How to make System 1 LLMs show System 2 competencies?

To make Large Language Models (LLMs) show System 2 competencies, researchers suggest using a combination of techniques such as providing concrete instructions, using domain-specific knowledge, and creating a sense of accountability. This can help LLMs mimic human-like reasoning and problem-solving abilities. The authors of the study propose a framework for measuring System 2 competencies in LLMs, which involves evaluating their ability to reason, plan, and justify their decisions. They also discuss the importance of considering the context and nuances of language when designing LLMs to exhibit System 2 competencies.