Rubrik's Cube: Testing a New Rubric for Evaluating Explanations on the CUBE dataset

The performance and usability of Large-Language Models (LLMs) are driving their use in explanation generation tasks. However, despite their widespread adoption, LLM explanations have been found to be unreliable, making it difficult for users to distinguish good from bad explanations. To address this issue, we present Rubrik's CUBE-an education-inspired rubric and a dataset of 26k explanations, written and later quality-annotated using the rubric by both humans and six open-and closedsource LLMs. The CUBE dataset focuses on two reasoning and two language tasks, providing the necessary diversity for us to effectively test our proposed rubric. Using Rubrik, we find that explanations are influenced by both task and perceived difficulty. Low quality stems primarily from a lack of conciseness in LLMgenerated explanations, rather than cohesion and word choice. The full dataset, rubric, and code are available at https://github.com/ RubriksCube/rubriks_cube . 1 See StackOverflow's policy on the use of ChatGPT and other LLMs: https:// COMPONENTS DIMENSIONS necessary parts of an explanation necessary qualities of a good explanation Typology of Explanations Language Content Typ1. COMMENTARY 1.a) Action Grammaticality Conciseness 1.b) Reason Word Choice Appropriateness Cohesion Coherence Typ2. JUSTIFICATION 2.a) Evidence Plausibility Typ3. ARGUMENT 3.a) Affective appeal(s) and Qualifier(s) Stance Clarity