"DeepMind's 'Superhuman' AI Revolutionizes Fact-Checking and Accuracy"
Google DeepMind's new AI system, SAFE, outperforms human fact-checkers in evaluating the accuracy of information generated by large language models, with a 72% match rate and significant cost savings. However, experts question the "superhuman" label, emphasizing the need for benchmarking against expert human fact-checkers. The system's open-sourced code and dataset provide transparency, but further details on the human baselines used in the study are needed. While AI fact-checking tools like SAFE could help combat misinformation, rigorous benchmarking against human experts is crucial for measuring true progress.