That, along with the cost of computing power, makes it difficult for startups like Xie to build the kind of massive, sweeping models their US equivalents are trying to create, so most focus on the application level, rather than their own models .
Founded last year, Vrch.io is developing an AI-powered image generator for voice input. In the past, interior designers may have had to use renderings created in Photoshop to show to clients. Now, if people want to redesign a space, they can do it on the spot using generative AI. “For those of us who work in the design world,” says Xie, “we spent most of our time converting information that was difficult to express accurately in words into images, and then using those images to communicate with customers.”
Although vrch.io has an investment from Miracle Plus (formerly Y Combinator China), a startup incubator in China, it is not currently targeting the Chinese market. This is due to the lack of clarity in the regulations.
“As a small company,” Xie says, “we cannot guarantee that every segment of the business, whether it be the algorithms, the data sources or the training of the models themselves, is compliant with regulations.”
In July this year, China’s Cyberspace Administration published interim guidelines on generative AI that focused on privacy, protection of personal information, transparency of algorithms and intellectual property rights. They haven’t set compliance standards for the technology that differ substantially from existing technology regulations, but startups like Xie’s are waiting for more details.
“Regulators clearly do not want to immediately overregulate to discourage innovation and further widen the gap in AI development between China and the US,” said Xu. According to Xu, the rules show that regulators are “willing to integrate the needs and input of technology companies, allowing relatively unfettered development in private environments and company-specific areas, as long as certain red lines are not crossed in the public sphere.” ”
Vrch.io is more concerned about getting its product to overseas markets first. It will wait until major models – most likely developed by Chinese Big Tech companies – become available before rolling out in the home market.
The economic climate also casts a shadow over the technology sector. Slower growth, declining consumer spending, real estate market problems and concerns about local government debt have contributed to an overwhelming sense of uncertainty. The Chinese government has stopped reporting statistics on youth unemployment in urban areas, an indicator of a general economic slowdown.
“When I start a business in this economic environment, I have to pick very specific low-hanging fruit problems,” said Pei Hao, founder of AI startup Lingua Technologies.
His company aims to compete with translation agencies in Beijing and professional editors in Britain and the US, who charge Chinese scientists fees to make their work readable to an international audience.
Hao says partnerships between Chinese academics and non-Chinese counterparts are often hampered by the added workload for native English speakers. “There’s so much cognitive load involved in fixing these papers, some of which are five to 10,000 words long,” Hao says.