To fix the way we test and measure models, AI is learning tricks from social science. It’s not easy being one of Silicon Valley’s favorite benchmarks. SWE-Bench (pronounced “swee bench”) launched in ...

Researchers are racing to develop more challenging, interpretable, and fair assessments of AI models that reflect real-world use cases. The stakes are high. Benchmarks are often reduced to leaderboard ...

Artificial intelligence has traditionally advanced through automatic accuracy tests in tasks meant to approximate human knowledge. Carefully crafted benchmark tests such as The General Language ...

AI agents are becoming a promising new research direction with potential applications in the real world. These agents use foundation models such as large language models (LLMs) and vision language ...

They could offer a more nuanced way to measure AI’s bias and its understanding of the world. New AI benchmarks could help developers reduce bias in AI models, potentially making them fairer and less ...