In a new study published in The Lancet Digital Health, researchers from the University of Adelaide have identified major hurdles in the use of artificial intelligence (AI) for medical diagnosis compared to human clinicians. The study highlights the ‘AI chasm’, where the development and commercialization of AI decision-making systems have outpaced our understanding of their value for clinicians and their impact on human decision-making.
Lead researcher Lana Tikhomirov, along with Professor Carolyn Semmler and their team, found that clinicians are contextually motivated and mentally resourceful decision makers, while AI models lack the ability to understand correlations in data and patient context. The researchers suggest that AI should be used more like a clinical drug rather than a device, and that misconceptions about AI technology limit its potential to augment human decision-making in the medical field.
The study underscores the importance of clinicians’ ability to utilize domain-relevant information and cues in the clinical environment, which is crucial for accurate and rapid diagnosis. The researchers emphasize the need for AI models to be able to question their dataset and exhibit epistemic humility, similar to how clinicians critically evaluate the information they are presented with.
Overall, the study sheds light on the challenges in implementing AI technology in clinical settings and highlights the importance of understanding the limitations of AI in order to maximize its potential to enhance medical diagnosis. The researchers advocate for a more nuanced approach to utilizing AI in healthcare, taking into account the unique expertise and decision-making processes of human clinicians.
Source
Photo credit www.newswise.com
