(Yicai Global) July 3 -- The financial industry needs customized large language models, as general LLMs are unsuitable, according to the chief technology officer of Baidu’s fintech unit Duxiaoman Financial.

The financial industry stores data locally because of security and privacy protection and has high risk control, accuracy, and real-time data requirements, Xu Dongliang said at an industry forum yesterday. Against this backdrop, general LLMs cannot meet the industry’s minimum requirements on financial knowledge, controllability, and accuracy, so customized LLMs are needed, he said.

Small- and medium-sized financial institutions lag far behind large ones in terms of business scale and capabilities, Xu added. But as all of them can use advanced financial LLMs to solve their business problems, smaller players will be able to narrow the digital and intelligence gap with larger peers, he pointed out.

Hundreds of financial institutions have already begun using Xuan Yuan, a financial chatbot that Duxiaoman Financial launched in May, Xu said.

Editor: Futura Costaglione