Taiwan’s Taoyuan launches AI lung cancer screening

The author argues that while governments and companies successfully use AI for narrow pattern recognition and immediate data triage, relying on predictive algorithms for high-stakes, life-changing decisions presents severe ethical and technical risks. Predictive models often fail due to missing context, susceptibility to gaming, overreliance by human operators, and inappropriate training data. The author concludes that individual-level predictions should remain advisory and contestable rather than being automated away by unexplainable black-box systems. Open article →