'We have begun testing with Ernie Bot': How marketers are using China's AI tools

Back in March, Chinese digital giant Baidu unveiled what was then dubbed as its answer to ChatGPT in the form of ‘Ernie Bot,’ an artificial intelligence-powered chatbot that was shown in a series of pre-recorded videos to be able to speak in Chinese dialects and generate a video and image with text prompts. However, the launch of Ernie Bot disappointed investors and sent shares tumbling, largely because it was not given a public launch and was only made available to a select number of users.

Similarly, a month later, Chinese tech giant Alibaba released their large language model chatbot called 'Tongyi Qianwen,' which can communicate in both Chinese and English and can turn text into images and short videos. However, the software is currently only available in beta tests for enterprise customers in China.

Among marketing agencies in China, there has been some experimenting with various Chinese AI tools, but nothing substantial or on the level of what the US tech giants offer.

"We have begun testing with the Ernie Bot," says Tomy Chan, managing director of data, technology and innovation at Publicis Groupe China, but in terms of scaled usage, we are currently leveraging non-China products from OpenAI and Stable Diffusion."

Kathy Shen, managing director of Tag, China, says that her teams are in the process of trialing a few of the Chinese AI tools, but at the moment, the main tool they are experimenting with is Kuaizi (筷子科技)—an AI-powered platform for generating and producing original content for advertisers or marketing content agencies.

"We see it [Kuaizi] as a potential tool for short video editing," says Shen. "This would enable us to make more efficient cuts and create video content for a variety of channels and different length videos as appropriate."

Meanwhile, Vincent Wong, managing director of Labs3.io, a consultancy that helps brands unlock growth in Web3, says that while they're keen to try a number of AI tools, they mostly use ChatGPT because it's the best language-based AI tool on the market today.

"A project called Super Clue benchmarks AI tools, measuring performance across three dimensions: general intelligence, domain expertise, and Chinese proficiency," says Wong. "And at least for the time being, ChatGPT4, the latest version of ChatGPT, scores higher than any other AI tool in the world."

Ashley Dudarenok, a China digital expert and founder of ChoZan and Alarice, says she maintains an open mindset towards embracing Chinese AI tools when they become widely available.

"Chinese AI technologies, such as Huawei's Mindspore and Baidu's PaddlePaddle, have demonstrated great potential in advancing the field," says  Dudarenok. "Their availability could offer us access to innovative solutions and capabilities to enhance our operations and drive further efficiency."

Can China catch up?

Certainly, there appears to be little resistance from Chinese consumers when it comes to the use of AI. A recent survey by Ipsos (chart below) measured public sentiment towards artificial intelligence in a number of countries around the world. Asked whether products and services using AI have more benefits than drawbacks, people in China viewed AI products more positively than any other country.

But while the appetite is clearly there, can China develop tools that will be enough to rival the hugely popular US-produced AI products from the likes of OpenAI and Stable Diffusion?

"We believe China could catch up in five years’ time with coordinated government support," says Chan. "However, the market is facing a number of major barriers at the moment, including data quality used for training data as well as data volume and processing power. For example, the online landscape in China currently suffers from an over-saturation of marketing and sales-driven content. As such, it can be difficult to distill quality content from promotional content, which dilutes overall data quality."

Chan also points out that corporations in China tend to be comparatively more siloed than in the US. Without synergy or willingness to share across sectors or industries, it's difficult to achieve economy of scale for data that is critical for effective training.

"Processing power is also impeded as a result of limited regulatory support," adds Chan. "We feel progress could only be accelerated with tangible government support. By driving synergy and coordination across efforts and resources, we are confident that China could make headways over the next five years."

However, that government support could soon be coming, with China's President Xi recently talking up the vision of ‘Digital China.’

"This means increased support and encouragement for digital initiatives and tech R&D are highly likely to happen or have already happened, to push for stronger AI offerings," says Tony Chan, regional strategy director Assembly APAC.

Has the US benefited from a first-mover advantage?

While it's clear that American AI tools like ChatGPT have had a first-mover advantage, it's worth remembering that being first to market hasn't always led to long-term dominance.

"Services like TikTok in short videos, and now Temu in social commerce, show that Chinese companies are able to capture large chunks of the global market without being first to market," says Wong.

One interesting side note is that even if Chinese companies develop a smarter AI tool, based on current AI training methods, it's likely that the tool's English proficiency will still be better than its Chinese proficiency.

"That's because these current AI tools are developed using large language models based on text," says Wong. "At least for now, over 50% of the text-based Internet is in English, with other languages accounting for less than 10% each."

Jonas Tillman, senior developer, Media.Monks China, says that while the waiting lists for developer access to a lot of the AI services in China have been slow, he remains hopeful that when China does roll out its products at scale, they will be more refined.

"Chinese companies not being first in this field hopefully results in more polished first products that are more accessible by less tech-savvy users," says Tillman. "The first-mover advantage with gained know-how is, of course, an asset, but the cost of gaining that asset for OpenAI, for example, has been extremely high. The ability to look at the result of those endeavours and to invest wisely can be equally valuable."

Will Beijing's tight internet controls and regulations hamper the adoption rate of China's Al technology?

Whereas American artificial technologies have pretty much had free rein, China has already mandated that it will require a security review of generative AI services before they’re allowed to operate. But many feel that more stringent controls would not hamper China's AI but instead improve it.

"Some of the tight measures may enhance its safety and ethical use," says Ashley Dudarenok. "Beijing's focus on user privacy, content discrimination, and model development demonstrates their commitment to responsible AI practices and aims to establish clear boundaries for AI development without stifling innovation too much, allowing companies to innovate within defined limits. This proactive approach instills confidence in AI technology and encourages wider adoption."

Others agree that China's tight internet controls and regulations may even cause AI technology to flourish, at least domestically.

"ChatGPT has been banned in China since February, giving space for home-grown companies to grow their AI capabilities and become potential AI powerhouse that provides market-relevant solutions, leading to uniquely different AI offerings and expertise vs other markets," says Chan, regional strategy director, Assembly APAC. "This is history repeating itself when China banned Google, WhatsApp and Twitter, which elevated digital expertise locally and gave rise to well-known platforms like Baidu, WeChat and Twitter."

However, as with everything else, the development and success of AI in China will be determined in Chinese terms.

"China has at least three times more digital users than the US," says Chan. "This means China does not rely on external factors to grow its unique capabilities and scale, as supported by its technological self-reliance push. This also means that what is deemed a success in China may not be viewed similarly outside China."

For now, while the US appears to be ahead in the AI race at the moment, there is optimism and interest about what China will bring to the table.

"China’s tools are largely an interim response to the AI wave at this point," says James Bay, chief commerce officer at Wunderman Thompson China. "However, we can expect announcements coming around September / October from the Chinese internet giants and actual viable product rollouts in 2024. China, being China, is no stranger to not being first to market, but have always found ways to ingeniously improve on things, so we are definitely very optimistic about what is to come."

And Jonas Tillman, senior developer, Media.Monks China says he's excited to explore the new AI products that China brings onto the market.

"China has many mature tech companies, and given the right incentives, I have no doubt that there will be an ecosystem that may have a different flavour than what the Western players offer."