Study finds rapid LLM development is outrunning the research meant to explain it

The analysis warns that capability gains in large language models are outpacing scientific understanding of how they work.