Abstract
Current bioinformatics research has been impacted by the rapid advancements in AI and will create faster or more accurate ways to analyze biological data and provide doctors with personalized medicine. In addition, algorithmic bias is a significant issue because the historical inequalities and structural constraints that exist in the training dataset and model evaluation will lead to using AI in a way that perpetuates discrimination. Complex machine learning models are difficult to understand and interpret, which is a problem that arises from the lack of transparency inherent in their design. This review examines ethical, social and trust issues related to artificial intelligence in bioinformatics, with a focus on the root causes of bias throughout the data/modelling lifecycle and their impact. The review also outlines key concepts related to transparency, explainability and accountability as critical tenets of reliable biomedical AI, and discusses emerging standards for responsible usage. Finally, the review discusses future directions for bioinformatics and includes an overview of the need for causal and explainable methods in bioinformatics, the need for inclusive governance structures, and the need for equitable pathways of global innovation in AI and highlights ethical integrity/trust as enablers of long-term, progressive bioinformatics innovation.