Joe Alcantara
AI Safety Researcher exploring the epistemic foundations of AI and infrastructure sovereignty
The Agency Trap: Why Automating Judgment Creates Catastrophic Risk
We have a habit of trading human judgment for a few milliseconds of efficiency. From the cockpits of commercial airliners to the high-frequency trading floors of Wall Street, the pattern is the same: we take a human “safety check,” replace it with an algorithm, and hope the speed is worth the risk. But as we move toward Agentic AI (systems that don’t just process data but take autonomous actions across multiple domains), we aren’t just automating tasks. We are automating away the very “sanity checks” that prevent systemic collapse. ...