top of page

Why Airbnb's 'cars and highways' approach to AI makes perfect sense

“The role of the designer is that of a very good, thoughtful host, anticipating the needs of his guests.”

- Charles Eames


Co-Founder and CEO of Airbnb, Brian Chesky, was asked in a recent interview with Jason Calacanis about the role of AI in Airbnb's future roadmap. The answer starts with Chesky's view of AI’s place in the world as a tool. Chesky likens AI to the infrastructure upon which creative things are built. The foundations of AI are the base models - large language models (LLMs) like GPT4. Chesky describes these base models as being like 'highways'; the infrastructure on which cars travel. Like many others in the field, Chesky recognises that in previous generations, this infrastructure would likely have been constructed by governments and not private companies. At Airbnb, Chesky says they think of themselves as designing 'the cars on the highway'. Rather than being concerned with the infrastructure, Airbnb are focused on tuning these models. The tuning is only as good as your sensibility and customer data, which Chesky sees as Airbnb's ultimate advantage. Whilst over the past few months, OTAs and travel brands have been rushing to implement GPT plugin-driven chatbots in a race to stay at the forefront of the Hype, Airbnb are taking a longer-term view which aligns with their company-wide roadmap and big-picture approach. Within this approach, it all comes down to personalisation and anticipating the needs of the guests. This is how to become a really good host and Chesky says that Airbnb wants to be one of the best companies in the world for AI personalisation. This means developing really well-trained models and leveraging the rich guest data they possess. To get there, Chesky acknowledges that the business needs to undergo significant change and move away from the guest inputting what they want, to the platform already knowing what they want. “It's not about where you're going but why you are going”, advice given to Chesky by Jony Ive. This approach relies on intimate knowledge of each guest, with Chesky saying that in the future "we won't have a search problem, we'll have a matching problem." Like cars on top of highways, the design and interface are more important for Airbnb than for the LLMs they build on top of. The best interface is unlikely to be the long-format text outputs that many travel brands are settling for now. For Airbnb, the interface needs to be more visual and much richer, in keeping with the rest of their product. For this reason, they have held back on realising early features that utilise LLMs. Chesky's goal is for Airbnb to be more of a 'travel community' than a product or service. The goal is to become the ultimate AI concierge and the ultimate host. In another recent interview with Skift, Brian Chesky said that Airbnb is not rushing the implementation of AI because he wants to make sure that the company gets it right. He wants to make sure that Airbnb uses AI in a way that enhances the user experience and does not replace the human touch that is essential to the Airbnb brand. We would summarise Airbnb’s approach as:

  • focusing on the tuning of models to personalise the platform utilising guest data

  • building visual and rich interfaces to support the delivery of AI personalisation

  • enhancing the user experience without replacing the human touch

 


You can listen to Brian Chesky's interview on TWIS here - https://podcasts.apple.com/ca/podcast/airbnb-ceo-brian-chesky-on-early-rejection-customer/id315114957?i=1000611761112


You can read Chesky's interview with Skift here - https://skift.com/2023/05/03/interview-airbnb-ceo-on-how-its-service-will-radically-change-with-ai-by-next-year/

Explore Solutions

Global, Dynamic Destination Content

bottom of page