According to Andrej Karpathy, referencing a detailed analysis from The Works in Progress Newsletter, the expectation that rapid advances in image recognition AI would eliminate radiology jobs has not materialized (source: Karpathy on X, 2025; worksinprogress.news). Despite predictions from leading AI figures like Geoff Hinton nearly a decade ago, radiology as a field is expanding, not contracting. The article highlights several reasons: current AI benchmarks do not comprehensively reflect real-world scenarios; the radiologist’s role is multifaceted, extending well beyond image recognition; and significant deployment barriers exist, including regulatory, insurance, and institutional hurdles. Furthermore, Karpathy cites the Jevons paradox—AI tools may increase efficiency, but also drive up demand for radiology services. For AI industry stakeholders, this underscores that practical AI adoption in healthcare is complex, with opportunities lying more in augmenting professionals rather than replacing them. The trend suggests that AI will act as a productivity tool, requiring businesses to focus on workflow integration, compliance, and support services rather than direct job replacement.