This study aims to provide a complete overview of the rate of retraction in the field of artificial intelligence (AI) and machine learning, including major reasons for retraction, journal analysis, author analysis, citation pattern in retraction and keyword analysis.
This research collected 3,456 AI and machine learning retraction publications from the Lens.org database covering the period of 1974–2024. Retraction status and reasons were cross-checked with OpenAlex, Crossref and Retraction Watch. Data fetching was performed using OpenRefine through REST API calls, whereas data visualization was done using VosViewer, Scimago Graphica and MS Excel.
This study found a 0.05% retraction rate, with 71.7% published in OA. The papers often continue to be cited, with top-cited papers receiving up to eight times more citations in post-retraction. Retractions were mainly due to error and concern (60.50%), falsification and manipulation (23.21%) and ethical issues (7.09%). Most retracted papers appeared in Scopus-indexed Q1 journals, primarily from Hindawi Limited publisher. China, India, the USA, the UK and Iran had the highest retraction rate, with top authors and institutions frequently affiliated with China. Notably, 32.72% of retracted papers consist of double-authored.
This study was limited to investigating the rate of retraction articles in the domain of AI and machine learning indexed in Lens.org.
There were several retraction studies in different disciplines, but no study explores retraction in the domain of AI and machine learning. To bridge the research gap, this study provides a comprehensive overview of retraction in the field of AI and machine learning.
