China’s business leaders are barrelling ahead of their global peers in adopting generative AI, according to a new survey from North Carolina-based software company SAS.
Over 80% of Chinese business leaders surveyed are currently using GenAI in their operations, above the global average of 54% and the U.S. at 65%, according to SAS’s survey, conducted with Coleman Parkes Research.
“China is noticeably ahead, not only in the practical aspects of orchestrating AI into their existing systems and processes, but also in embedding trust by preparing to adhere to GenAI regulations,” the report says.
China is turning to generative AI to help manage its “massive population” and the data collected from them, Bryan Harris, chief technology officer at SAS, suggests. “Pioneering work in generative AI can give China a competitive advantage in the world, with regards to unlocking value and data,” he says.
Still, safety and security issues like hallucination, deepfakes, and data privacy are pushing governments to quickly develop rules around the new technology.
SAS’s survey reports that around 70% of Asia-Pacific business leaders report feel “fully prepared” or “moderately prepared” to comply with incoming generative AI regulations, compared to 59% in North America and 52% in Northern Europe.
Last year, Beijing released some of the world’s first rules governing generative AI. The rules required chatbots to adhere to “core socialist values” and protect national security. Yet penalties were not as strict as observers feared, which analysts read as Beijing signaling support for the fledgling industry.
In October, the Biden administration passed an executive order that, among other measures, called upon AI developers to share safety test data with the U.S. government. Then, earlier this year, the European Union passed its own comprehensive set of AI regulations, dubbed the AI Act.
Still, AI companies may find it difficult to navigate a patchwork of different AI rules. “It’s good to have regulations. It’s just not good to have three different ones,” Harris says. “It’s inefficient.”
China’s AI investment
Since the release of OpenAI’s ChatGPT in 2022, Chinese tech firms large and small are rushing to create their own large language models and generative AI programs.
Big tech firms like Baidu, Alibaba, and JD.com are investing in their own large language models; Baidu claims its ERNIE LLM outperforms the latest version of GPT-4 in some Chinese-language tasks. Several Chinese startups—like 01.AI, founded by former Google China president Kai-Fu Lee—are also racing to release chatbots and other generative AI products.
Yet China’s AI companies are also struggling with a shortage of AI chips, made worse by U.S. controls barring the export of advanced processors (like those made from Nvidia) to China. Startups are reportedly rationing the use of their products due to a lack of computing power.
On Monday, iFlyTek–one of China’s leading AI developers–said it was likely to post a net loss of up to $64.53 million for the first half of the year, which it partly blamed on the “ultimate exertion of pressures from the U.S.” in a Shenzhen stock filing. Shares are down 5.4% since Friday’s close.
“All countries are looking for information advantage as a result of AI,” Harris says, from the level of national security all the way down to commercial applications. “Generative AI becomes the new nuclear arms race in that regard,” he says.
Yet Chinese companies have some advantages, such as being “more aggressive” than their U.S. peers when it comes to combining datasets and applying them to AI models, Harris says.
He’s optimistic that Chinese companies like Alibaba can still catch up to their U.S. peers like OpenAI, even with a shortage of cutting-edge GPUs. “What’s cool about this whole movement is that there’s innovation happening almost simultaneously around the world. That’s a good thing, right?”
Fortune is bringing Brainstorm AI to Asia for the first time, convening executives, founders, venture investors, and regulators from around the world in Singapore on July 30–31. Register here!
