Beijing [China]March 4 (ANI): In China, censorship of social, political and religious topics almost inevitably affects AI-generated material, and there is evidence it has already, Federico Giuliani writes in Insideover.
Recently, China directed major tech companies not to provide access to ChatGPT services on their platforms, either directly or through third parties, people familiar with the matter told Nikkei Asia.
Ant Group, the fintech subsidiary of Tencent Holdings and Alibaba Group Holding, has been instructed not to provide access to ChatGPT services on its platforms.
Tech companies will also need to report to regulators before launching their own ChatGPT-like services, the sources added.
ChatGPT, developed by Microsoft-backed startup OpenAI, is not yet officially available in China. Some internet users were still able to access it using a virtual private network (VPN), Nikkei Asia reported.
Domestically, political content monitoring, censorship, and public oversight are often conducted using AI-driven methods, Insideover reported.
If a machine-learning tool draws information primarily from within China’s notorious Great Firewall, its output will reflect the omissions and biases of an information environment instilled by the state’s strict regulations and propaganda.
For example, researchers Margaret Roberts and Eddie Yang found that a natural language processing system trained on articles from the Baidu Baike online encyclopedia was as Alternative systems differ in perspective.
Elections and democracy are rated positively by globally trained algorithms, or associated with words like “stability”. On the contrary, people who have studied on Baidu Baike gave good evaluations to keywords such as “surveillance” and “CCP”, and associated concepts such as “democracy” with unfavorable adjectives such as “chaos”.
When Chinese tech giant Baidu released its ERNIE-ViLG text-to-image generator in 2022, users such as dissident artist Badiucao discovered vulnerabilities and manipulation, Giuliani reported.
In fact, Chinese IT and social media companies have their own private blacklists and censorship techniques in practice, despite the government and CCP’s detailed regulations and instructions on censorship.
Chinese users have limited access to ChatGPT, while users around the world are trying it out. The Great Firewall hasn’t blocked it yet, though the login requires a phone number from one of the specific countries outside of China.
For example, Baidu’s ERNIE-Bot is rumored to be unveiled next month. Censorship and other forms of manipulation may also be visible in the chatbot’s output, given the company’s heavily regulated search engine and the findings surrounding its AI text-to-image generator, as Insideover reports.
Given that ERNIE-Bot is clearly educated on global data, users should be wary of any unintentional mistakes against the will of the CCP, Giuliani warned. (Arnie)
(This is an unedited and auto-generated story from a Syndicated News feed, the content body may not have been modified or edited by LatestLY staff)
share now