Why reward models are still key to understanding LLM alignment

Interconnects AI - Why reward models are still key to understanding LLM alignment