A new study published on arXiv investigates the impact of Large Language Models (LLMs) on real-effort tasks, commonly used in economic experiments. Researchers found that most of these tasks, which require cognitive effort and depend on actual performance, can now be accurately solved by LLMs at minimal cost. The study highlights that performance on these tasks improves with newer AI model generations, and even mid-tier models are rapidly automating them. This raises concerns about the validity of real-effort tasks in unsupervised settings, as observed performance may no longer reflect genuine human effort due to the ease of outsourcing to AI. AI
IMPACT Challenges the validity of traditional economic experiments relying on human effort, necessitating new methodologies to account for AI automation.
RANK_REASON The cluster contains an academic paper detailing research findings. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →