Network Security Internet Technology Development Database Servers Mobile Phone Android Software Apple Software Computer Software News IT Information

In addition to Weibo, there is also WeChat

Please pay attention

WeChat public account

Shulou

Nature | all over the world, 1 AI 3 postdoctoral students use ChatGPT every day, without using the AI tool to affect job hunting.

2025-03-29 Update From: SLTechnology News&Howtos shulou NAV: SLTechnology News&Howtos > IT Information >

Share

Shulou(Shulou.com)11/24 Report--

[guide to Xin Zhiyuan] Nature published a survey of postdoctoral researchers around the world. Most people think that chat robot tools like ChatGPT greatly facilitate their scientific research and daily work, showing great potential for this revolutionary tool.

A global postdoctoral survey conducted by Nature shows that 1/3 of respondents are using AI chatbots to help them modify text, generate or edit code, and organize literature in related fields.

Recently, Nature published an article describing, from the perspective of postdoctoral researchers from all over the world, how ChatGPT can help postdoctoral researchers adapt to life in a foreign country, overcome language barriers, focus on scientific research, and save researchers a lot of time in scientific research.

Chairman OpenAI also forwarded this article as an example of academic recognition of the effectiveness of ChatGPT.

Netizens are even more direct, thinking that "there are postdoctoral users of 1Accord 3", which means that only 1Accord 3 is willing to admit that the percentage of people who really use the meaning is higher than this.

The advent of ChatGPT has changed everything. Rafael Bretas, a postdoctoral fellow from Brazil, has lived in Japan for more than a decade and speaks spoken Japanese very well.

But the cumbersome requirements of written Japanese, such as strict honorific norms and hierarchy, still confuse the Brazilian postdoctoral student in a foreign country.

As a result, he often has to write emails to his superiors and colleagues in English. However, due to the limitation of the English proficiency of both sides, this way of communicating in the second language of both sides often leads to misunderstandings.

When OpenAI launched ChatGPT in November 2022, Bretas, which studies the cognitive development of primates, was quickly tested at RIKEN, the national research institute in Kobe, Japan.

He hopes ChatGPT can help him write standard written Japanese easily.

At first his expectations were not high because he heard that chatbots were not very good at languages other than English.

Moreover, he conducted experiments in his native Portuguese and found that the resulting text "looks very naive".

However, when he used a chat robot to modify several Japanese emails and asked his Japanese friends if they were in line with written Japanese etiquette, he received very positive feedback.

Now, Bretas relies on this chat robot to write formal Japanese emails every day, which is very easy to use.

This not only saves him time, but also reduces his sense of frustration. He is now able to express his views and ideas in Japanese more quickly and accurately. "it makes me more confident to continue my current research work," he said.

Excitement and fear coexist since the launch of ChatGPT, workers worried that it would "change their lives", triggering a wave of job losses and leading to an economic downturn.

Researchers can no longer sit still and begin to explore the potential of this magical tool.

Researchers have found that it can help them deal with a range of daily tasks, from writing abstracts to writing code.

Most researchers think that ChatGPT can greatly save time, while others worry that it may lead to the proliferation of low-quality research.

Last month, the journal Nature published the results of a survey of researchers' views on the widespread use of artificial intelligence in science, resulting in both excitement and fear.

However, few studies have seriously discussed how researchers should use ChatGPT to help them with their research.

To delve deeper into this issue, Nature added questions about artificial intelligence applications in its global postdoctoral survey in June and July.

The survey showed that 31 per cent of working respondents had used chat robots, but 67 per cent said AI had not changed their daily work or career plans.

Forty-three per cent of researchers who use chatbots use it once a week, while only 17 per cent use it every day, such as the Brazilian postdoctoral Bretas mentioned above.

Mushtaq Bilal, a postdoctoral fellow in comparative literature at the University of Southern Denmark, often expresses his views on the application of artificial intelligence in the academic field, which he believes will soon increase in the future.

"it's too early for postdocs to say whether artificial intelligence has changed their daily work," he said. According to his observation, because of the inherent institutional inertia, it is often difficult for researchers and scholars to respond sensitively to the emergence of new tools and technologies immediately.

How to use this all-around digital assistant, however, the Nature survey is only aimed at researchers, and it is hard to say whether it can reflect the use of chatbots in other professional scenarios.

A July survey by the Pew Research Center, a Washington-based think-tank, found that only 24% of people in the United States had heard of ChatGPT and tried to use it.

Article address: https://www.pewresearch.org/ short-reads / 2023-08-28 / most-americans-havent-used-chatgpt-few-think-it-will-have-a-major-impact-on-their-job/ but the actual usage rate is less than 1%, while college-educated people account for 1/3 of the total user population.

Another survey of Swedish college students in April and May found that 35 per cent of 5894 respondents regularly used ChatGPT. In Japan, 32 per cent of college students surveyed said they had used ChatGPT.

According to a survey by Nature magazine, the most common use of chatbots is text rewriting, which accounts for 63%.

Teng Xinzhi, a postdoctoral fellow in radiology at the Hong Kong Polytechnic University, says he uses chatbots every day to refine English texts, draft documents and write presentations because English is not his mother tongue.

He said he might ask ChatGPT to "polish" a paragraph to make it look "fluent and professional".

Or he will ask ChatGPT to generate an alternative title for the summary he wrote, and then he will choose the most appropriate title according to the style he needs from the suggestions.

In this way, he can save the money he used to spend on professional editing services.

Ashley Burke, a postdoctoral researcher on malaria at the University of Whitwater Sland in Johannesburg, South Africa, said she used ChatGPT to help when she needed to "help write a short beginning" when she was writing stuck.

For example, she asked ChatGPT to "write an introduction to the incidence of malaria in Zambia". The results generated by AI could lead to more creative content of her own.

She also uses ChatGPT to simplify scientific concepts, which can not only help her understand scientific concepts, but also help her to convey these complex scientific concepts to others in simple language.

She thinks this is "the most useful place she has found artificial intelligence so far".

For example, when writing the "Research methods" section of the paper, she was not sure how to describe the "DNA sequence analysis" description. She would ask ChatGPT, "how do you check for DNA sequence polymorphism?" "

ChatGPT will come up with a complete plan of 10 steps: from the beginning of data collection to the end of the report.

This will help her solve the "thorny problems" in the process of writing her thesis.

Bilal notes that surveys show that engineers and social scientists are more likely to use chat robots, which is consistent with his observations. However, he also found that biomedical scientists in Denmark also actively use chatbots.

However, a large number of engineering postdoctoral students rely on ChatGPT to polish their texts (up to 82%), which makes him vaguely worried.

Because such a high proportion may indicate that engineers do not have enough training in science writing.

He pointed out: "although the AI chat robot can alleviate the problem of science writing to some extent, engineers should also strengthen the practice of science writing." Because it's a vital skill for scientists. "

According to a survey by Nature magazine, about 56% of postdocs use chatbots to generate, edit and troubleshoot code errors.

For example, Iza Romanowska, a postdoctoral fellow in archaeology at Aarhus University in Denmark, points out that ChatGPT is a great help in guiding self-taught programming, adding more comments to the code.

Although these comments do not affect the functionality of the code, they help others understand the code.

It also improves the transparency of the code, she added, because many non-professional programmers are reluctant to open source their own code because of the extra work they need to do to clean up the code.

Archaeologist Iza Romanowska uses ChatGPT to solve the problem of self-taught writing code. Antonio Sclocchi is a postdoctoral student in physics who works in machine learning at the Federal Institute of Technology in Lobsang, Switzerland. She is also using ChatGPT to assist in writing her own code.

Also, she upgraded her ChatGPT to GPT-4 because GPT-4 performed better on some code tasks.

In addition, she uses ChatGPT to generate exam questions and illustrations in LaTeX format.

Emery Berger, a computer scientist at the University of Massachusetts Amherst, said he was pleased with the "Nature" survey and said the results were of profound significance.

Although the percentage of postdoctoral researchers who use chatbots is lower than expected, academia is "shockingly skeptical" of artificial intelligence tools such as ChatGPT.

Critics of chatbots usually never try to use ChatGPT in person.

And when people really try, they often only see the shortcomings, rather than explore the revolutionary potential of the technology.

"it's like waving a magic wand and the Statue of Liberty suddenly appears in front of her, but some people will only focus on her missing an eyebrow rather than marvel at her amazing ability to create the Statue of Liberty! "

Chatbots are very helpful for junior researchers whose native language is not English.

Researchers use ChatGPT as an editorial assistant to help them correct students' papers and application letters. It has played an important role in polishing the abstracts of papers and other written materials that need to be submitted to journals for review.

"you can clearly see that their English has improved a lot. "

Rafael Bretas uses ChatGPT to help improve how Japanese emails he writes to colleagues use AI tools. Lack of specification and guidance Berger believes that most postdocs will actively look for and try various artificial intelligence tools.

But of the three researchers, Bretas, Romanowska and Sclocchi, only Bretas mentioned that his organization had issued official guidelines on how employees use artificial intelligence chatbots.

The guidelines emphasize principles such as prohibiting employees from entering private or personal information into chatbots because they cannot guarantee the privacy of tools such as ChatGPT to process data.

The guide, released in May, also advises users to ensure that the use of chatbots does not violate the agency's copyright rules, and to separately verify the accuracy of the results generated by chatbots.

Romanowska says her university has not issued any formal guidelines or advice on how to use chatbots responsibly.

This seems to be quite common: in a survey of Swedish students, 55% of respondents said they did not know whether their school had guidelines on the use of artificial intelligence.

Romanowska added: "the only rule of my university is that students are not allowed to use ChatGPT for homework or exams. "

She thinks the school's response is a little out of date:

"this is a tool and students should be taught how to use it. We all use it at work, and pretending it doesn't exist doesn't change the fact that everyone is using ChatGPT. "

Tina Persson, a career planner in Copenhagen, says many of her new researchers' clients are pessimistic about artificial intelligence tools.

"it's bad for their careers," because many of them may end up in industry because they haven't got permanent academic positions, but industry is rapidly embracing the new technology.

Dirty work AI doing, invention and creation of their own academia may be relatively slow to accept artificial intelligence. According to a survey by Nature, about 2/3 of postdoctoral students do not think that artificial intelligence has changed their daily work and career plans.

However, 2/3 of those who did use ChatGPT said it had an impact on the way they worked.

Postdoctoral interviewees agree that ChatGPT is an excellent tool to help academic work get rid of tedious tasks.

Romanowska says that as she coaches students, she encourages people to write code in ChatGPT, especially when they are trying to Debug, she thinks:

"copying and pasting the problematic code into ChatGPT and asking what's wrong with it makes the Debug process very easy. ChatGPT can not only accurately point out problems, but also analyze potential problems. "

Mushtaq Bilal said that artificial intelligence chatbots should not replace good science writing training programs. Most of the respondents also realized the limitations of the tool. 29% of postdoctoral respondents said they used it to find the literature, while Bilal expressed his concern:

These chatbots may falsify references to papers that do not exist, and untrained researchers may end up wasting a lot of time on verification.

Sclocchi also pointed out that if users rely too much on chatbots, there will still be problems.

Although these tools can provide advice in article writing, including the reformulation of structures and paragraphs, in the end, decide which story to tell, how to tell your story to the audience, and how to integrate all kinds of information. this is something that artificial intelligence cannot replace researchers to do.

Writing code with artificial intelligence tools can improve productivity, but thinking about how to build code and how to connect code to your domain is a task that artificial intelligence cannot accomplish.

"all this work requires deep thinking. "

Romanowska believes that chatbots can solve some of the tasks in her work, but many problems really can't help.

For example, dealing with important administrative tasks, responding to reviewers' suggestions, writing cover letters, applying for positions, and writing abstracts are all technical tasks that chatbots can effectively help handle.

However, academic work takes time, thoughtfulness and originality, and chatbots cannot meet these needs.

This is the core task that we really need to accomplish ourselves. "

Reference:

Https://doi.org/10.1038/d41586-023-03235-8

This article comes from the official account of Wechat: Xin Zhiyuan (ID:AI_era)

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.

Views: 0

*The comments in the above article only represent the author's personal views and do not represent the views and positions of this website. If you have more insights, please feel free to contribute and share.

Share To

IT Information

Wechat

© 2024 shulou.com SLNews company. All rights reserved.

12
Report