b'TECHNOLOGY|TECHNOLOGY TRANSACTIONSmay be changed or how much advance notice mustHowever, the use of AI in business raises a number be given. For mission-critical cloud services,of unique issues. In addition to the usual heightened focus is paid to ensuring quality,requirements for license and SaaS agreements, availability and resiliency through contractualthese AI-specific concerns include issues regarding incentives such as service levels (i.e., benchmarkscompliance with applicable laws and regulations for measuring the performance of a system), andand intellectual property rights and ownership. still more so if the provider will host customer-facingIn regard to legal compliance, 2021 saw rising aspects of the insurers business. This moreconcern that AI decision-making processes will strategic approach to cloud, and contracting fordiscriminate against members of a protected class cloud, was a key focus for 2021. because the outputs looked only at correlation instead of causal factors or relied on data that AI Adoption itself reflects a discriminatory past. The strict For years, insurance professionals have discussedregulatory environment in which insurers operate how AI will change the insurance industry. Thesemakes this point particularly noteworthy, as conversations have been mostly forward-looking,highlighted by the New York Department of but in 2021 we saw insurers and producers begin toFinancial Services recent position to address the realize this vision by increasingly delegating tasksrisk of unlawful discrimination and lack of data to AI, including tasks relating to marketing,transparency in insurance decisions. In underwriting, claims processing, customer servicetransactions, a key issue is how to secure (e.g., chatbots), and fraud detection, to name acommitments from AI providers that AI tools and few. In regard to claims processing, for instance,AI-based systems will not discriminate against ML can assist insurers by assessing damageprotected classes of applicants or employees, severity and predicting repair costs from historicalwhich is a difficult data science challenge.data, sensors and images.In regard to protection of intellectual property These AI tools can be built in-house but the(IP) rights, a key concern is that IP laws do not investment in the AI space in recent years hasprotect AI output because they protect only human produced an abundance of leading AI solutionscreativity. As a result, contractual provisions have from by technology players big and small. Webeen evolving to capture and protect value created have seen clients turn to these solutionsthrough AI. For example, the insurance business increasingly in 2021, whether licensing specificmay seek to own AI output as a trade secret, tools for specific functions, contracting forrequiring both specific language in agreements and development of AI solutions, or by outsourcingoperational procedures to preserve secrecy.wider functions to service providers able to offer dramatic productivity gains and otherData Monetizationimprovements through robotic processIf AI and the cloud are the engines driving automation (RPA) and other AI technologies.transformation for insurance businesses, data can 128|Global Insurance Industry Year in Review 2021'