Airbnb CEO:自研的AI代理非常依赖阿里千问模型,比OpenAI速度更快、更便宜

Core Insights - The choice of AI models by companies is shifting from technological superiority to cost-effectiveness, with Airbnb's CEO Brian Chesky highlighting the use of Alibaba's Qianwen model for its speed and lower costs [1] - Airbnb's AI customer service agent, launched in May for English-speaking users in the U.S., has significantly reduced the need for human customer service by 15% and decreased average problem resolution time from nearly three hours to six seconds [1] Group 1 - Airbnb relies heavily on Alibaba's Qianwen model for its AI customer service, despite having access to OpenAI's latest models, due to better speed and cost [1] - The AI customer service tool has been upgraded to include operational buttons and links, allowing users to make booking changes or cancellations without human intervention [1] Group 2 - Currently, there are no plans to integrate Airbnb's application with ChatGPT, as the connection tools are not fully ready, contrasting with competitors like Booking Holdings and Expedia Group [2] - Chesky mentioned that OpenAI needs to develop a robust platform for Airbnb's application to operate almost independently within ChatGPT [2] Group 3 - Airbnb has introduced new social features to encourage user connections and improve travel recommendations within the app [3] - Users can now share their Airbnb profiles after booking experiences and message each other directly, with privacy measures in place [3] - More social features are expected to be launched next year, potentially enhancing user-generated content and travel inspiration within the Airbnb platform [3]