Person
Collin Burns
AI researcher on language-model evaluation, alignment, and empirical assessment of model capabilities.
Why this matters
Burns treats benchmarks as stress tests—capabilities that look fluent until a systematic probe finds the fracture.
