Researchers from the University of Texas at Dallas, Tencent AI Lab, and the University of Southern California have introduced DSBench, a comprehensive benchmark designed to evaluate the performance of large language models (LLMs) on real-world data science problems. The benchmark reveals significant limitations in current data science agents, with GPT-4o scoring only 28% accuracy compared to 66% achieved by human experts. This highlights a substantial gap in the capability of AI systems to handle complex data analysis and modeling tasks, presenting an exciting challenge for future advancements in AI.