Challenges and risks about ChatGPT under discussion among professors and experts
Professors and experts have found risks and challenges with the usage of the buzzy ChatGPT and other AI generative or AIGC tools, covering academic plagiarism, biased information, artwork copyright and "replacement" threat for certain professions.
When asked the question "what you will do when your students are found to use ChatGPT in paperwork?" Professors from local universities gave their answers.
It's certainly a "No" and regarded it as plagiarism, said Xie Xuefang, professor and the vice dean of the School of Humanities at Tongji University.
Zang Zhipeng, a professor at East China University of Politics and Science and Law, welcomes the new AIGC tools for teaching and learning. But for research and official papers, he noted that ChatGPT and AI-generated content should be clearly marked – reflecting various university standards.
All Shanghai universities have adopted tools to fight plagiarism and they are developing or upgrading them for the new ChatGPT wave, experts said.
Fudan University has developed its own ChatGPT-like service MOSS. MOSS is an AI-powered language model with one-tenth capabilities of OpenAI's ChatGPT. It's sufficient for academic research, according to media reports.
Experts and professors shared their opinions in a forum held in Shanghai recently, organized by the Metaverse on Real Joint Research Institute in Shanghai.
Professor Zang is also a key member of the Cultural Metaverse Professional Committee of the China Cultural Industry Association. The committee also mentioned AIGC risks covering biased information, regulation and several ethical issues.
Professor Xie mentioned AI's tech "black holes" like biased information, fake news with Deepfake techs and copyright issues.
Deepfake, coming from deep learning and fake (faces), has brought security concerns. AI-featured Deepfake technology can "simulate" faces used in social media and applications like the face-swapping app ZAO, which was once popular in China.
Copyright of the AI-generated content doesn't make sense. Each copyright verifying process requires registered cost, which will be unrealistic in the huge volume of AIGC work, said Xiaoice, a Chinese AIGC firm spun off Microsoft.
Images in a graphic novel that were created using the artificial intelligence system Midjourney should not have been granted copyright protection, the US Copyright Office said, cited by media reports.
The AIGC and ChatGPT tools are regarded to "replace" several certain professionals, such as journalists, painters and other content creators, according to industry and media reports.
In the AIGC wave, people will find a new collaborative way to work with AI, while Xiaoice aims to contribute to its capabilities, Li Di, chief executive of Xiaoice, said in the AIGC forum of the Global AI Developer Conference, or GAIDC.
GAIDC, a city-level AI conference held over the weekend in Lingang of the Pudong New Area, is a part of the World Artificial Intelligence Conference, or WAIC. WAIC will be held in Shanghai from July 6 to 8 this year.
Last but not least, Shanghai Daily also asked a question to ChatGPT about its potential risks. It brought new ideas for the story.
"However, as time went on, some challenges and negative impacts of ChatGPT became apparent," said ChatGPT.
The risks include "over-reliance on ChatGPT for decision-making," "bias and misinformation," "a negative impact on human communication and social interaction" and "using ChatGPT threatened to displace many jobs and industries," said ChatGPT, with each point using several sentences to support.