College students write papers and perform AI big model magic show

The student paper was misjudged by AI, highlighting the double-edged sword effect of AI-assisted writing. Although it improves efficiency, the risk of misjudgment and academic integrity issues coexist.

Who is the busiest during the college graduation season? Besides the graduating students, it is AI big models. Wenxin Yiyan, iFlytek Spark, Tongyi Qianwen, Kimi... all of them are extremely busy.

Because AI not only helps graduates write their papers, but also detects their AI content and then uses AI to revise them; AI is busy, very busy, extremely busy.

However, papers written with the participation of AI are easily exposed at a glance. In the references of the papers, the authors are Sun Wukong and Zhu Bajie; in the content of the papers, the actual exchange rate is positively correlated with net exports; and the "AI-accented" language throughout the papers exposes the true nature of some papers before they are even tested.

Although using AI to write graduation thesis may result in "crazy literature", AI's work efficiency and knowledge reserves far exceed those of humans, allowing it to become friendly partners with many students during the thesis writing process.

At the same time, with the introduction of relevant regulations, the risk of using AI to write papers has increased; therefore, more people use AI as a learning partner: translating literature, speeding up information, learning English, reviewing before exams...

In the process of learning, is it better to get half the result with twice the effort or twice the result with half the effort? Maybe it does not depend on IQ, but on the tools.

1. AI papers reveal their true colors

What is frustrating for graduates? Answer: The paper duplication rate is too high. What is even more frustrating? Answer: The paper duplication rate and AI assistance rate are both high.

Nowadays, more and more college students use AI to assist them in writing papers. Wenxin Yiyan, iFlytek Spark, Tongyi Qianwen, Kimi, GPT4, etc. have become their study partners.

However, in the process of paper detection, major universities have also embraced technology, using AI to detect the AI ​​content of papers, using magic to defeat magic; and the reason why an AI-assisted paper will reveal its true colors is mainly due to three reasons:

First, the references are fabricated.

To manually identify the AI ​​content of a paper, sometimes you only need to turn to the end of the article and check the cited references; because the writing information of AI comes from the Internet, and the sources of many references are pure nonsense.

When graduate student Xiao Bai used GPT4 to write a paper, he found that among the authors of the references, there were not only Li Siguang and Zhang San, but even Sun Wukong and Zhu Bajie. It seems that the F4 who went to the West to obtain the scriptures also have to concentrate on scientific research.

Meanwhile, graduate student Taozi also said: "In the paper I wrote using Wenxinyiyan, Zhang San, Li Si, Wang Wu, and Zhao Liu in the references are all studying artificial intelligence."

Not only that, there are also AI-recommended documents, whose titles and authors are of high imitation level, but cannot withstand careful scrutiny.

For example, college student Xiaoxin told Brother Pan: "I want to understand the error model of rope-driven parallel mechanisms, and asked Wenxinyiyan to recommend some literature. I searched the titles and found that I couldn't find them online. I went to the corresponding journals to look for papers with the same volume and issue, but the titles were inconsistent with the actual results."

Secondly, it is a mistake in common sense knowledge.

When reading AI-generated papers, readers can easily encounter "crazy literature" in the sentences. This is because the information AI absorbs is incorrect, resulting in "random reading and replying".

For example, Xiao Mo, a college student, said: "When I was writing my thesis, I asked chatGPT an economic question: What is the relationship between the real exchange rate and net exports? But it told me that they are positively correlated. It was so wrong."

Xiaoxia, a college teacher, also told Brother Pang: "I found 'Brand Research Expert Daniel Chen Weiting' in my student's graduation thesis. I don't know which AI wrote it."

Finally, there is the unique language style of AI.

Similar to the "translation tone" and "classical Chinese tone", as AI produces more and more texts, the specific "AI tone" language is gradually becoming familiar to people.

For example, some netizens used Tongyi Qianwen to read literature and asked it to summarize the content of the article. Although the AI ​​reads quickly and summarizes the content accurately, the extremely long sentences it generates and the stacking of various conjunctions make the text appear cold and ruthless.

At the same time, Xiaoxia also told Brother Pan: "You can tell at a glance that a paper written by AI is full of empty words and clichés. If the style of the paper is inconsistent, it is likely that AI was used."

2. AI becomes a research paper partner

Since AI is very likely to be exposed as a fraud when writing papers, why do students still practice AI writing? The main reason seems to be that AI is indeed more efficient.

For example, Xiao Hao, a college student, said: "My paper was due, but I still had a large chapter to write. I simply copied and pasted it to Wenxinyiyan to rewrite it, and the final plagiarism rate was only 7%."

Brother Pang discovered that many large AI models today can complete paper writing. Just tell it the title, theme, and other relevant requirements of the paper, and AI will write for you.

For example, Brother Pan asked Tencent’s newly launched AI model - Yuanbao: "Can you help me write a paper?" The other party asked Brother Pan to provide: paper topic, length requirement, research background, data source, time requirement, references, etc., and said that he would generate a paper outline and then help you complete the writing.

At the same time, the official accounts of some large AI models are also guiding you on how to use them to complete your paper. For example, the Wenxinyiyan App account said: "Just enter the paper topic, and you can generate the paper outline and framework."

In addition to relying completely on AI, some students do not let AI take over everything when writing papers, but use it to assist them in writing.

Graduate student Daning told Pang Ge: "I first find the parts of the paper with high repetition rate, then paste these sentences into Kimi and tell it the relevant instructions for rewriting, such as replacing synonyms, abbreviations and then expanding them, and finally optimizing the paragraph as a whole."

On the other hand, some college students said: "I am a computer idiot, but I can find answers to all the colloquial questions about iFlytek Spark Excel formulas. I must have a place for it in the acknowledgments of my graduation thesis."

Although various AI models make paper writing more efficient, papers written by students themselves may be recognized as AI by AI, thus causing a doubling of insult to their academic ability and mental state.

For example, college student Xiaohong told Brother Pan: "I wrote a paper that I made up, but it was detected that the AI-assisted rate was 71%. Is it because my content is too empty or there is something wrong with the expression?"

Not only that, there were also college students who were detected by AI to have too high AI content in their papers. Then they used AI again to reduce the AI ​​content, playing a game of "academic nesting dolls" and staging a magic showdown.

3. It can also be a learning partner

Under the protection of various AI masters, even if you are a poor student, you don’t have to worry about your thesis. As long as you find the right way to use it, AI will take care of everything for you.

However, in August 2023, the draft of China's Academic Degree Law was submitted to the Standing Committee of the 14th National People's Congress for deliberation. It stipulates corresponding legal responsibilities for using artificial intelligence to write degree theses, and clarifies that using AI to write degree theses is academic misconduct and the degree certificate may be revoked.

It seems that "academic tailors" who use AI to assist in paper writing and stitch together the achievements of others face higher risks while gaining high returns.

Therefore, in addition to paper writing, more and more college students use AI as a learning partner. They do not rely on it to write papers, but only use it to help improve their cognition.

For example, some netizens use AI big models to translate foreign literature. They feed foreign literature that is difficult to read to the AI ​​big model, and it can translate the Chinese version for you in a short time.

At the same time, some college students use AI to speed-read the content of literature and generate a refined outline. Friends who study hard have doubled their efficiency and become "speed-reading experts" in their circle of friends.

Not only that, AI is upgrading its capabilities in many aspects related to learning.

For example, the Wenxin Yiyan App has built-in columns such as "Exam Review Assistant", "Golden Sentence Generator", and "Mind Map Assistant". When you enter a question in a specific column, AI will help you solve it.

Brother Pan entered "Basic Knowledge of Microeconomics" in the "Exam Review Assistant" column and found that the AI ​​not only summarized the knowledge points involved, but also gave example questions and answers; it is no longer unimaginable for a poor student to quickly upgrade to a top student.

On the other hand, Tongyi Qianwen also has built-in columns such as "Science Q&A", "English Level 4 and 6", and "Primary School to Junior High School Tutors" in the App to ensure that users of all ages can obtain the knowledge they need.

It can be seen that in addition to writing papers, AI big models also have many applications in learning scenarios. If these AI tools are used properly, turning things around may not be a dream.

Author: Pan Ge

Source: IQ Tax Research Center (ID: 1089356)