Recently, researchers from the Chinese Academy of Sciences and Yale University proposed a new way of using large language models for reasoning and problem-solving.
In their paper, the team experimented with letting large Large Language Models reference similar problems and solutions when being asked a question to "enhance the complex reasoning ability of LLMs", and found that this new approach enjoys "a substantial improvement over the baselines" in areas such as Shortest-path Reasoning and Creative Writing.
Sources:
https://arxiv.org/abs/2310.03965