Scott Alexander, curated
← Back to curation

Preliminary Thoughts On The Midjourney Scanner

Quality
74
Strong
Claude Shift
48
Moderate
RWI
2
of 10

Summary

A skeptical, domain-reasoned take on Midjourney's pivot to a whole-body ultrasound-tomography scanner (a water tank ringed by ultrasound + AI reconstruction). Scott (flagging he's a psychiatrist, not a radiologist) argues it can't replace MRI/CT (ultrasound can't penetrate bone or air -> misses brain, lungs, bowels) nor bedside ultrasound (you can't lower a frail 90-year-old into a tank), and the one real gap (dense-breast tomography) is already served by existing devices. Its actual play is whole-body SCREENING for healthy people -- which the medical consensus opposes because of the false-positive problem ('a zit but inside your organs'). Reframes the whole thing as a BET that future AI will benefit ultrasound tomography IN PARTICULAR (cheaper sensors, AI radiology beating incidentaloma-vs-tumor, full-waveform-inversion through bone), plus the radiologist Twitter responses (earlier detection != better outcomes) and the skin/pimple intuition pump.

Why this score

Quality 74 · Strong. Strong (upper): a lucid, genuinely clarifying analysis that dismantles the SF-hype and reframes it into a specific testable bet, carrying the durable screening-false-positive discussion; strong but topical to one product announcement.

Claude’s paradigm shift 48 · Moderate. Moderate: mostly applies existing radiology/screening knowledge to a new product; the 'AI-must-benefit-ultrasound-in-particular' framing is the freshest element.

Real-world impact 2 · Minor. Minor: commentary on a medical-tech product; no material footprint of its own.