- Essays··8 min read
The Compute Bill Sitting Under J-MID
Japan and Korea are deploying AI across every teaching-hospital radiology workflow — yet neither ministry nor hospital consortium has published the electricity budget for the inference stack. Continuous report drafting, MRI classification and admission triage draw power differently from the scanners they augment, and the gap between an H100 rack and a Furiosa chip is a hospital-substation decision. The operating energy assumption under each country's AI hospital programme remains unwritten.
radiology-aimedical-imagingcompute-costshealthcare-aiRead - Essays··8 min read
Your Scan Is Done, Your Report Is Not
England's diagnostic imaging backlog — 1.8 million people waiting, a radiologist shortfall projected to reach 39% by 2029 — is now kept moving by AI triage systems whose error rates and training data do not appear on the consent form you signed. Liability for the countersigned report remains with the clinician, not the vendor. The capital is moving faster than any framework that would tell you whose risk you are sitting inside.
nhsmedical-imagingai-healthcarepatient-rightsRead