(Human) Attention Is (Still) All You Need: Human oversight makes AI-assisted social science reliable

In three linesExperimental study across 280 runs showing unconstrained LLMs fail in 72% of economic research tasks versus 16% with Human-in-the-Loop Economic Research (HLER). Architecture enforces pre-commitment, decision sequencing, and three human gates. Significant result (p<0.001): humans validate, LLMs reason but do not execute data work.

Summary generated by Claude — human-verified