🧵 5. The study challenges the belief that bigger AI models equate to better performance, highlighting the gap between complexity and consistency in model outcomes.