https://workatloss.com/I wanted to parody where AI seems to be heading. You start as a manual inference unit pressing a button for the next token, then systems take over; agents, sub-agents, approvals, swarms. It’s Universal Paperclips for the agentic AI era, and a small exploration of where predicting the next token might eventually take us. What’s the worst that could happen?