Taipei, Taiwan – China is scrambling to develop its own version as artificial intelligence-powered chatbots shock the global tech industry.
Chinese search engine giant Baidu has launched a chatbot called ERNIE in March, following the pioneering launch of ChatGPT, which raised existential questions about the future of fields ranging from education to journalism to medicine. announced plans to release
Chinese tech stocks rose on the news, with officials pledging to boost support for the sector. Similar projects to ERNIE are underway at top institutions including Chinese tech giants Huawei, Alibaba, Tencent, JD.com, and Beijing Academy of Artificial Intelligence.
China’s Ministry of Science and Technology last week said it would push the integration of AI across China’s industries, and cities, including Beijing, also announced plans to help developers.
But while China appears to be producing fast followers that rival ChatGPT, developed by California-based OpenAI, how the technology works within an ecosystem that includes strict internet controls. I have a big question about
“Artificial intelligence, the most general technology we have, should be super general,” said Jeffrey Ding, an assistant professor at George Washington University who studies China’s technology sector. told Al Jazeera.
“But in reality they are shaped by the specific political, cultural and linguistic contexts in which these models are developed and deployed.”
Bots like ChatGPT rely on generative AI to create responses extracted from billions of data points scraped from the internet. This can also make the response difficult to predict.
A long conversation between ChatGPT and its users has derailed, so Microsoft is limiting its ChatGPT-powered search engine, Bing, to a maximum of 5 questions to keep the work going. ChatGPT’s response also slammed US conservatives who accused the bot of “waking up” on hot social issues such as affirmative action and transgender rights.
In China, Internet censors routinely ban keywords, remove posts, and ban users, following the cautious stance of the ruling Communist Party of China (CCP). As such, creative internet users use homonyms, coded messages, and screenshots to circumvent information management.
For chatbots, censorship means a very limited pool of trusted information.
Baidu’s ERNIE chatbot relies on information collected both inside and outside China’s firewalls to get the right data set, including sources such as Wikipedia and the notoriously unwieldy Reddit. I am using the source.
Assuming their products technically perform on par with ChatGPT, Chinese tech companies either limit what chatbots can do, like Microsoft’s Bing, or limit what chatbots can do. may have to choose.
Matt Sheehan, a fellow at the Carnegie Endowment for International Peace who studies AI and China, told Al Jazeera:
“Historically, whenever they face a trade-off between information management and … business opportunity, they always stand on the side of information management and assume that the enterprise will solve it.”
In 2017, Tencent removed two chatbots from its QQ messaging app after allegedly making comments deemed unpatriotic. One of his chatbots, developed by Microsoft, tells users that they dream of moving to the United States, and another of him, developed by Chinese tech firm Turing Robot, tells them they don’t like the Chinese Communist Party. told the user.
Most recently, Hangzhou-based startup YuanYu Intelligence suspended its chatbot earlier this month after providing negative responses about the Chinese economy, but the company’s lead developer said in an interview with The Washington Post claimed that the suspension was simply due to a technical error. .
Baidu itself has long anticipated Beijing’s red line, as seen with the ERNIE VilG Image and art generator released in demo form last year.
The app is widely praised for performing as well or better than its western rivals, but blocks users from content related to politically sensitive topics such as Tiananmen Square, democracy, Xi Jinping and Mao Zedong. .
“With generative AI, the power of this tool is its ability to be creative, connect things you don’t think will be connected, and do things in a variety of styles that you would expect,” Sheehan said. I’m here.
“But how can we prevent perhaps subtle or indirect criticism of the Communist Party’s core beliefs without completely neutralizing the tool itself? seems like a very difficult problem.”
Before ChatGPT was released, China had already taken steps to regulate AI. On Wednesday, the Cybersecurity Administration began enforcing new rules governing search engine recommendations, giving users more control over how their personal data is used by search engines. Did.
In January, China also passed a law regulating deep synthesis, a type of generative AI that can be used to create “deep fakes,” and set up a registry of algorithms last year. seen as unknown.
As part of a broader crackdown on the tech industry from 2020 onwards, authorities will curb companies deemed to be acting beyond their powers, such as unplugging the large IPOs of Ant Group and ride-hailing services. I have no hesitation in doing so. app Didi has conducted an investigation over the allegations of concern about the data.
Despite being blocked by China’s firewalls, ChatGPT has generated a lot of buzz among Chinese users who access the site via virtual private networks (VPNs) and other roundabout methods.
Much of that excitement stems from ChatGPT’s ability to run in Chinese and other languages, even though it was trained in English, said Ding, a professor at George Washington University.
“The excitement isn’t really about business applications. Part of it is the excitement and surprise of how impressive the natural language capabilities of this technology are,” he said.
“And one aspect of it is that ChatGPT has not even been trained on Chinese text. Yes, but it performs very well in other languages as well.”
That said, Chinese can be a particularly difficult language for AI due to its heavy use of idioms and proverbs based on historical context, Ding said.
Chinese developers have already released many chatbots, such as Inspur’s Yuan 1.0 and Fudan University’s MOSS, but none match ChatGPT’s capabilities.
Unlike in Silicon Valley, Chinese tech companies have historically tended to focus on consumer products with short development cycles, said Chim Li, China technology analyst at The Economist Intelligence Unit. He said it puts them at a disadvantage in nascent areas like AI.
The emergence of ChatGPT has provided a “proof of concept” for Chinese companies, showing both the potential of generative AI and the need for long-term investment, Lee said.
“Baidu has been looking at this kind of model for quite some time, but we need to justify this kind of investment just to train the model, and we don’t want to research or talk about the long-term underlying data related to the algorithm. It goes without saying that it will be a lot of fun,” Lee said. Al Jazeera.
“What’s really helpful about ChatGPT is that these companies are now able to say, ‘We want to develop this kind of thing, so I can tell the government what I want to do. ”
Rui Ma, a tech analyst and creator of Tech Buzz China, said that while no one can guess which Chinese firm will come out on top in a race to rival ChatGPT, Baidu appears to have walked out of the gate first. looks like
“I think the most exciting thing right now is still at the model level,” said Ma.
Alibaba told Al Jazeera that it is internally testing a Chat GPT-style bot for use in its apps and cloud services, but would not provide further details or answer censorship questions. I didn’t.
JD.com issued a statement to Al Jazeera last week about its plans to deploy ChatJD, an industrial chatbot for use on retail and financial websites, based on 10 years of data from various platforms. bottom.
Baidu, Tencent and Huawei did not respond to requests for comment.
Aside from heavy government scrutiny, Chinese tech companies also face hurdles from abroad in the form of export restrictions.
In August, US President Joe Biden signed the CHIPS and Sciences Act into law. The law requires government-subsidized tech companies to move production of advanced chips out of China.
Chinese tech companies strategically stockpile chips, but Washington’s efforts to sabotage the sector pose a long-term threat, said EIU’s Lee.
“The United States has explicitly banned the export of these highly advanced AI chips that are used for training models or just for employment, so all these factors put China’s AI developers at a disadvantage in many respects. placed,” he said.
“In fact, many Chinese companies and research institutes have stockpiled some chips that are used for this kind of application, but looking at the size of chips that ChatGPT needs, they will run out at some point. It is very likely that you will make a point,” he added.
