Significant disparities persist in how researchers from low- and middle-income countries (LMICs) and high-income countries (HICs) participate in agenda-setting and knowledge production. Rapid advancement in artificial intelligence (AI) might contribute to improving research capacity in LMICs. This review aimed to synthesize evidence on AI for research capacity strengthening in LMICs towards shifting power in global health. We conducted a systematic review of current evidence on AI for research capacity strengthening and a review of reviews on the decolonization of knowledge generation, searching PubMed, Scopus, and SciELO for relevant literature. Articles were included in the systematic review if they included primary data on using AI for research purposes. Reviews were included in the review of reviews if they addressed issues related to knowledge generation. Each review was assigned two independent reviewers for title and abstract screening, full-text review, and data extraction. A narrative synthesis of the extracted data from both reviews was then performed. Given study designs for the inclusion-eligible papers, we did not conduct a formal risk-of-bias assessment. The systematic review identified 305 papers, of which 8 met the inclusion criteria. The review of reviews identified 14 papers, of which 8 were included in the final analysis. Key themes identified from the systematic review include data analysis and research productivity, literature reviews and knowledge management, training and capacity strengthening, expanding access to methodological support, and writing support. The review of reviews found a recurrent theme in the need to address power imbalances rooted in colonial legacies. These reviews demonstrate the potential for AI to transform research capacity in LMICs by democratizing access to advanced analytical tools, providing methodological support, and helping overcome resource limitations that have historically restricted research opportunities. However, equitable governance and local leadership are crucial to prevent AI from widening the gap between LMICs and HICs, perpetuating the power asymmetries that current efforts seek to dismantle.
Wahl et al. (Thu,) studied this question.