LG CNS announced on the 24th that it has developed a finance-tailored AI evaluation tool for corporations seeking to adopt generative artificial intelligence (AI).
This evaluation tool utilizes pre-built datasets to assess dozens of publicly available open large language models (LLMs) and identifies the most suitable AI models for financial services, including banking, insurance, and securities. Notable examples of open LLMs include LG AI Research's EXAONE 3.5, Meta's Llama, and Alibaba's Qwen 2.5.
Financial corporations fine-tune open LLMs to build their own models due to security reasons such as preventing data breaches. Fine-tuning is the process of optimizing an AI model for a specific purpose by training it with separate data. In contrast, closed LLMs like OpenAI's ChatGPT or Google's Gemini do not have public source code, making it impossible to build and utilize a proprietary AI model; they are available only through subscription.
LG CNS's finance-specialized evaluation tool consists of 29 evaluation indicators and about 1,200 datasets. It comprehensively evaluates inference abilities based on financial knowledge, mathematical calculation skills, understanding of complex questions, document summarization capabilities, comprehension of financial terminology, and the ability to utilize AI agents (Agentic Tool), and is specifically designed to apply financial-related inference problems to the evaluation metrics for accurate performance verification.
Additionally, LG CNS collaborated with financial experts to ensure that the evaluation data accurately reflects actual financial services and expertise, enhancing its completeness. This enables verification of whether AI can provide accurate responses within the regulatory requirements and complex service structures of the financial industry.
LG CNS plans to provide differentiated value across all phases of generative AI adoption for financial corporations, from selecting the optimal LLM to customization and operational stabilization, through this evaluation tool.
LG CNS is continuing to drive business innovation based on generative AI in collaboration with financial corporations. Recently, it began a generative AI platform development project with NH NongHyup Bank and is developing an AI-based customer service response system with Shinhan Card. Last year, it developed an LLM-based future-oriented branch called 'AI Branch' in collaboration with Shinhan Bank and is operating a future-oriented contact center (FCC, Future Contact Center) utilizing AI and cloud technology for KB Financial Group.
Hyun Shin-kyun, president of LG CNS, noted, "The biggest concern for corporations aiming to introduce generative AI into financial services is finding the right AI model," adding, "LG CNS's finance-specialized AI evaluation tool will be the optimal solution to quickly resolve such concerns for financial corporations."