The revolutionary potential of Artificial Intelligence (AI) has infiltrated nearly every facet of contemporary life, none more so than the healthcare sector. Researchers have progressively turned to AI to improve the early detection, diagnosis and treatment of cancer. A prime example of AI's promising developments in cancer management is the work at the Massachusetts Institute of Technology (MIT) Jameel Clinic and Mass General Cancer Centre. Their collaborative team, led by MIT Jameel Clinic AI faculty lead Regina Barzilay, developed an AI tool known as Sybil, which was trained on low-dose chest computed tomography scans. It accurately predicted the risk of patients developing lung cancer within six years.
Despite the encouraging developments, experts warn of significant risks and biases linked to AI's use in cancer detection and treatment. Overdiagnosis is a significant concern, potentially leading to unnecessary treatments and patient anxiety. Experts argue that the purpose of AI cancer detection is to identify cancers that can potentially cause death, not merely to find more cancers. Even more worrying is the evidence of racial biases in AI-driven diagnosis. Research has shown that AI systems risk being less accurate for people with darker skin. While the use of AI in cancer detection and treatment is promising, there are necessary considerations to ensure its responsible development and deployment, and experts stress the importance of ongoing conversations among clinicians, AI developers and patients about when and how to use AI-assisted diagnosis.