Post by Dhruv Tandon

Stanford researchers: "We must study whether AI will replace workers." Also Stanford researchers: "GPT-4.1, analyze these 17,000 papers for us." To determine which jobs AI could automate, they had GPT-4o-mini process 5,000 startups and 17,000 research papers The AI basically did all the boring parts—reading, categorizing, matching companies to job tasks. You know, research assistant stuff. While studying whether AI would replace research assistants. It's ironically meta - like hiring a robot to write a report on whether robots will take your job. The robot says no, prints its own termination notice. But here's the thing—it actually worked great. The humans handled the thinking parts, the AI handled the reading-thousands-of-documents parts. Nobody had to spend six months in spreadsheet hell. Meanwhile everyone's writing think pieces about the "future of work" like it's some distant philosophical question. But these researchers just... did it. No hand-wringing, no 50-page framework. They had too many documents, an LLM was right there, problem solved. This is how it'll actually happen. Not some master plan to eliminate tasks, just economic gravity. Someone has 17,000 papers to read, an LLM exists, the choice makes itself. People will reach for these tools the same way they reached for calculators or spreadsheets—because why wouldn't you?