Cactus Communications (Representative Director Makoto Yuasa) announced on April 25 that a survey of approximately 450 researchers (including users of its services) on the use of AI in the writing of academic papers reveals the rapid spread of AI tool use. 75% of the respondents reported an increase in their use of AI tools compared with the year before last. The study also found that approximately 70% of the most-used AI tools are free and publicly available.
The survey was conducted between March 25 and April 8 this year among 452 researchers and collaborating researchers using Cactus' English-language paper editing service "Editage" and its AI English-language paper writing support tool "Paperpal." Respondents were asked about the use of AI tools when writing papers, the types of tools used and the purpose for using them. 75% of respondents (36.9% indicated a "significant increase" and 38.1% a "moderate increase") reported an increase in the use of AI tools compared with the year before last. Furthermore, 8.8% had submitted papers using only AI tools (excluding data). Of these, 75% were accepted (regardless of the journal in which the papers were published).
The top three most frequently used AI tools were DeepL (41.8%), ChatGPT (20.4%) and Google Translate (11.1%), with free-of-charge services accounting for approximately 73% of the total. In terms of purpose and use, 28.9% of the respondents used the tools to proofread or review articles and 28.2% for translation. The next most common purposes were idea generation (9.7%) and writing research (7.1%). In addition, 92.9% of the respondents said that being able to effectively use AI tools would be advantageous in the development of their research careers.
When asked to select the impact of using AI tools, many chose productivity benefits in terms of work efficiency, such as "reduced time spent on research" and "increased speed in writing papers," or in terms of the productivity of "increased number of papers written." However, when asked whether the use of AI tools would mean that they could "contribute to enhancing the level of papers to be submitted," only 23% of respondents answered, "I think so," whereas 26.8% answered "I do not particularly think so" and 41.2% answered "I do not know."
Kentaro Iwata, a senior manager at Cactus who led the survey, said that the use of AI tools is already widespread among researchers and that they are moving to a phase where their ability to master AI tools is influencing their career development. Iwata identified the need for mechanisms to control or verify issues such as unintentional fabrications caused by generative AI in the future and stated that Cactus will continue its investigations in the future.
This article has been translated by JST with permission from The Science News Ltd. (https://sci-news.co.jp/). Unauthorized reproduction of the article and photographs is prohibited.