Pinterest and Airbnb are navigating the complex world of artificial intelligence (AI) by leveraging various models, including those developed in the United States and China. As both companies explore the strengths of these models, they are adapting their approaches to enhance their services, particularly in recommendation systems.
Pinterest’s Chief Technology Officer, Matt Madrigal, highlighted the company’s strategy of utilizing multiple AI models. These operate not just U.S.-homegrown models, but on models developed by Chinese developers. The company securely hosts these models within its own proprietary infrastructure, allowing for maximum control and customization to ensure results. As Madrigal pointed out, the strength of these models is in their flexibility.
“Open source techniques that we use to train our own in-house models are 30% more accurate than the leading off-the-shelf models,” – Matt Madrigal
This flexibility gives enterprises such as Pinterest the ability to download and personalize their own AI models to better fit their business objectives. Recently, Pinterest has been testing out Chinese AI models to improve its recommendation engine. This transition sheds light on a continuing pattern within the tech industry to branch out their AI investments.
Airbnb is successful because they use models, including ones made here in the U.S. Second, they provide a safe and secure operating environment for their AI infrastructure. This strategy is part of a larger industry movement, as organizations of all shapes and sizes look for powerful solutions to improve user experiences with tailored recommendations.
And of course, Meta, the parent company of Facebook has made huge waves in the AI world. Earlier this year, in 2023, it released its open-source Llama AI models. These models rapidly became the gold standard among developers building bespoke applications. This dominance was challenged as new competitors arose with the development of models such as DeepSeek, and Alibaba’s offerings.
Mark Zuckerberg, CEO of Meta, has already pledged billions in support of reaching “superintelligence,” highlighting the depth of the company’s investment in AI technology. Sir Nick Clegg, who has just stepped down as global affairs director at Meta, had one job. This shakeup may reflect imminent shifts in the company’s long-term priorities.
In the highly competitive and secretive world of AI development, OpenAI has certainly made a splash. Under the leadership of Sam Altman, OpenAI has gone on an every-offensive, striking partnerships that would provide the computing power and infrastructure needed to support its growing operations. Today, the company is under enormous pressure to ramp revenue and become profitable. In short, OpenAI is looking to advertising as a way to shore up its financial reserves.
The war for the future supremacy in AI has escalated with Chinese challengers proving to be stiff competition. While Alibaba’s Qwen and Moonshot’s Kimi are perhaps less well-known, they’re giants in the tech world. ByteDance, TikTok’s parent company, is just as invested in these technologies.
The reality is changing drastically, at least as reflected by the dramatic rise of one particular model. In September, Qwen shot past Meta’s Llama to become the most downloaded large language model family on the Hugging Face platform. This pivot is one indicator of the changing tastes within the developer community.
A new report from Stanford University, released last month, shows that Chinese state-supported AI models are now on par with their global counterparts. In many instances, they have outperformed them on performance. This judgment is emblematic of an increasing awareness of the professional development options available to us from Chinese creators.
“If you look at the top trending models on Hugging Face – the ones that are most downloaded and liked by the community – typically, Chinese models from Chinese labs occupy many of the top 10 spots,” – Jeff Boudier
Industry experts are cautiously tracking this trend as it poses serious questions over the future development environment of AI technology. Fellow tech company Airbnb decided to take an opposite approach. Cooperation across regions may be crucial in developing thoughtful new models.