Prompting Science Report 1: Prompt Engineering is Complicated and Contingent
This highlights the complexity and contingency in benchmarking and prompt engineering for business, education, and policy leaders, emphasizing that these methods are not universally applicable.
The report demonstrates that there is no single standard for measuring LLM benchmark performance, with the choice of standard significantly impacting results, and that prompting approaches such as politeness or answer constraints can unpredictably help or harm performance depending on the context.
This is the first of a series of short reports that seek to help business, education, and policy leaders understand the technical details of working with AI through rigorous testing. In this report, we demonstrate two things: - There is no single standard for measuring whether a Large Language Model (LLM) passes a benchmark, and that choosing a standard has a big impact on how well the LLM does on that benchmark. The standard you choose will depend on your goals for using an LLM in a particular case. - It is hard to know in advance whether a particular prompting approach will help or harm the LLM's ability to answer any particular question. Specifically, we find that sometimes being polite to the LLM helps performance, and sometimes it lowers performance. We also find that constraining the AI's answers helps performance in some cases, though it may lower performance in other cases. Taken together, this suggests that benchmarking AI performance is not one-size-fits-all, and also that particular prompting formulas or approaches, like being polite to the AI, are not universally valuable.