Martin Varsavsky cites AI paperwork relief

Healthcare AI keeps getting sold as a doctor substitute. But the thing actually working right now is much less dramatic. It is software that listens, drafts notes, cleans up documentation, and gives clinicians some of their day back. That is the point Martin Varsavsky was making when he pointed to JAMA research and said the real value is paperwork relief, not synthetic bedside empathy. (jamanetwork.com) ### What is the claim here? Basically, the claim is that medicine’s biggest AI bottleneck is not diagnosis alone. It is administration. Clinicians spend huge chunks of the day inside the electronic health record, writing notes, clicking boxes, and finishing charts after hours. If AI can take even part of that load away, the payoff lands immediately — more face time, less clerical drag, and maybe a little less burnout. (jamanetwork.com) ### Why paperwork first? Because this is the low-risk, high-friction part of care. An ambient scribe can draft a note from a conversation. A documentation assistant can turn a rough summary into billing-ready language. A scheduling or intake tool can move forms and messages around without pretending to be a physician. That is a much easier problem than asking AI to weigh symptoms, uncertainty, tradeoffs, and emotion the way a good clinician does. (jamanetwork.com) ### What does the JAMA evidence actually show? The signal is real, but it is not magic. One JAMA Network Open study on an ambient documentation platform found better clinician satisfaction and lower cognitive load after adoption. Another study found AI scribe users spent less time in the EHR and less time writing notes per appointment than matched nonusers. A JAM(jamanetwork.com)I tool in healthcare. (jamanetwork.com) ### So does this free up time with patients? That is the hope — and sometimes yes. If a clinician is not staring at the screen trying to reconstruct the visit, that attention can shift back to the person in the room. Some of the early reporting around these studies highlights exactly that: lower cognitive burden, less after-hours charting, and a better ability to(jamanetwork.com)ates room for it. (uchicagomedicine.org) ### Why not push straight to autonomous clinical AI? Because the hard part of medicine is not just pattern recognition. It is judgment under uncertainty, responsibility, trust, and communication. Varsavsky himself is backing a much more ambitious company — Certuma, which wants AI to dia(uchicagomedicine.org)odes are more manageable than a bad diagnosis. (forbes.com) ### Is there a catch? Yes — several. Time savings look modest in some studies, and after-hours work does not always fall as much as people hope. AI scribes also cost money, usually a few hundred dollars per clinician per month, and there is a live debate over whether better documentation can also mean more aggressive coding and higher spending. So this is not a pure free lunch. It is workflow software with tradeoffs. (hitconsultant.net) ### Why does this matter beyond one post? Because it points to a more realistic roadmap for healthcare AI. The flashy vision says AI will become the doctor. The practical vision says AI will clear the underbrush first — notes, inboxes, intake, coding, discharge paperwork, scheduling. Turns out that may be the version hospitals can adopt fastest and clinicians will thank them for first. (jamanetwork.com) ### Bottom line? The near-term winner in healthcare AI may be the least glamorous one. Not a machine with perfect bedside manner. Just a tool that kills the clicks, drafts the note, and gives the human clinician more time to actually care.