for you. Instead of paying private accountants or expensive software companies, you could resolve your taxes through a brief conversation with a public AI. Demanding a Theory of Public Good Moving past a defensive mindset means establishing a clear theory of public utility. We need a system that acts as a personal concierge, guiding you through complex public services because it understands your specific situation. Securing this future requires shifting our focus from merely preventing harm to actively demanding public benefits.