The huge technological leap that machine learning models have shown in the last few months is getting everyone excited about the future of artificial intelligence, but also nervous about its uncomfortable consequences. The ability to hold intelligent conversations is the new obsession in sectors across the board.

Entrepreneurs, researchers, and investors in China are looking for ways to make their mark in the generative artificial intelligence space. Open source models are being used by tech firms to create tools for consumers and businesses. Individuals are using artificial intelligence to make money. Text, image, and video synthesis should be used according to the regulators. Concerns are being raised about China's ability to keep up with artificial intelligence.

As generative artificial intelligence takes the world by storm, let's take a look at how it is being used in China.

Chinese flavors

generative artificial intelligence is on everyone's lips thanks to popular art creation platforms. Halfway across the world, Chinese tech giants have captured the public's attention with their equivalent products, adding a twist to suit the country's tastes.

Baidu, which made its name in search engines and has in recent years been stepping up its game in self-driving cars, operates a 10-billion parameter model trained on a data set of Chinese image-text pairs. Is it fair against the Americans? The results from the prompt " kids eating shumai in New York Chinatown" were given to Stable Diffusion, whereas the same prompt in Chinese was given to ERNIE-Vig.

The Diffusion is stable.

ERNIE- VILG.

I think the results are a tie, because I grew up eating dim sum in China and Chinatown. They didn't get the right shumai, which is a type of pork and shrimp dumplings in a half- open yellow wrapping. Stable Diffusion nail's the atmosphere of a Chinatown dim sum restaurant, but its shumai is off. The Cantonese version of shumai is more popular in eastern China than it is in the west.

The quick test shows how difficult it is to capture cultural nuances when the data sets used are biased.

Different Dimension Me is a Chinese tool that can turn a photo of a person into a character. The generator has its own biases. It took off in other countries because it was intended for Chinese people. Users realized that the platform failed to identify black and plus-size individuals, which resulted in offensive results.

Of course also clearly not having the model adjusted properly for darker-skinned folks, sigh

Anyway Different Dimension Me is the name, but sorry they already blocked / limit overseas users as couldn’t handle the traffic pic.twitter.com/cYi6rJwTaC

— Rui Ma 马睿 (@ruima) December 7, 2022

Harry Shum, a computer scientist who co-founded Microsoft's largest research branch outside the U.S., is the creator of Taiyi, a large-scale Chinese text-to- image model. There are 20 million Chinese image-text pairs trained on the open source model.

IDEA is one of a few institutions backed by local governments that work on cutting-edge technologies. The center is likely to enjoy more research freedom because of that. It is supported by one of China's wealthiest cities and is based in the tech hub of Shenzhen.

Rules of AI

China's generative artificial intelligence tools are shaped by local laws. The MIT Technology Review pointed out that the text-to-image model filters out politically sensitiveKeywords. It is expected that the Chinese internet has always been a place where censorship is common.

There is a new set of regulatory measures targeting what the government calls "deep synthesis tech", which is technology that uses deep learning, virtual reality, and other synthesis methods to generate text, images, audio, video, and virtual scenes. The fact that prompt can be traced to one's real identity inevitably has a negative effect on user behavior.

The rules could lead to more responsible use of generative artificial intelligence, which is already being used to make explicit content in other countries. People are not allowed to create and spread fake news in China. The service providers will be in charge of implementing that.

China’s generative AI rules set boundaries and punishments for misuse

Yoav Shoham, founder of AI21 Labs, an Israel-based OpenAI rival, said in an interview that China is at the forefront of regulating generative artificial intelligence. There are many companies that are putting limits on the use of artificial intelligence.

There is no consensus on how to govern the field. Shoham thinks it's an area we're learning together. It has to be more than one person working on it. It has to involve technologists who understand technology and what it does and what it doesn't do, the public sector, social scientists, and people who are impacted by the technology as well as the government, including the sort of commercial and legal aspect of the regulation.

Monetizing AI

Machine learning is being used to make money in a variety of ways in China. They aren't from the crowd that's tech savvy. They're opportunists or stay- home mothers looking for an extra source of income. They realize that if they improve their prompt, they can trick the artificial intelligence into making beautiful images, which they can post on social media and charge for downloads. The skilled ones are selling their skills to other people who want to join the game and make money.

The rest of the world uses artificial intelligence in their jobs. Light fiction writers can cheaply make illustrations for their works, a genre that is shorter than novels. Artificial intelligence can be used to design T-shirts, press-on nails, and prints for other consumer goods. By generating large batches of prototypes, manufacturers can save money.

It is not yet known how different generative artificial intelligence is in China and the west. Entrepreneurs make decisions based on their initial observations. Entrepreneurs tell me that businesses and professionals are happy to pay for artificial intelligence because they see a direct return on investment. When China kept its borders shut, e- commerce sellers were struggling to find foreign models, and one clever application came from the Hillhouse-backed Zmo.ai. There is a solution. The two companies created models of all shapes and colors.

Some entrepreneurs don't believe that the type of skyrocketing valuation and rapid growth their Western counterparts are experiencing will be possible with the use of an artificial intelligence-powered software. Many Chinese startup have told me that they worry that China's enterprise customers are less willing to pay for software as a service than those in developed economies, which is why they start expanding overseas.

Dog-eat-dog competition is what is taking place in China. Product-led software, which doesn't rely on human services, can do well in the U.S. In China, even if you have a great product, your rival could steal your source code overnight and hire dozens of customer support staff, which don't cost that much, to outrace you.

Chinese companies tend to prioritize short-term returns over long-term innovation according to Shi Yi, founder and CEO of flashcloud. He said that Chinese tech firms tend to focus on getting skilled at applications and generating quick money. Major breakthrough in generative artificial intelligence this year are all happening outside of China, according to one investor.

Wrapped in red tape, China’s startups give up their mainland dreams

Roadblocks ahead

Chinese tech firms may not have the best tools to train large neural networks. The U.S. government imposed export controls on high-end artificial intelligence chips from China. An enterprise software investor at a top Chinese VC firm said that computing will take longer and cost more if a startup uses less powerful chips. He said that the sanctions are pushing China to invest in advanced technologies.

The impact of the U.S. chip sanction on the company's artificial intelligence business is limited in the short and long term. A large portion of the company's cloud business doesn't rely too much on the advanced chips. In cases where it needs high-end chips, it has already stocked enough in hand to support its business.

Is the future something you have in mind? The executive said that they have their own developed artificial intelligence chip called Kunlun. By using our Kunlun chips in large language models, the efficiency to perform text and image recognition tasks has been improved by 40% and the cost has been reduced by 20% to 30%.

Time will tell if Kunlun and other indigenous artificial intelligence chips will give China an advantage.