What is Research Misconduct?
Research misconduct means “Fabrication, falsification or plagiarism in proposing, performing or reviewing research or in reporting results.”
Elements of Research Misconduct
Fabrication – Refers tomaking up data or results and recording or reporting them
Falsification – Refers to manipulating research materials, equipment or processes, or changing or omitting data or results such that the research is not accurately represented in the research record
Plagiarism – Refers to appropriation of another person’s idea, processes, results or words without giving appropriate credit
What does not constitute research misconduct
Honest error or difference of opinion
Categories of academic misconduct in Artificial Intelligence (AI) era
Data fabrication: Use of artificial intelligence to generate false data or manipulate data to conform to desired outcomes – Misconduct of High severity
Content plagiarism: Use of artificial intelligence technology for text auto-generation without proper citation or acknowledgement of original sources – Misconduct of medium to high severity
Opacity of results: Use of artificial intelligence for data processing and result generation without adequately disclosing methodologies or data sources, lacking replicability and verifiability – Misconduct of medium severity
Consequences of Research misconduct in AI era
Inaccurate findings- manipulated data generates misleading or false findings leading to erroneous conclusions
Reproducibility challenges- Data manipulation erodes reproducibility making it difficult for other researchers to replicate results
Damage to scientific integrity- Research misconduct erodes public trust in the scientific community and tarnishes the reputation of the individuals and institution involved
Decision making- Data manipulation can have severe consequences on public health, safety and well-being of the society
Mitigating the threats of scientific research misconduct using AI
SWOT Analysis: Strength, Weakness, Opportunities and Threats
Rigorous data Governance: Institutions must establish robust protocols for data collection, storage and access
Developing advanced detection tools: To check for check plagiarism
Digital watermarking: Increases the traceability and decreases visual realism, image duplication and manipulation issues
Transparent and open science: Fosters collaboration
Peer review: plays critical role in evaluating quality and research integrity
Ethical guidelines and oversight: play crucial role in evaluating ethical implications of AI research projects and compliance
Education and awareness: Researchers must be educated about the risks of scientific misconduct and data manipulation with AI
Tools for detection and prevention of academic research misconduct in AI era
Data integrity checkers: Scrutinize datasets for anomalies and inconsistencies serving as crucial mechanisms to detect signs of data fabrication or falsifications
Plagiarism detection software: Modern plagiarism detection software can pin-point AI-generated texts and pseudo-original content
Transparency and explainability tools for AI algorithms: designed to shed light on opaque decision-making processes inherent to AI models and thereby promote transparency in scientific research applications
Enhancement with data provenance: Traces lifecycle of data, documenting its origin, movements and transformations
Enhancement with AI model auditing: Comprises systematic evaluations of AI algorithms, accesses their fairness, accuracy, transparency and ethical implications
Mitigation of unethical AI practices through education and training programs for researchers
Offering mandatory courses on “AI Ethics and Regulations”
Introducing elective courses such as “Big Data Management” and “AI-Assisted Statistical Analysis”
Organizing regular AI ethics seminar and workshops
Encouraging interdisciplinary collaborations involving law and scientific community
Establishing online self-study platforms on AI ethics, data management and related topics
Stringent measures to curb research misconduct
Penalizing the individuals involved in research misconduct with penalty
Blacklisting the individuals involved in research misconduct
Retraction of the article
Concluding remarks:
Whether AI is a friend or a foe in a scientific community depends on the purpose for which it is utilized
AI can be a friend when it is employed with the right intent to gather information, assist in the process, accelerate the process, format, organize, analyse, sort and segregate the data
AI may turn out to a foe if it used in a wrong intent solely for the purpose of decision-making, synthesis, innovation or generating new content because the originality of the content may get compromised.
It would be better if researcher relies on his expertise and experience for such activities
Use of AI in a wrong intent may pave way for research misconduct and its consequences.
References:
- https://ori.hhs.gov/definition-research-misconduct
- Chen Z, Chen C, Yang G, He X, Chi X, Zeng Z, et al. Research Integrity in the era of artificial intelligence challenges and responses. Medicine (Baltimore) 2024;103(27):e38811. doi: 10.1097/MD.0000000000038811
- Nair A. Increasing Threat of Scientific Misconduct and Data Manipulation With AI. Enago Academy. 2023/08/14. https://www.enago.com/academy/scientific-misconduct-and-datamanipulation-with-ai/
- Birks D, Clare J. Linking artificial intelligence facilitated misconduct to existing prevention frame works. International Journal of Educational Integrity 2023;19:20. https://doi.org/10.1007/s40979-023-00142-3
