Language models (LMs) have achieved impressive accuracy across a variety of tasks but remain vulnerable to high-confidence misclassifications, also referred to as unknown unknowns (UUs). These UUs cluster into blind spots in the feature space, leading to significant risks in high-stakes applications. This is particularly relevant for smaller, lightweight LMs that are more susceptible to such errors. While the identification of UUs has been extensively studied, their mitigation remains an open challenge, including how to use identified UUs to eliminate unseen blind spots. In this work, we propose a novel approach to address blind spot mitigation through the use of intelligent agents -- either humans or large LMs -- as teachers to characterize UU-type errors. By leveraging the generalization capabilities of intelligent agents, we identify patterns in high-confidence misclassifications and use them to generate targeted synthetic samples to improve model robustness and reduce blind spots. We conduct an extensive evaluation of our method on three classification tasks and demonstrate its effectiveness in reducing the number of UUs, all while maintaining a similar level of accuracy. We find that the effectiveness of human computation has a high ceiling but is highly dependent on familiarity with the underlying task. Moreover, the cost gap between humans and LMs surpasses an order of magnitude, as LMs attain human-like generalization and generation performance while being more scalable.
Blind Spot Navigation in Large Language Model Reasoning with Thought Space Explorer
Clemson University(克莱姆森大学);Université de Montréal(蒙特利尔大学);Center for Advanced AI, Accenture(Accenture高级人工智能中心);The University of Alabama(阿拉巴马大学);University of Kansas(堪萨斯大学);Arizona State University(亚利桑那州立大学)
On the Blind Spots of Model-Based Evaluation Metrics for Text Generation
Univ. of Washington(华盛顿大学);Johns Hopkins Univ.(约翰斯·霍普金斯大学);Shanghai Jiao Tong Univ.(上海交通大学);Carnegie Mellon Univ.(卡内基梅隆大学);New York Univ.(纽约大学);Mass. Institute of Technology(麻省理工学院)
The Truncation Blind Spot: How Decoding Strategies Systematically Exclude Human-Like Token Choices
Department of Statistics, Ludwig Maximilian University, Munich, Germany(统计系,路德维希-马克西米利安大学,慕尼黑,德国);Munich Center for Machine Learning (MCML)(慕尼黑机器学习中心(MCML))
Catching The Correct Answer Trap: Characterising AI Tutor Blind Spots When Analysing Student Reasoning
Department of Computer Science, University College London, UK(英国伦敦大学学院计算机科学系);Centre for Artificial Intelligence, University College London, UK(英国伦敦大学学院人工智能中心)
Mind the Gap: Exposing LLM Translation Blind Spots Using the AlphaMWE Multilingual Parallel Corpus
Leiden University(莱顿大学);Leiden University Medical Centre(莱顿大学医学中心);Institute of Computer Science PAS, Warsaw(波兰科学院计算机科学研究所);University of Ljubljana(卢布尔雅那大学);University of Granada(格拉纳达大学);Technical University of Munich(慕尼黑工业大学);Heinrich Heine University Düsseldorf(杜塞尔多夫海因里希·海涅大学);Google DeepMind(谷歌DeepMind)
Good human thinking populates AI's semantic blind spots
Georgetown University(乔治城大学);Washington and Lee University(华盛顿与李大学);The Wharton School, University of Pennsylvania(宾夕法尼亚大学沃顿商学院);University of San Francisco(旧金山大学);Neag School of Education, University of Connecticut(康涅狄格大学尼格教育学院);Pennsylvania State University(宾夕法尼亚州立大学);University of Pennsylvania(宾夕法尼亚大学)