Mechanistic interpretability is the program of explaining what AI systems are doing in terms of their internal mechanisms. I analyze some aspects of the program, along with setting out some concrete challenges and assessing progress to date. I argue for the importance of propositional interpretability, which involves interpreting a system's mechanisms and behavior in terms of propositional attitudes: attitudes (such as belief, desire, or subjective probability) to propositions (e.g. the proposition that it is hot outside). Propositional attitudes are the central way that we interpret and explain human beings and they are likely to be central in AI too. A central challenge is what I call thought logging: creating systems that log all of the relevant propositional attitudes in an AI system over time. I examine currently popular methods of interpretability (such as probing, sparse auto-encoders, and chain of thought methods) as well as philosophical methods of interpretation (including those grounded in psychosemantics) to assess their strengths and weaknesses as methods of propositional interpretability.
Propositional Knowledge Representation and Reasoning in Restricted Boltzmann Machines
The Australian E-health research centre, CSIRO(澳大利亚电子健康研究中心(CSIRO));Level 5, UQ Health Science Building(昆士兰大学健康科学大楼5层);Brisbane, QLD 4026, Australia(布里斯班,昆士兰州4026,澳大利亚);CSIRO(英联邦科学与工业研究组织)
How Do LLMs and VLMs Understand Viewpoint Rotation Without Vision? An Interpretability Study
School of Computer Science and Technology, Beijing Institute of Technology(北京理工大学计算机科学与技术学院);Key Laboratory of Computing Power Network and Information Security, Ministry of Education, Shandong Computer Science Center (National Supercomputer Center in Jinan), Qilu University of Technology (Shandong Academy of Sciences)(教育部计算功率网络与信息安全部重点实验室,山东计算机科学中心(济南国家超级计算中心),齐鲁工业大学(山东省科学院))
Morphologic for knowledge dynamics: revision, fusion, abduction
LTCI, Télécom ParisTech, Université Paris-Saclay(LTCI,巴黎高等电信学院,巴黎萨克雷大学);Université Paris-Dauphine, PSL Research University, CNRS(巴黎第九大学,巴黎文理研究大学,法国国家科学研究中心);Universidad de Los Andes(洛斯安第斯大学);Universidad Industrial de Santander(桑坦德工业大学)
AI Readiness in Healthcare through Storytelling XAI
Center for Artificial Intelligence in Public Health Research (ZKI-PH) at Robert Koch Institute(罗伯特·科赫研究所公共卫生研究人工智能中心);Free University of Berlin(柏林自由大学)
针对医疗 AI 落地中的信任与差异化解释需求,本文提出结合多任务蒸馏和可解释性技术的叙事式 XAI,以支持复杂深度模型并提升专家信任。
Explainable Artificial Intelligence: a Systematic Review
School of Computer Science, College of Science and Health, Technological University Dublin(都柏林理工大学科学与健康学院计算机科学学院);Technological University Dublin(都柏林理工大学)
Interplay between Federated Learning and Explainable Artificial Intelligence: a Scoping Review
Simula Metropolitan Center for Digital Engineering(西米拉数字工程中心);Karlsruhe Institute of Technology(卡尔斯鲁厄理工学院);University Hospital Bonn(波恩大学医院);Norwegian University of Science and Technology(挪威科技大学);Charité - Universitätsmedizin Berlin(柏林夏里特医学院);Birmingham City University(伯明翰城市大学);ETH Zurich(苏黎世联邦理工学院);Technical University of Munich(慕尼黑工业大学)
Enabling Collaborative Clinical Diagnosis of Infectious Keratitis by Integrating Expert Knowledge and Interpretable Data-driven Intelligence
Sir Run Run Shaw Hospital, Zhejiang University School of Medicine(浙江大学医学院附属邵逸夫医院);College of Computer Science and Technology, Zhejiang University(浙江大学计算机科学与技术学院);Key Laboratory for Corneal Diseases Research of Zhejiang Province(浙江省角膜病研究重点实验室);School of Software Technology, Zhejiang University(浙江大学软件学院)
Explainable Artificial Intelligence (XAI) on TimeSeries Data: A Survey
ENSTA Paris, Institut Polytechnique Paris(巴黎高等先进技术学院,巴黎综合理工学院);Inria Flowers(法国国家信息与自动化研究所Flowers团队);Renault(雷诺公司);TECNALIA(TECNALIA研究中心);Basque Research and Technology Alliance (BRTA)(巴斯克研究与技术联盟);University of the Basque Country (UPV/EHU)(巴斯克大学)
Tensor Networks for Explainable Machine Learning in Cybersecurity
Multiverse Computing;Department of Basic Sciences, Tecnun - University of Navarra(基础科学系,塔努大学);Donostia International Physics Center(Donostia国际物理中心);Ikerbasque Foundation for Science(Ikerbasque科学基金会)
Applications of Explainable artificial intelligence in Earth system science
Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai)(南方海洋科学与工程广东省实验室(珠海));Guangdong Province Key Laboratory for Climate Change and Natural Disaster Studies(广东省气候变化与自然灾害研究重点实验室);Max Planck Institute for Biogeochemistry(马克斯·普朗克生物地球化学研究所);Shanwei Meteorological Service(汕尾市气象局);College of Computer Science and Technology, Changchun Normal University(长春师范大学计算机科学与技术学院);ELLIS Unit Jena(ELLIS耶拿单元)
Enabling Machine Learning Algorithms for Credit Scoring -- Explainable Artificial Intelligence (XAI) methods for clear understanding complex predictive models
Warsaw University of Technology(华沙理工大学);University of Warsaw(华沙大学);Data Juice Lab sp. z o.o.(Data Juice Lab 有限责任公司);Polish Credit Bureau S. A. (Biuro Informacji Kredytowej S.A.)(波兰信贷局)
What the Waveform Knows: Transparent-first Speech and Audio Intelligence with Caption Studio
Faculty of Science, Agriculture, and Engineering, Newcastle University Singapore(新加坡纽卡斯尔大学科学、农业与工程学院);School of Information and Control Engineering, Qingdao University of Technology(青岛理工大学信息与控制工程学院);Department of EEE, Amrita School of Engineering, Amrita Vishwa Vidyapeetham(阿姆瑞塔工程学院电气与电子工程系,阿姆瑞塔大学)
Explainable AI meets Healthcare: A Study on Heart Disease Dataset
Pandit Deendayal Petroleum University(潘迪特·迪恩达亚尔石油大学);Confirm SFI Centre for Smart Manufacturing(Confirm SFI智能制造中心);Data Science Institute(数据科学研究所);National University of Ireland Galway(爱尔兰国立大学戈尔韦分校)