Recently, ChatGPT has become the biggest internet sensation globally, capable of editing text, generating images, writing code, and even assisting with homework and writing papers… Overnight, ChatGPT became the focus of discussion among internet users worldwide, with companies and investors rapidly making plans. Even the CEO of OpenAI, the developer, did not expect such success. Launched on November 30, 2022, it gained 100 million monthly active users in just two months, becoming the fastest-growing consumer application in history.
After AlphaGo’s “crushing” of professional chess players, ChatGPT emerged, reigniting public interest in artificial intelligence. Currently, there are numerous rumors about ChatGPT, and today, Upstream News will take a look at the five major myths surrounding it.
Truth: Human Creativity is Irreplaceable
ChatGPT is undoubtedly the hottest technology right now, and the topic of artificial intelligence replacing humans has resurfaced. A large number of professions that ChatGPT is said to replace have been listed. Experts believe that in the future, media workers, legal professionals, accountants, customer service representatives, teachers, writers, legal assistants, market research analysts, programmers, and more will be replaced by ChatGPT to varying degrees. As a result, some internet users are understandably anxious: Will ChatGPT really take away our jobs?
Experts’ responses to this are negative.
Some analyses suggest that since ChatGPT’s knowledge is based on past data (its database only updates to 2021), it lacks innovative capability and the answers it provides are quite generic, lacking unique insights. “Its knowledge comes from its materials: books, the internet, and some information it has interacted with,” said Huang Li, a lecturer at the School of Computer Science and Artificial Intelligence at Southwest University of Finance and Economics, who believes that this operational mechanism determines that human creativity remains irreplaceable by AI.
“I believe it shouldn’t be simply viewed as replacing or not replacing,” said Professor Zhou Qiang from the University of Electronic Science and Technology of China and a doctoral supervisor. “It can replace part of our work, allowing us to free up time to do newer tasks that it cannot perform, which is actually a mutual promotion between humans and technology.”
▲ Shi Yu believes that technological development can replace simple repetitive labor but will not seriously threaten jobs. Image source: Visual Chongqing
Shi Yu, director of the Intelligent Safety Technology Research Center at the Chongqing Academy of Sciences, shares a similar view. He believes, “Technological development can improve human productivity and replace simple repetitive labor, but it will not seriously threaten jobs.” He stated that technological advancement will promote the emergence of new professions, providing people with new “jobs.” Humans and artificial intelligence will coexist and collaborate rather than replace each other.
Truth: Authenticity and Accuracy Require Human Review
It is said that ChatGPT can not only chat but also take exams, do homework, write papers, compose poetry, write scripts, and even code. But are these outputs reliable? After deeper interactions, some have pointed out that what ChatGPT says is not reliable and is often just “seriously talking nonsense,” with many limitations.
From a usage perspective, Gong Liang, assistant general manager of the Information Technology Department at China Minsheng Bank Chengdu Branch, believes that caution is necessary when using ChatGPT. “Its answers are not necessarily correct; more often, they just appear correct. And these seemingly correct answers cannot be directly used in many fields, such as healthcare, investment, and market trends,” he stated, noting that certain areas may involve regulation or industry self-discipline. If ChatGPT and similar technologies are not regulated, they may cause certain troubles.
Researchers share similar concerns. According to a report by The New York Times on February 8, the U.S. news credibility assessment and research organization (NewsGuard) tested ChatGPT and found that the software could adapt information in seconds, generating a large amount of content that appears “convincing” but lacks sources. They stated that currently, there is no feasible method to effectively address this issue.
▲ Zhou Xiangdong believes that the authenticity and accuracy of content generated by ChatGPT may still require human review. Image source: Visual Chongqing
“As a language model, ChatGPT has become adept at understanding and processing human natural language. However, its development focus is on fluency in language, on how to better understand human language and communicate, making ‘chatting’ more enjoyable and interesting, rather than ensuring the authenticity and accuracy of generated content,” said Zhou Xiangdong, a researcher at the Chinese Academy of Sciences and CTO of Zhongke Hongtuo. He believes that the issues of authenticity and accuracy in ChatGPT’s generated content may still require human review.
Truth: Currently Free for the Public
Currently, ChatGPT offers a free service to the public, but its developer, OpenAI, plans to profit from this service in the future. On February 2, OpenAI’s official account announced that its membership service is coming soon, with a monthly fee of $20 (approximately 134 RMB). The announced privileges of the ChatGPT Plus service include “normal access during peak times,” “faster response times,” and “priority access to new features.”
However, before OpenAI’s paid operations, many products claiming to be associated with ChatGPT have already started charging. Their service methods mostly involve free trials, after which fees are charged once the free usage limit is reached.
▲ “ChatGPT Online” offers paid options. Image source: Internet
For instance, “ChatGPT Online” provides users with four free conversation quotas; after that, continued usage requires a recharge, with recharge amounts of 9.99 RMB for 20 times (valid for three months), 99.99 RMB for 1300 times (valid for six months), 199.99 RMB for 3000 times (valid for one year), and 999.99 RMB for unlimited use (valid for one year). Another similar service, “GPT Deep Blue,” also shows a membership recharge mechanism with options for 199 RMB monthly, 399 RMB quarterly, and 999 RMB annually, with a page even featuring an option to “join as an agent to earn money.”
The WeChat public account “GPT Dada” states: “The backend connects to the ChatGPT open window, limited to a trial (5 experiences per day).” This public account’s technical interface is provided by Wuhan Nightingale Technology Co., Ltd., which was renamed “Chat Dada” on December 10, 2022, and later changed to “GPT Dada.” After entering the chat window and following the prompts to send “GPT,” the public account automatically replies: “Congratulations on receiving 5 experience times… Daily experience times are limited!”
The WeChat public account “Huachen Pro” is certified by Chuzhou Huachen Advertising Media Co., Ltd. After entering the “ChatGPT” function provided by the public account, it offers options for health assistants, translation assistants, and interviewers, with the chat box avatar using the ChatGPT logo, albeit with a different color. In the personal homepage, users can choose to “open a membership for unlimited dialogue,” including monthly, quarterly, and annual membership types, priced at 199 RMB, 399 RMB, and 9999 RMB, respectively.
Truth: Counterfeit Versions Involved in Trademark Infringement
On the WeChat platform, simply searching for keywords like “ChatGPT” or “AI” in the search box reveals a series of products that claim to be related to ChatGPT and use its official logo as their avatar. Many of these accounts were registered in the year when ChatGPT “went viral.”
On WeChat, “ChatGPT Online” introduces itself as “the Chinese version of ChatGPT, a high-level intelligent AI chat and writing tool.” The operator is Shanghai Entropy Cloud Network Technology Co., Ltd. According to the Tianyancha app, this company was established in January 2018 with a registered capital of 2 million RMB.
▲ Many names that appear to be related to ChatGPT can be found on WeChat. Image source: Internet
Recent information shows that “ChatGPT Online” has been renamed “AI Dialogue | Intelligent Writing Q&A,” but it still provides related services on its public account, and the avatar has removed the official ChatGPT logo, although the customer service name added by this public account remains “ChatGPT Online Customer Service.”
Another public account named “OpenAI Artificial Intelligence ChatGPT” shows that its parent company is Guizhou Star Power Information Consulting Service Co., Ltd. According to the Tianyancha app, this company was established in July 2014 with a registered capital of 1 million RMB. The name “OpenAI Artificial Intelligence ChatGPT” has now been changed to “Unnamed Public Account.”
However, before the press release, the reporter searched again on WeChat, and public accounts containing “ChatGPT” in their names have almost disappeared, with some accounts having already changed their names.
It is essential to know that currently, the developers of ChatGPT, the artificial intelligence research organization OpenAI, do not provide any services related to ChatGPT in mainland China, nor do they have corresponding companies operating.
Regarding the use of the names and trademarks of OpenAI and ChatGPT, Senior Lawyer Chu Jiang from Beijing Zhou Tai Law Firm stated that such products’ activities online involve a series of legal issues, with potential violations of trademark law, anti-unfair competition law, consumer rights protection law, cybersecurity law, and even criminal law. Chu Jiang emphasized that ChatGPT is a product launched by OpenAI, which holds all property rights, including the “ChatGPT” name as a source identifier for products/services.
▲ “ChatGPT Online” has been renamed “AI Dialogue | Intelligent Writing Q&A.” Image source: Internet
Truth: Either Middlemen or Counterfeits
In addition to using the name and trademark of ChatGPT, some public accounts and mini-programs claim to “transfer” ChatGPT services to domestic users through technological means, charging fees for convenience. But are they truly providing “transferred” ChatGPT services?
Previously, a media reporter called the staff of Shanghai Entropy Cloud Network Technology Co., Ltd. to inquire about the situation. The staff claimed that all dialogue responses sent through their platform were obtained via a direct connection to OpenAI’s API. “We also need to pay OpenAI for each question and answer, which incurs significant costs, depending on the length of the dialogue characters,” said the staff member.
After directly questioning multiple domestic ChatGPT service providers and comparing their responses with the official ChatGPT, it was found that some domestic providers likely offer interfaces directed at ChatGPT, while others may not be linked to ChatGPT services and are using other AI dialogue models. Domestic “ChatGPT service providers” require users to view an advertisement for each question answered.
Network security practitioners have stated that currently, there are only two models of “ChatGPT service providers” in China: one obtains the official ChatGPT API and provides it to domestic users as a “middleman,” and the other is a complete “counterfeit,” using AI dialogue models of inferior quality compared to ChatGPT, and therefore cannot provide high-quality responses.
