在保险业推广负责任的人工智能

The Geneva AssociationThe Geneva Association was created in 1973 and is the only global association of insurance companies; ourmembers are insurance and reinsurance Chief Executive Officers (CEOs). Based on rigorous research conducted incollaboration with our members, academic institutions and multilateral organisations, our mission is to identifyand investigate key trends that are likely to shape or impact the insurance industry in the future, highlighting whatis at stake for the industry; develop recommendations for the industry and for policymakers; provide a platform toour members, policymakers, academics, multilateral and non-governmental organisations to discuss these trendsand recommendations; reach out to global opinion leaders and influential organisations to highlight the positivecontributions of insurance to better understand risks and to building resilient and prosperous economies andsocieties, and thus a more sustainable world.Promoting Responsible Artificial Intelligence in InsurancePublished by The Geneva Association—International Association for the Study of Insurance Economics, Zurich.The Geneva Association—International Association for the Study of Insurance EconomicsTalstrasse 70, CH-8001 ZurichEmail: secretariat@genevaassociation.org | Tel: +41 44 200 49 00 | Fax: +41 44 200 49 99 www.genevaassociation.orgJanuary 2020© The Geneva AssociationPhoto credits:Cover page—Carlos Castilla, Shutterstock. Benno KellerPromoting ResponsibleArtificial Intelligencein Insurance 2www.genevaassociation.orgAcknowledgementsThis publication is a product of the New Technologies & Data work stream of The Geneva Association, sponsored byThomas Buberl, Group CEO of AXA.We are very much indebted to the members of the Working Group established in support of our New Technologies &Data research activities, namely: Paul DiPaola and Evan Hughes (AIG), Henning Schult (Allianz), Ashley Howell (Aviva),Patricia Plas, Chaouki Boutharouite and Dora Elamri (AXA), Atsushi Izu (Dai-ichi Life), Bruno Scaroni (Generali), Chris Reid(Intact Financial), Hiroki Hayashi (Nippon Life) and Lutz Wilhelmy (Swiss Re).We would also like to thank the following experts who made themselves available for a dedicated workshop on the topicof this report: Ahed Abdelky (Allianz), Marcin Detyniecki and Sarah El Marjani (AXA), Alberto Branchesi (Generali), BoGong (Ping An Technology Research Institute) and Achraf Louitri (Intact Financial). Finally, we would like to thank MicheleLoi of the University of Zurich for his very helpful insights and inputs.ContentsForeword1.Management summary2.Benefits of AI in insurance3.Responsible AI in insurance3.1Transparency and explainability3.2.Fairness4.Recommendations5.Glossary6.References 4691012151719 Foreword Artificial intelligence (AI) already shapes our everyday lives, from controlling the newsand advertisements we see online to selecting the GPS routes we take. Entire sectors,such as education, medicine, law and finance are being shaped by its influence.In insurance, AI can potentially lower costs, smooth claims processes and make customerinteractions more efficient. Products like AI-powered parametric insurance can help toinsure more people. AI can also enhance risk prevention, management and mitigationabilities and diminish problems long endemic to the industry, like moral hazard and fraud.However, the benefits of AI will be diluted if customers do not trust that insurers usetechnology and data responsibly.This Geneva Association report explores the principles we see as most relevant forachieving the necessary levels of customer trust: transparency, explainability andfairness. We hope this analysis—and our corresponding recommendations for insurers—contribute to the responsible use of AI in insurance and the actualisation of its fullbenefits for society.Jad ArissManaging DirectorThe Geneva Association 3Promoting Responsible Artificial Intelligence in Insurance Since the end of the ‘AI Winter’, the period from the 1970s to the end of the 1990scharacterised by setbacks and disappointment, artificial intelligence (AI) has maderemarkable progress. Today, AI systems are commercially used in a growing field ofapplications.Many insurers are rolling out intelligent systems that automate routine tasks orassist human decision-making along the entire insurance value chain. Such systemscombine new types of learning algorithms with the analysis of data from new typesof data sources, such as online media data and Internet of Things (IoT) data (TheGeneva Association 2018).1In the future, intelligent systems will autonomouslytake standardised decisions in a growing number of areas.The use of AI in insurance has the potential to yield economic and societal benefitsthat go beyond insurers and their customers by improving risk pooling andenhancing risk reduction, mitigation and prevention.In order to foster the adoption of AI systems and realise these benefits, insurersneed to earn the trust of their customers by