University professor 'stole' ChatGPT to help students write recommendation letters! Successfully won Cambridge Scholarship

ChatGPT has mixed reputations in the education community. Teachers believe that AI makes the homework they leave meaningless, while students believe that it is better to let AI write the homework that is also meaningless in the first place to save time.

However, things may have reversed recently, but the protagonist of the story has changed from "students cheating" to "professors improving work efficiency."

Recently, Ian Bogost, a professor, author and game designer at the School of Computer Science and Engineering at Washington University in St. Louis, wrote an article. After interviewing many university professors, he found that ChatGPT is indeed useful to professors. It is a productivity tool that can get twice the result with half the effort when writing letters of recommendation and making course outlines.

Even a professor used the template recommendation letter written by ChatGPT and reversed it to write the "most humane recommendation letter" and helped the student obtain the Cambridge scholarship.

However, regarding the issue of recommendation letters, some netizens left a message saying that they had already solved it and there was no need for AI to take action. I was responsible for recommending and the students were responsible for writing.

ChatGPT scandal reversed in two episodes

Since the birth of ChatGPT, major universities have successively launched pursuit orders against ChatGPT, as if it was shrouded in There is a dark cloud over the education world, and many negative arguments such as "the college essay is dead", "the end of high school English", "students' homework will all be written by AI" are constantly being spread.

Some colleges and universities have launched investigations into ChatGPT, updated academic integrity warnings in their syllabuses, and even launched special courses to discuss it.

However, as the drawbacks of ChatGPT are gradually exposed, these professors no longer care about the "cheating problem" and are thinking more about how to use ChatGPT to automatically complete their tasks?

Large language models may be quite useless at generating "accurate facts and knowledge", but they are very good at solving tasks that are not related to work output and can give very reasonable output.

ChatGPT unifies the student and teacher teams on the same front for tasks that are very repetitive and unimportant.

Take "Recommendation Letter" as an example. ChatGPT cannot explain why a professor would (or would not) recommend a specific candidate for a specific role, but ChatGPT can provide a detailed template. You only need Just modify a few key information.

A professor at the University of Texas who wishes to remain anonymous uses ChatGPT to type a draft every time before "giving a lecture" or "writing a letter of recommendation". He believes that this phenomenon is very common. , but this shortcut may be considered as avoiding work, but there is still so much important work to be done, writing recommendation letters is definitely not a high priority, and ChatGPT can cut the time of writing letters in half.

Another scholar who uses AI writing tools, Matt Huculak, head of advanced research services at the University of Victoria Library, believes that there is a hidden secret in academia, and most professors will be based on "excellent" , "good" (good), "average" (average) to classify letters, and then fine-tune and reuse them according to the specific situation.

But Huculak wanted to know whether the emergence of ChatGPT could end this phenomenon, especially for "top" students who could not be defined by the template, so he conducted an experiment and asked ChatGPT to write a letter to outstanding students. The letter of recommendation was not used as a template. Instead, the output of ChatGPT was used as a negative teaching material. Then Huculak began to write a completely different and anti-formula letter of recommendation.

Huculak said that the writing process was a feeling he had not experienced in a long time. It was a very humane and heartfelt recommendation letter. This student also successfully obtained an admission to Cambridge University with the recommendation letter. Scholarship (prestigious scholarship).

With this successful experience, Huculak began to use ChatGPT in his own work, and re-wrote an anti-formula text based on the output. He felt that the feeling of "rearranging the material" was very comfortable, and then Don’t be afraid of blank documents anymore.

Stephanie Kane, a lecturer at George Mason University, also believes that ChatGPT perfectly solves the problem of "everything is difficult at the beginning".

Every time Kane starts developing a syllabus for a new course, she asks ChatGPT for ideas. “ChatGPT is a bit like a rubber duck that talks back,” she said. (By explaining codes and documents to the little yellow duck, you can inspire inspiration and discover contradictions.)

But Kane soon discovered that ChatGPT cannot provide real reading materials such as relevant books and papers that exist in reality. Related topics or concepts.

Compared with asking colleagues, Kane thinks ChatGPT is better. At least it will not increase pressure on colleagues. I can ask any questions without worrying about being said to be stupid or asking questions without preparation.

Huculak and Kane both use ChatGPT to break away from formulaic templates, but Hank Blumenthal, a filmmaker who has straddled both industry and academia, hopes to gain some secular insights through ChatGPT (cliché) .

Blumenthal, who had not been offered academic jobs in his field, wondered whether the position statement he requested on diversity, equity and inclusion was too unusual for academic tastes.

He believes, "My diversity statement is about all the movies I've made, I've hired blacks, Asians, women, diverse crews, directors and actors. However, I think schools want It’s something else."

As for what it is specifically, Blumenthal believes that ChatGPT has the ability to help write expected text without being based on its own pluralistic stance.

Another American university professor who asked not to be named admitted that he used ChatGPT to generate formal "assessment criteria" (assessment criteria), and these criteria are now part of course and degree applications.

What the professor meant was that the research was excellent and sounded like what someone with no knowledge of the field would want to hear when evaluating a course. The material generated was good enough to become a real proposal.

A common regret for large language models is that ChatGPT does not offer originality after being trained on piles of existing material.

But professors are not often asked to come up with something truly new. Most of the work they do every day is office work, such as writing letters, processing forms, and writing reports. Artificial intelligence is fully capable of this. Labor, or at least provide a sense of superiority over AI.

The same is true for students, who can also feel overwhelmed and overworked: exhausted by the demands of different professors and not fully understanding the specific requirements; overwhelmed by tuition fees coming; feeling confused about their future prospects; and also undergoing the test of transition to adulthood.

Students come to college first to gain a college experience and secondly to learn and earn certificates.

College lecturers may view classwork as a mere commodity that can be tainted by chatbot intervention; however, students will view these assignments as distractions that prevent them from realizing what they really want. do what.

So in these areas, AI just helps remove annoying roadblocks so we can all get on with doing the things that really matter.

Reference materials:

https://www.php.cn/link/cfa3a0bc94975cb9c346a585ccb3ad9e

https://www.php .cn/link/be16d5d77fc088f250f94227280ec528

The above is the detailed content of University professor 'stole' ChatGPT to help students write recommendation letters! Successfully won Cambridge Scholarship. For more information, please follow other related articles on the PHP Chinese website!