In an interview this week, Brian Chesky, CEO of Airbnb, acknowledged that while the company intends to eventually integrate ChatGPT-style AI capabilities into its platform, the current technology isn’t yet ready for the demands of travel and lodging. Speaking on CNBC’s Squawk Box, he said the software development kit (SDK) “wasn’t quite robust enough for the things we want to do.”
Chesky noted that Airbnb likely will move toward integrating ChatGPT at some point, but for now, the company is proceeding with caution. The travel platform launched new social features on Tuesday, including direct messaging between hosts and guests, and an upgraded in-app chatbot in North America that can cancel or change reservations.
Apart from that in a separate interview with Bloomberg this week, Chesky revealed that Twitter’s rumored integration is not the route Airbnb is taking, currently the platform uses a model built from 13 different chatbots and is “depending heavily” on Qwen, a model by Alibaba Group. He said the OpenAI-based ChatGPT model simply doesn’t yet meet Airbnb’s enterprise-grade requirements.
Chesky argued that we are only “at the beginning” of the AI revolution. “We’re all going to have to work together,” he said, adding that if companies try to vertically integrate every piece of the stack themselves, it will be “very, very difficult.” His message: AI will lift up many companies, but success will depend on collaboration and readiness.
In my opinion, it is not usual to see a CEO admit that the hype around AI, particularly ChatGPT, still outpaces actual practical readiness. Airbnb is dealing with real risk: travel bookings involve payments, cancellations, regulatory complexity, and high stakes if AI makes a mistake. Choosing to hold off on full ChatGPT integration until the tech meets rigorous standards shows prudence.
At the same time, leaning on 13 different models and Alibaba’s Qwen suggests Airbnb doesn’t want to wait as well and is chasing incremental gains rather than betting everything on one flashy model. So, at the end, I think this will serve them better: delivering reliable service now, and integrating the newer tech when it’s robust.
