How AI systems explain things to people with different levels of expertise, across three efforts:
- ELI-Why — a dataset of 13.4K “why” questions benchmarking whether LLMs tailor explanations to expertise level; GPT-4 matched intended level only 50% of the time → ELI-Why: Evaluating the Pedagogical Utility of Language Model Explanations
- CodeQL for novices — a system bridging LLMs with CodeQL so novice programmers can analyze large codebases; 3.7× accuracy gain, 32% time reduction → Synthesizing Program Analyzers to Help Programmers Answer Questions About Code
- Social discourse around papers — an online system that retrieves and labels relevant social media discussion to enrich how researchers read literature → Beyond the Page: Enriching Academic Paper Reading with Social Media Discussions
Related
USC — Adaptive Computing Experience (ACE) Lab — the lab. Research Agenda — covers the “users from different knowledge backgrounds” domain.