Waterdrop Internally Tests a ChatGPT-Like Application and Will Apply It Into Insurance Marketing and Services
In a media interview, Huang Mingxing, who heads Waterdrop’s AI business, said Waterdrop has internally tested a ChatGPT like application, and will apply it to insurance marketing and services.
As an AI-enabled chatbot application developed by Open AI, ChatGPT can chat with users and even write emails, do copy writing or translation and even edit code by learning and understanding human languages. Since its launch in November 2022, ChatGPT has caused a sensation in the industry, with the number of monthly active users exceeding 100 million in just two months.
Huang Mingxing said, with a consistent focus on frontier AI technologies, Waterdrop attracted the attention of Open AI when it released GPT-2 in 2019. Built on the GPT-3.5 model, ChatGPT is an intelligent chatbot with AIGC at the core. Compared with other intelligent chatbots, ChatGPT generates more intelligent content and thus better interacts with users through chat. That’s why it gains popularity worldwide very quickly.
The intelligent chatbot technology is not invented by Open AI, but ChatGPT offers great inspiration to relevant enterprises across the world. Huang Mingxing added that by using big models and large-scale annotation in general scenarios, ChatGPT produces good results through massive hash rate consumption, but the cost may go up accordingly. “We try to use a smaller model and proper annotation for our insurance marketing and service scenario, which will prove equally effective with proper hash rate consumption.”
Recommended AI News: Complete Verkada Platform is Now Available Across the UK and Europe
“Waterdrop Assistant”, the digital worker launched by Waterdrop in 2022, is designed to help insurance service workers improve working efficiency and precision through intelligent chatbot technology. Now, Waterdrop takes a further step forward by internally testing a ChatGPT-like application, and expects the intelligent chatbot can use texts or voice to independently complete simple insurance marketing jobs, such as introducing short-term insurance products with simpler insurance clauses to customers and answering customer questions in the process.
Huang Mingxing said, “The key to the application of intelligent chatbots to insurance is to train them with massive data.” To build the intelligent chatbots aligned with insurance demand, they must be first trained with plenty of knowledge about insurance products and dialogues with users in insurance sales, underwriting, claim settlement and other scenarios to understand insurance related knowledge and generate contents more accurately. Without data training, even access to advanced algorithms and models like ChatGPT will be useless.
Recommended AI News: Raising the Ceiling of Code-Free Programming for Robotics and Industrial Automation
[To share your insights with us, please write to firstname.lastname@example.org]