Raw LLM Responses

Inspect the exact model output for any coded comment.

Comment
I skimmed the actual paper in Nature, and it seems pretty legit. That being said, as a radiologist that currently uses commercially available "AI" assisted software (NeuroQuant, RAPID and VIZ.AI), this kind of stuff is often way less useful out in the real world where you are dealing with subpar scanners, artifacts, technologists, etc. Right now, computers are a lot better than humans at estimating volumes of things and finding small abnormalities in large data sets (i.e. small nodule in the lung or breast), but they are really bad at common sense decisions like obvious artifact. Viz.ai in particular has an unacceptable number of false positives for large vessel occlusions in the real world despite many papers saying that it has a low false positive rate in a controlled environment.
reddit Cross-Cultural 1577928203.0 ♥ 74
Coding Result
DimensionValue
Responsibilitynone
Reasoningconsequentialist
Policynone
Emotionindifference
Coded at2026-04-25T08:33:43.502452
Raw LLM Response
[{"id":"rdc_fcst7ck","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"indifference"},{"id":"rdc_fcsyd2f","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"indifference"},{"id":"rdc_fctc77k","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"indifference"},{"id":"rdc_fcs99w9","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"approval"},{"id":"rdc_fcss1wq","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"indifference"}]