Since its launch in November 2022, the artificial intelligence chat robot ChatGPT has caused huge waves in the education field. Many students have made ChatGPT a god, and even called ChatGPT "thesis father". Although I don't agree with this name, I have to admit that ChatGPT is powerful.
The robot's ability to simulate human speech and writing so realistically has many educators question how essay-based learning can continue when students have such a free shortcut at their fingertips. In fact, college students have already been caught cheating by using this tool.
In order to gain a deeper understanding of this issue, I conducted a large number of online surveys in February, and inquired about the understanding and use of ChatGPT by students in various schools. The result is not surprising. After a lot of searching and analysis, The general result is as follows:
Thirty percent of college students use ChatGPT for their written assignments, and of this usage group, nearly 60 percent use it for more than half of their assignments. 3/4 of ChatGPT users consider using it as cheating, but still use it ChatGPT, while 54% of respondents did not know. Of those 46% who were familiar with ChatGPT, 64% (30% of the total sample) said they had used it to help with written assignments.
Among ChatGPT users, 60 percent estimate they use the tool for 50 percent or more of their total written work.
According to the latest news, the University of Hong Kong (HKU) temporarily banned students from using ChatGPT or other artificial intelligence-based tools in courses, assignments and assessments last Friday, becoming the first and only school in the region to implement such a ban .
We learned from the email provided to us by a Zoe classmate of the University of Hong Kong that we sent out last year: Ian Holliday, the vice president of teaching at the University of Hong Kong, said that the University of Hong Kong needs some time to draft a long-term AI tool policy.
He said in the email: "We need to adopt a short-term policy as a temporary measure. We prohibit the use of ChatGPT or any other AI-based tools in HKU, whether in class, assignments, or other aspects." Ian Holliday also said that if Students who want to use AI-based applications must obtain written consent from the course instructor, otherwise any violation will be considered as potential plagiarism.
Currently HKU teachers are asking students to take additional oral or other tests and take other steps to see if there is a suspicion of irregularities, while more universities plan to change written assignments by having students start writing drafts in class ways to prevent plagiarism.
In addition to HKU, many educators we heard in our survey are also agonizing over whether to include ChatGPT in their curriculum or ban it outright. According to our survey results, 46% of respondents indicated that their professor and/or school had banned ChatGPT for assignments, while 29% said they had not banned it, and 26% were unsure. The argument against ChatGPT is that these AI tools undermine the purpose of education, which is to acquire knowledge and skills.
There are both spears and shields, and ChatGPT is the spear of the students, so GPTZero is the shield in the hands of the college instructors.
GPTZero is a tool designed to detect writing generated by ChatGPT. GPTZero was created by Edward Tian, a senior computer science student at Princeton University. This name may be the top reward on the assassination list of students who use ChatGPT heavily in the future.
The eldest brother was born in Tokyo and grew up between Beijing and Toronto. His parents are both working as software engineers in Beijing, and his grandmother is also an electronic engineer graduated from Tsinghua University. He believes that his grandmother is the biggest in his life. one of the inspirations. Edward Tian is currently studying at Princeton University and is also a student reporter for the school newspaper "Princetonian Daily". During the summer, he worked as a software engineering intern at Microsoft Corporation.
After understanding the background, I seem to understand why this big brother invented GPTZero. After all, Brother Tian’s grandmother got Ye Wenjie’s script. (There should be applause from the Three Body Party here)
The principle of GPTZero is to detect AI-generated text by measuring the "perplexity" and "explosiveness" of the text. "Perplexity" is a measure of the randomness of the text, but also a measure of whether the text is familiar to a language model (such as ChatGPT). If a piece of text is very random, confusing, or foreign to a language model, it has a high perplexity score and is more likely to be human-generated. Conversely, if the text is very familiar and the language model has likely seen it, it will not be very confusing to the model and is more likely to be AI-generated.
The "burst" refers to the complexity of the sentence. Humans tend to vary sentence length and have "bursts" in their writing, while AI language models are more consistent. This can be seen by creating a sentence change graph. With a human essay, the variation would be all over the graph, with ups and downs, whereas a machine-written essay would be rather boring, with a constant baseline.
So I can basically be sure that any non-man-made content in the future will face challenges from detection tools like GPTZero. I also understand that even as ChatGPT and other AI language models improve, technologies like GPTZero and other AI detection software will keep pace. After all, training a detection model is much easier than training these giant language models. It costs millions of dollars to train these huge language models such as ChatGPT, and the training cost of detection tools is almost negligible in comparison.
Another piece of good news for universities is that GPTZero has made it clear that it will be freely available to major educational institutions to ensure academic integrity.
I believe this is another reason for a crowdfunding killer.
At the same time, some students find ChatGPT helpful, but schools need to consider how to integrate AI tools into teaching and learning more carefully.
“I am taking ChatGPT to help me teach graphic design, which means that I work with students to explore how to communicate in creative ways through the relationship between text, image and space,” Lisa Maione, a professor at Kansas City Art Institute, once publicly commented .
She said: "I have ChatGPT as a partner in our project. As our department embraces an inquiry-based education model, ChatGPT is another counterparty where we can consider how a question affects our understanding of an answer...I hope ChatGPT As with all the analog and digital tools we use, it has been an interesting partner for our creative and critical work.”
"ChatGPT is not a substitute for critical thinking, critical reading or critical writing," Maione cautions. “In a way, I feel this tool will encourage my students and I to read, write and edit more.” Another 28% of students also think that their advisors “maybe” (23%) or “ Absolutely” (5%) did not know they used the tool in their assignment.
While I don't consider using ChatGPT to be cheating, as long as the resulting work is still original and not plagiarized. However, some may consider it a form of academic dishonesty. Using ChatGPT can be a useful tool for students, but it is important to use it responsibly and ethically, and one of the main reasons students turn to ChatGPT and similar tools is to save time and effort. However, using these tools to complete assignments does not really reflect a student's understanding and true ability of the coursework itself.
In my opinion, it is best to use ChatGPT or any similar tool as a source of inspiration, but all the actual writing and structure should be done by the students themselves. The university learning process is essentially about developing students’ critical awareness and independent thinking to ensure that ChatGPT can be a fun tool to help students read, write and edit more, rather than replace these critical skills.
In the end, I believe that the best solution to this whole thing is to find a professional and caring mentor to solve the problem fundamentally, without crowdfunding or running the risk of being caught by the immature AI tool. Years of college life down to become a glorious: "Prompt Engineering (Prompt Engineering)". Zoe from the University of Hong Kong is feeling good now, with the help of LunwenHelp academic tutors, he is not worried about the school's ban policy on ChatGPT, I wish him success in his studies and good health.