A living survey of benchmarks that compare large language models with humans.

At the core of the Hugging Face hacking incident were a series of human choices that traded security for speed.

Research suggests that the combination of incentives to publish and the use of large language models will lead to more papers, but they will be less refined.