19DECEMBER 2021journey, mainly based on traditional insurance risk and customer data enriched with alternative demographic and medical data (See Figure 1). If insurers are still in the data accumulation phase, we enable them with evidence-based research and proxies for underwriting assessment. To maximise efficiency, these AI models have also been fully integrated into Magnum Pure, our Life &Health automated underwriting solution platform, to orchestrate and utilise both Human and AI intelligence at scale. With Magnum Pure, this expert-backed, AI, and machine-learning enabled risk assessment solution can be easily deployed across multiple distribution channels. We also support insurers on their data-driven transformation across the value chain providing risk insights with pricing analytics, claims analytics, and portfolio deep dives. As insurers access more (digital) data points from their customers, opportunities exist for them to connect and analyse the potential for offering more customised protection solutions with other external data sets (from their digital partners or open-source).In Swiss Re's experience, system design and management often fall short when insurers attempt to implement machine intelligence (MI) into existing cross-functional processes. Insufficient resources are dedicated to integrating models and algorithms into existing workflows to transform the customer journey (see Figure 2). In an interview with Swiss Re Institute, one insurer seeking to eliminate unnecessary underwriting questions said it leveraged banking transaction data to offer accelerated underwriting to prospects. The MI-enabled underwriting model performed well in classifying individuals into the relevant segments. The marketing department, however, could not leverage these results, as it did not invest in a propensity-to-buy exercise to take full advantage of the new underwriting system.Many other hurdles exist in the industry to scale deployment of MI solutions today, such as limited innovation management and empowerment, slow feedback loops to validate model performance (such as for mortality risk), lack of strategy to acquire and retain data, partners, and talent, the lack of long term clarity from regulation on AI applications. Etc. The list can go on, but what we need is a holistic view to make MI successful. The criteria to evaluate a new process should include the integration of direct (development and running - i.e. ML Ops) and indirect (organisational and opportunity) benefits and costs.While chief data officers and scientists have become common-place at insurers, inchoate firm-wide data strategies and inadequate underlying technology can hinder their effectiveness. System design, deployment plans, and success criteria should focus on business workflow context, decision support, and enterprise productivity. It may mean streamlining and redesigning workflows. Insurers must develop standards of syntax, semantics, and terminology, and clean and fix data at the source as much as possible. Initiatives to clean up the data should be led outside of IT, but with a strong operational relationship with IT. Regulatory risks regarding tech-linked innovation in insurance, in particular around data privacy and use (i.e. fairness and transparency), also need to be considered. Many machine intelligence approaches require large amounts of high-quality data that are complete, clean, and timely to train algorithms. Enterprise-scale transformative benefits could be delivered with a production-ready data strategy and more investment in data engineering (See Figure 3). If deployed correctly, models/algorithms can deliver a substantial return on investment. Without said capabilities, the value from MI integrated solutions cannot make an impact at scale across the firm relative to existing human-centric processes.Most importantly, a machine intelligence project also needs clear and understandable communication across all facets, to secure senior management buy-in and funding. Focus on retaining talent; create near-term opportunities and incentives to apply machine intelligence in a way that interests skilled developers and data scientists. Finally, executives must be continually challenged to translate technical metrics into financial return metrics. In summary - to successfully transform their enterprises with machine intelligence-enabled technology, we recommend that insurers focus on corporate data strategy and other non-model characteristics of end-to-end enterprise deployment. any machine intelligence approaches require large amounts of high-quality data that are complete, clean, and timely to train algorithmsJonathan Anchen
< Page 9 | Page 11 >