Generative large language models and academic integrity: ethical risks, detection challenges, and Governance in the age of AI

This paper explores methods to automatically predict lexical complexity in a multilingual setting using advanced natural language processing models. More precisely, it investigates the use of transfer learning and data augmentation techniques in the context of supervised learning, showing the great interest of multilingual approaches. We also assess the potential of generative large language models for predicting lexical complexity. Through different prompting strategies (zero-shot, one-shot, and chain-of-thought prompts), we analyze model performance in diverse languages. Our findings reveal that while generative models achieve high correlation scores, their predictive quality varies. The comparative study illustrates that while generative large language models have potential, optimized task-specific models still outperform them in accuracy and reliability.

The quality of teaching, learning and assessment is compromised by the growing problem of academic dishonesty, especially in large class sizes as a result of the ‘massification’ of education. In South Africa and around the world, student plagiarism and cheating has become a matter of concern, especially when it comes to teaching large classes. This concern has received much attention as it impacts negatively on the maintenance of academic standards and integrity at many universities. Academics have a major role to play in the process of maintaining academic integrity. Through an ‘interpretivist’ and qualitative approach, we explored the experiences of three emerging academics within the Discipline of Curriculum Studies at a university in South Africa. We used Pinar’s method of currere as a lens that focuses on academics’ experiences of assessment and plagiarism in teaching large classes and its effect on academic integrity. The findings suggest that although ‘massification’ of education in South Africa is commended for addressing past social injustices and for facilitating accessibility to education, quality teaching and learning including assessment is seriously compromised. This demands a serious rethink of assessment strategies to deter academic dishonesty, and a reconsideration of the way academics and institutions think about plagiarism detection tools in teaching large classes. Significance: Understanding academics’ experiences of assessment and addressing the growing problem of plagiarism can contribute significantly to efforts towards improving teaching and assessment practices in large classes, and to upholding academic honesty within higher education institutions in South Africa. A rethink of effective assessment strategies is needed to provide a worthwhile quality educational experience. In the context of this study, ethics within the teacher education curriculum should be prioritised.

Evaluating generative patent language models

As you know, the principles of academic culture and integrity in the current state of university education in Ukraine are often simply perceived as an abstract, but necessary for the European integration of education postulate. Codes of ethics and regulations on combating academic plagiarism are created on the basis of recommendations of the Ministry of Education and Science, regulate the activities of employees and students in the context of prevention and detection of falsifications, violations of academic integrity, including academic plagiarism. The establishment of the principles of academic culture and integrity in educational institutions is a necessary prerequisite for building a strong civic culture in society. Expert evaluation of scientific papers, including anti-plagiarism, in our opinion is an important prerequisite for the adequacy of complex scientific topics and objectifies the overall assessment of the quality of presentation of research results, including dissertations,monographs, manuals, articles and abstracts. The aim of the work was to improve measures to combat academic dishonesty and plagiarismbased on the analysis of current legislation, orders and regulations of Danylo Halytskyi Lviv National Medical University, creation and implementation of a new discipline for PhD students to acquire the necessary knowledge, skills and abilities. based on the topics of meaningful modules, created taking into account the experience of examination of scientific and educational works. In total for the three-year period (2018-2020) 5992 inspections were carried out and violations of academic integrity were confirmed in 611 works, which was the reason for their rejection. Algorithmic logical functional sequence of anti-plagiarism examination of dissertations and abstracts with analysis of the percentage of uniqueness of the text by the method of shingles with optimized settings was used to conduct the initial examination of dissertations to detect academic plagiarism. Deep search parameters and deep verification by Unichek, Plagiarism Detector Pro v. 1092”, etc. with further maternity check and establishment of the percentage of uniqueness of the text by the freely accessible program “AdvegoPlagiatus”. Based on the analysisof the causes, trends and spread of academic plagiarism in the university community for applicants and graduate students was created and implemented the discipline “Information support, academic integrity and research ethics” which is aimed at gaining analytical competencies, flexible skills “soft skills”. necessary during the planning and implementation of dissertation research and ensuring adequate communication in the scientific community. It consists of three content modules “Academic Integrity and Anti-Plagiarism”, “Scientific Ethics”, “Information Technology and Patent and Literary Support of Research Papers”. Lectures, seminars and topics of independent work are aimed at acquiring analytical competencies, flexible skills “soft skills” needed during the planning and implementation of dissertation research and ensuring adequate communication in the scientific community, the formation of graduate students’ theoretical knowledge and practical skills and skills in compliance with ethical and moral and legal principles in the planning and implementation of scientific research with human participation and with the use of laboratory animals, obtaining by graduate students and applicants knowledge of research activities and acquisition of skills of independent research, features of patent research and preparation of reporting documentation. Postgraduate students acquire knowledge of current legislation on academic integrity and anti-plagiarism; ethical aspects of scientist behavior and communication; research planning, organization and tactics strategies; liability for breach of academic integrity; basics of statistical conclusion, counteraction to conflicts of interest, elimination of risk factors and errors in ensuring academic integrity, which determines the social and professional mobility of research and teaching staff, promotes diversification of higher education, increasing the value of scientific knowledge and professional experience. Prospects for further research are to improve algorithms for detecting academic plagiarism and scientific falsifications, optimizing measures to prevent anddetect manifestations of academic dishonesty.

Higher education is founded on academic integrity although plagiarism has continued to be a challenge in Pakistan. As the use of digital resources has also become more common, the threat of academic dishonesty has increased, and the application of effective tools that can check plagiarism has become a necessity. Turnitin and iThenticate are artificial intelligence (AI)-powered plagiarism detector tools that have opened a novel set of possibilities in enhancing originality, responsibility, and academic integrity. The research paper has a cross-sectional quantitative research design that investigates the effect of AI-based plagiarism detection tools on academic honesty among Pakistani tertiary education. A sample of 200 respondents in 50 participants were then sampled in general and private universities using stratified random sampling method. The sample considered undergraduate and post graduate students between 18-30 years. The data were gathered by using a structured questionnaire aimed at the measurement of awareness, frequency of use, and perception of AI plagiarism tools and their perceived effect on the writing behavior and commitment to academic integrity among the students. Descriptive statistics, chi-square tests and logistic regression were applied to the data to determine the relationships between demographics, the usage of the tool and reported academic honesty. The results indicate that AI-based plagiarism detection improves the knowledge of academic integrity and deters unethical activities, but such problems as excessive reliance on machine-generated reports and insufficient training are still major obstacles. The research also notes that incorporation of plagiarism detecting tools with supplementary actions, including ethics education and institutional policy reinforcement, is the key way to enhance the culture of academic honesty in Pakistani higher education.

The rapid advancement of digital technologies has significantly impacted academic practices, particularly in the area of plagiarism detection. As universities and research institutions adopt tools to safeguard academic integrity, concerns arise about their effectiveness and potential limitations. This study investigates the role of automated plagiarism detection tools in higher education, examining how they influence academic practices and the detection of both traditional and AI-generated plagiarism. Despite the sophistication of tools like Turnitin, PlagScan, GPTZero, and QuillBot, the research finds that these systems often struggle with accurately interpreting context, resulting in false positives and overlooked instances of plagiarism. The study underscores the necessity of combining technology with human judgment, recognizing that such tools should be seen as supplementary rather than definitive measures of originality. Grounded in theoretical frameworks such as Technological Determinism, Actor-Network Theory (ANT), and Socio-Technical Systems Theory (STS), the research highlights the complex relationship between technology, academia, and societal expectations. Through a qualitative analysis of existing literature, the study identifies key challenges and suggests that hybrid approaches, blending technological tools with human oversight, may offer a more balanced and effective approach to plagiarism detection. The findings encourage further exploration into the ethical implications of reliance on automated systems in education and their broader impact on academic integrity.

Academic integrity policies against assessment fraud in postgraduate studies: An analysis of the situation in Spanish universities

Artificial Intelligence (AI) has evolved into an indispensable tool in education. AI usage in education permeates tutoring systems, automated essay scoring, plagiarism detection, virtual reality simulations, and chatbot-based learning support. This ubiquity has threatened the tenets of academic integrity upon which the entire education system hinges. This present conceptualization focuses on demystifying the concepts and conversations at the nexus of AI adoption and academic integrity. The conceptualization deeply explored the development of AI and the motivation for its deployment in education. A broad overview of academic integrity highlights the core values of honesty, trust, fairness, respect, responsibility, and courage. This was followed by a detailed exploration of the techniques used by students to avoid detection of AI-generated work. A focal discussion was then provided on the impact of AI-generated writing tasks on students’ academic integrity, highlighting both opportunities and challenges. Next, the technical, procedural, educational, and collaborative strategies for detecting and minimizing the rate of AI-generated work among students were discussed. The Technology Acceptance Model and Academic Integrity Framework were discussed as conceptualizations' theoretical foundations. The conceptualization closes with a summary of recent empirical research emphasizing the need for further studies to explore all ramifications of the influence of AI on academic integrity. It is hoped that the conceptual clarity provided in this work will support the emerging scholarship on AI's influence on society.

Artificial intelligence (AI) generative large language models are powerful and increasingly accessible tools with potential applications in health care education and training. The annual Orthopaedic In-Training Examination (OITE) is widely used to assess resident academic progress and preparation for the American Board of Orthopaedic Surgery Part 1 Examination. Open AI's ChatGPT and Google's Bard generative language models were administered the 2022 OITE. Question stems that contained images were input without and then with a text-based description of the imaging findings. ChatGPT answered 69.1% of questions correctly. When provided with text describing accompanying media, this increased to 77.8% correct. In contrast, Bard answered 49.8% of questions correctly. This increased to 58% correct when text describing imaging in question stems was provided (P<.0001). ChatGPT was most accurate in questions within the shoulder category, with 90.9% correct. Bard performed best in the sports category, with 65.4% correct. ChatGPT performed above the published mean of Accreditation Council for Graduate Medical Education orthopedic resident test-takers (66%). There is significant variability in the accuracy of publicly available AI models on the OITE. AI generative language software may play numerous potential roles in the future in orthopedic education, including simulating patient presentations and clinical scenarios, customizing individual learning plans, and driving evidence-based case discussion. Further research and collaboration within the orthopedic community is required to safely adopt these tools and minimize risks associated with their use. [Orthopedics. 2024;47(3):e146-e150.].

The construction of hypernym taxonomic trees, a critical task in the field of natural language processing, involves extracting lexical relationships, specifically creating a tree structure that represents hypernym relationships among a given set of words within the same domain. In this work, we present a method for constructing hypernym taxonomy trees in the Chinese language domain, and we named it CHRRM (Chinese Hypernym Relationship Reasoning Model). Our method consists of two main steps: First, we utilize pre-trained models to predict hypernym relationships between pairs of words; second, we regard these relationships as edges to form a maximum spanning tree in the word graph. Our method enhances the effectiveness of constructing hypernym taxonomic trees based on pre-trained models through two key improvements: (1) We optimize the hyperparameter configuration for this task using pre-trained models from the Bert family and provide explanations for the configuration of these hyperparameters. (2) By employing generative large language models such as ChatGPT and ChatGLM to annotate words, we improve the accuracy of hypernym relationship identification and analyze the feasibility of applying generative large language models to the task of constructing taxonomy trees. We trained our model on subtrees of WORDNET and evaluated its performance on non-overlapping subtrees of WORDNET, demonstrating that our enhancements led to a significant relative improvement of 15.67%, achieving an F1 score of 67.9 on the Chinese WORDNET validation dataset compared to the previous score of 58.7. In conclusion, our study reveals the following key findings: (1) The Roberta-wwm-ext-large model consistently delivers outstanding results in constructing taxonomic trees. (2) Generative large language models, while capable of aiding pre-trained models in improving hypernym recognition accuracy, have limitations related to generation quality and computational resources. (3) Generative large language models can serve various NLP tasks either directly or indirectly; it is feasible to improve the downstream NLU task's performance through the generative content.

Plagiarism is a growing concern for academia across the globe. Several factors influence the behaviour of the researcher towards plagiarism. The UGC (Promotion of Academic Integrity and Prevention of Plagiarism in HEIs) Regulation, 2018 was notified to promote academic integrity in HEIs and curb plagiarism. However, this regulation has many gaps which need to be addressed in the quest for achieving academic integrity. This paper is an attempt to identify these gaps in the regulation. It also attempts to address the over reliance of academic fraternity on Plagiarism Detection Tools.

This study examines the effect of QuillBot on the writing skills and academic integrity of eleventh-grade students. This quantitative study addresses three main questions: 1) Is there any simultaneous significant effect of the use of the QuillBot on eleventh-grade students’ writing skills and academic integrity? 2) Is there any significant effect of the use of the QuillBot on eleventh-grade students’ writing skills? and 3) Is there any significant effect on the use of the QuillBot for eleventh-grade students’ academic integrity? Data were collected through writing tests and academic integrity questionnaire, with analysis conducted using MANOVA. The findings indicate that QuillBot significantly improves students’ writing skills by enhancing clarity, coherence, and accuracy. Furthermore, QuillBot influences academic integrity by promoting ethical writing practices, such as proper paraphrasing and plagiarism detection. The study concludes that is a valuable tool for improving both writing skills and maintaining academic integrity in educational settings when it is used responsibly and under supervision. Recommendations are provided for educators on integrating QuillBot into their teaching strategies to maximize its benefits while minimizing potential misuse.

Generative unigram language models have proven to be a simple though effective model for information retrieval tasks. In contrast to ad-hoc retrieval, topic tracking requires that matching scores are comparable across topics. Several ranking functions based on generative language models: straight likelihood, likelihood ratio, normalized likelihood ratio, and the related Kullback-Leibler divergence are evaluated in two orientations. Best performance is achieved by the models based on a normalized log-likelihood ratio. Key component of these models is the a-priori probability of a story with respect to a common reference distribution.

The public release of ChatGPT, a generative artificial intelligence language model, caused wide-spread public interest in its abilities but also concern about the implications of the application on academia, depending on whether it was deemed benevolent (e.g., supporting analysis and simplification of tasks) or malevolent (e.g., assignment writing and academic misconduct). While ChatGPT has been shown to provide answers of sufficient quality to pass some university exams, its capacity to write essays that require an exploration of value concepts is unknown. This paper presents the results of a study where ChatGPT-4 (released May 2023) was tasked with writing a 1500-word essay to discuss the nature of values used in the assessment of cultural heritage significance. Based on an analysis of 36 iterations, ChatGPT wrote essays of limited length with about 50% of the stipulated word count being primarily descriptive and without any depth or complexity. The concepts, which are often flawed and suffer from inverted logic, are presented in an arbitrary sequence with limited coherence and without any defined line of argument. Given that it is a generative language model, ChatGPT often splits concepts and uses one or more words to develop tangential arguments. While ChatGPT provides references as tasked, many are fictitious, albeit with plausible authors and titles. At present, ChatGPT has the ability to critique its own work but seems unable to incorporate that critique in a meaningful way to improve a previous draft. Setting aside conceptual flaws such as inverted logic, several of the essays could possibly pass as a junior high school assignment but fall short of what would be expected in senior school, let alone at a college or university level.

On Chatbots and Generative Artificial Intelligence.