Researchers have developed a system that uses AI to identify flaws in summaries written by other AI models. The AI-generated critiques are shown to humans, who are more likely to notice errors and inaccuracies when reviewing the summaries with the AI's feedback. The study found that larger AI models are better at self-critiquing and that this ability improves with scale, suggesting potential for using AI-assisted supervision in complex tasks. --- Why it matters: This matters because it shows promise for improving the reliability of AI systems by leveraging their own abilities to identify flaws and inaccuracies. This could lead to more accurate and trustworthy results from AI models, particularly in applications where precision is crucial. Source: https://openai.com/index/critiques