2018/08/22 by Matthew Arnold, Arnold, Matthew, Rachel K. E. Bellamy +24 · 8 citations
Computer Science · Decision Sciences · Social Sciences · #Artificial Intelligence (cs.AI) #Business #Computer science #Computer security #Computers and Society (cs.CY) #Conformity #Declaration #Ethics and Social Impacts of AI #FOS: Computer and information sciences #Internet privacy #Knowledge management #Law #Marketing #Political science #Product (mathematics) #Risk and Safety Analysis #Service (business) #Service provider #Set (abstract data type) #cs.AI #cs.CY
paper · pdf · doi:10.48550/arxiv.1808.07261
published in arXiv (Cornell University) (Cornell University) · 31 pages
openalex publication_date 2018/08/22 · arxiv created 2019/02/07 · arxiv updated 2019/02/08 · openalex created_date 2022/08/04 · openalex updated_date 2026/08/05
Accuracy is an important concern for suppliers of artificial intelligence\n(AI) services, but considerations beyond accuracy, such as safety (which\nincludes fairness and explainability), security, and provenance, are also\ncritical elements to engender consumers' trust in a service. Many industries\nuse transparent, standardized, but often not legally required documents called\nsupplier's declarations of conformity (SDoCs) to describe the lineage of a\nproduct along with the safety and performance testing it has undergone. SDoCs\nmay be considered multi-dimensional fact sheets that capture and quantify\nvarious aspects of the product and its development to make it worthy of\nconsumers' trust. Inspired by this practice, we propose FactSheets to help\nincrease trust in AI services. We envision such documents to contain purpose,\nperformance, safety, security, and provenance information to be completed by AI\nservice providers for examination by consumers. We suggest a comprehensive set\nof declaration items tailored to AI and provide examples for two fictitious AI\nservices in the appendix of the paper.\n