Reducing the black box problem by analyzing the AI prediction results process.
NH Nonghyup has numerous domestic and international success stories, including KB Insurance.
On the 5th, SAS Korea announced the latest version of its enterprise analytics platform, SAS Viya, which enhances AI interpretability and transparency. This allows both data scientists and general users to better understand and adjust the operating principles of machine learning models, resulting in more reliable analysis results.

According to a domestic AI trust perception survey conducted by the Electronics and Telecommunications Research Institute (ETRI), overall trust in AI was low at 4.7 out of 7, highlighting concerns about bias in algorithm design and accountability. While cutting-edge machine learning technologies offer more accurate analytical results, their complex operating principles pose a "black box" problem, making it difficult to clearly explain how specific predictions were derived. As AI applications expand to areas such as loan screening, job recommendations, and medical diagnosis, the importance of the ability to interpret AI models and derive transparent results is growing.

Accordingly, SAS provides new features through the latest version of SAS Viya, including ▲ transparent AI decision-making ▲ enhanced personal information protection with AI-based intelligent tagging automation technology ▲ integration of the latest open source AI technology. Advanced analytics capabilities, including built-in frameworks like Partial Dependency (PD), LIME (Likely abbreviated as LIME), and Individual Conditional Expectation (ICE), help enterprises ensure transparency and trust in their AI solutions. Furthermore, intelligent tagging technology, which identifies data characteristics, automatically identifies personal information such as age, address, and asset status, classifying and managing it in accordance with privacy policies to ensure compliance with relevant regulations such as the European General Data Protection Regulation (GDPR). Furthermore, Python, R, and SAS models can be integrated with SAS Viya, and it supports Red Hat, Oracle Linux, SUSE Linux (SLES), and Windows Server.

Kim Geun-tae, Director of SAS Korea, said, “While real business results are being achieved through AI, such as deep learning and natural language processing, the black box problem that arises as models become more complex and the resulting reliability issues are challenges that need to be resolved.” He continued, “Just as a mechanic inspects a car engine and provides an accurate diagnosis, data scientists must clearly understand AI models to derive valuable insights. The latest version of SAS Viya helps companies gain confidence in the operating principles and methods of AI models, ultimately enabling them to provide products and services that customers can trust.”


NH Nonghyup, KB Insurance, and other domestic and international companies have adopted the system to improve customer service.
Recently, NH Nonghyup Bank built a big data analysis platform using SAS Visual Data Mining and Machine Learning (SAS VDMML) based on SAS Viya. Through this, we effectively explored big data on 22 million individual, small office, and corporate customers, and expanded personalized services such as customer-specific product recommendations, customer management based on churn potential, and small office (SOHO) customer discovery.

KB Insurance is the first in the domestic insurance industry to adopt SAS Viya, further strengthening its AI competitiveness. Specifically, the company plans to leverage SAS AI technology to further enhance its existing big data analysis capabilities in areas such as customer management, personalized product development, and insurance fraud detection, thereby providing customers with the services they desire in a timely manner.

Meanwhile, globally, leading companies in various industries, including AAA Northeast, a U.S. auto insurance non-profit; Swiss Mobiliar, a leading Swiss insurer; Cary, North Carolina; Targetbase, a U.S. custom marketing agency; and Tieto-Tapiola Oy, a Finnish financial IT service provider, are adopting SAS Viya and the SAS Analytics Platform to increase data utilization and solve business challenges.