How Large Language Models are Transforming Machine-Paraphrased Plagiarism

Wahle, Jan Philip; Ruas, Terry; Kirstein, Frederic; Gipp, Béla

doi:10.22541/au.167528149.95939897/v1

Cited by 16 publications

(15 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One such factor is to what extent the ICMJE criteria are only applicable to human authors, which is not mentioned explicitly in the criteria themselves. The ICMJE criteria were constructed in 1978 and thus did not have to consider the emergence of artificial intelligence that can produce humanlike text, and text that is undetectable by plagiarism software [17]. It is thus important to have discussions, and an open mind in regards to what these systems can contribute to and what ethical and practical challenges lie ahead of us.…”

Section: Discussionmentioning

confidence: 99%

Does GPT-3 qualify as a co-author of a scientific paper publishable in peer-review journals according to the ICMJE criteria? A case study

Osmanovic-Thunström

Steingrímsson

2023

Discov Artif Intell

View full text Add to dashboard Cite

This paper explores the potential for a system to be a co-author on an academic paper based on the criteria proposed by the International Committee of Medical Journal Editors (ICMJE). We used a third generation generative pretrained transformer (GPT-3) to write a review paper on the topic of its choice: The Effects of Sleep Deprivation on Cognitive Function. The system was asked to fulfill the four main criteria for co-authorship as recommended by ICMJE, which includes contributions to the conception or design of the work; drafting the work or revising it critically for important intellectual content; final approval of the version to be published; and agreement to be accountable for all aspects of the work. Our results showed that the system was able to fulfill the criteria, with significant difficulties with accurate and reliable referencing. We also explored the ethical implications of using AI systems for research and found that it is important to take a cautious approach when considering its use for scientific authorship. This case study provides a methodology for further investigations into the possibilities and limitations of automated writing.

show abstract

Section: Discussionmentioning

confidence: 99%

Does GPT-3 qualify as a co-author of a scientific paper publishable in peer-review journals according to the ICMJE criteria? A case study

Osmanovic-Thunström

Steingrímsson

2023

Discov Artif Intell

View full text Add to dashboard Cite

show abstract

“…One such factor is to what extent the ICMJE criteria are only applicable to human authors, which is not mentioned explicitly in the criteria themselves. The ICMJE criteria were constructed in 1978 and thus did not have to consider the emergence of arti cial intelligence that can produce humanlike text, and text that is undetectable by plagiarism software (12). While systems like GPT-3 are tools which rarely generate original thought as they are trained on existing data, there is a merit in asking to what extent human authors are not experiencing the same phenomena when it comes to academic writing.…”

Section: Discussionmentioning

confidence: 99%

“…Association between reduced sleep duration and weight gain in women American Journal of Epidemiology 171 (11), 1159-1166 Phillips A C et al (2013). The role of poor sleep quality in depressive disorders Current Psychiatry Reports 15 (12), 425 Pires S O et al (2022). Effects of acute total sleep deprivation on cognitive processes: A meta-analysis Sleep Medicine Reviews 56(January) 101737 With the new draft we nally asked the system if the text was good enough for a peer-review journal.…”

Section: Gpt-3 Davinic Language Model: Strategies For Improving Sleep...mentioning

confidence: 99%

Does GPT-3 qualify as a co-author of a scientific paper publishable in peer-review journals according to the ICMJE criteria? - A Case Study.

Thunström

Transformer

Steingrímsson

2022

Preprint

View full text Add to dashboard Cite

This paper explores the potential for a system to be a co-author on an academic paper based on the criteria proposed by the International Committee of Medical Journal Editors (ICMJE). We used a third generation generative pretrained transformer (GPT-3) to write a review paper on the topic of its choice: The Effects of Sleep Deprivation on Cognitive Function. The system was asked to fulfill the four main criteria for co-authorship as recommended by ICMJE, which includes contributions to the conception or design of the work; drafting the work or revising it critically for important intellectual content; final approval of the version to be published; and agreement to be accountable for all aspects of the work. Our results showed that the system was able to fulfill the criteria, although it had difficulty with referencing. We also explored the ethical implications of using AI systems for research and found that it is important to take a cautious approach when considering its use for scientific authorship. This case study provides a methodology for further investigations into the possibilities and limitations of automated writing.

show abstract

“…In [5] the author contributed in machine paraphrased plagiarism. There are many other contributions in language model transforming with machine-paraphrased plagiarism [6].…”

Section: Related Workmentioning

confidence: 99%