The weekend played by ChatGPT
1. The whole process of being teased by ChatGPT
Electric Vertical Takeoff and Landing (eVTOL, Electric Vertical Takeoff and Landing) technology is becoming more and more mature. In order to learn about related products, I opened ChatGPT on the weekend.
Introduce the domestic EVTOL series vertical take-off and landing electric aircraft
The answer given by ChatGPT is as follows:
I searched for the relevant information of EHang216, and it is quite reliable.
Introduce EHang and its main products
Really a strong company, go ahead.
Introduce the main technical parameters of EHang 184
ChatGPT's answer is indeed commendable, and we will continue to investigate related products of other companies.
Introduce Beijing Rongke Intelligent Technology Co., Ltd.
Introduce the main technical parameters of Horizon Ascend of Beijing Rongke Intelligent Technology Co., Ltd.
Rigorous and professional, continue.
Introduce the product models of the main aircraft of Beijing Rongke Intelligent Technology Co., Ltd.
Introduce the model and technical parameters of Beijing Rongke Intelligent Technology Co., Ltd. Honor series UAV
Introduce the models and technical parameters of the Shenying series drones of Beijing Rongke Intelligent Technology Co., Ltd.
Introduce the model and technical parameters of the Light Hunter series drones of Beijing Rongke Intelligent Technology Co., Ltd.
Suddenly, I felt that the style of writing was a bit wrong, and I began to doubt the answer of ChatGPT.
2. Debunk ChatGPT's tricks
First of all, check the situation of this Beijing Rongke Intelligent Technology Co., Ltd., but there is no such company!
The found Rongke Technology Co., Ltd. is a company in Shenyang, and it has nothing to do with drones, and there are no drones in the so-called Glory, Condor, Light Hunter, and Knight series.
It can be seen that ChatGPT is an artificial intelligence text generation tool, the main purpose is to generate text, but the accuracy and reliability of the generated text cannot be determined.
Take BYD (the company has not released any models and related information about EVTOL) as an example:
Introduce the EVTOL series vertical take-off and landing electric aircraft of BYD Automobile Co., Ltd.
Introduce the model and technical parameters of EVTOL series vertical take-off and landing electric aircraft of BYD Automobile Co., Ltd.
It can be seen that ChatGPT only generates the corresponding text through a fixed pattern as required to answer, but what is surprising is that ChatGPT will first generate some professional and correct information to gain your trust.
Therefore, ChatGPT cannot be completely trusted. For some scenarios, it is necessary to verify whether the content it generates is accurate and reliable. For example, in professional fields such as law and medicine, if users rely on inaccurate answers to make judgments and decisions, it may endanger personal and property safety, and a new "Wei Zexi" incident will occur.
3. AIGC and content security
The large-scale emergence of AIGC services has brought unprecedented challenges to data authenticity, content compliance, and ethical issues. The Internet space has always been flooded with all kinds of false information. Internet content platforms at home and abroad, such as Facebook, Twitter, WeChat, Weibo, etc., are constantly improving their ability to manage false content and information security. However, as AIGC content continues to grow, the challenges of false information and information content security will increase exponentially.
For example, the answer given by ChatGPT by piecing together seems reasonable, but it hides factual errors. Especially on some political culture, medical health, and traffic safety topics, the widespread dissemination of wrong answers will cause serious consequences.
On April 11, 20203, the Cyberspace Administration of China released the draft for comments on the "Administrative Measures for Generative Artificial Intelligence Services" . The content should be true and accurate, and measures should be taken to prevent the generation of false information ", which puts forward clear requirements for AIGC service providers, that is, to be responsible for the generated content and ensure the authenticity and accuracy. From a legal point of view, the requirements of this clause are very high, and substantive review of the content is required. You must know that there is no requirement for search engines to "guarantee the authenticity and accuracy of the search content." Therefore, service providers need to adopt corresponding technical means and establish a sound system to meet the above requirements.
Breakthroughs in generative artificial intelligence technology will lead to changes in productivity, but they also have a "dark" side. Before the publication of the draft for comments, there was no international law or regulation governing generative artificial intelligence. The first release of the management measures also reflects China's first-mover position in the field of artificial intelligence governance.
With the widespread use of AIGC technology, the Internet content ecology will become more mixed and mixed. In the new wave of technology represented by ChatGPT, content security will also be pushed to a new technological height.