Oct 6, 2026 · 32 min · 12 segments
The Fifth Universal Definition of Myocardial Infarction (MI), better known colloquially as a heart attack, has replaced the previous Type 1-5 clinical classification with primary, secondary, and…
And so today I'm not going to talk about AI killing everyone, but I am going to talk about when AI can get something wrong, can make everyone wrong in the same way.
So if you ask five people to analyze a hospital problem, any agreement can be reassuring.
As five people use the same artificial intelligence model, the same source material and the same assumption, that agreement may mean something different.
Users of Epic and other hospital systems may frame their questions around what they can retrieve.
AI can extend that boundary, but can also make the boundary harder to see.
So EPIC has substantial reporting capabilities, such as its slicer, dicer tools, self-reporting exploration.
EPIC describes Sidekiq as an AI assistant that builds reports for employing language questions.
UC Davis Health also describes reporting a reporting workbench and the clarity and the caboodle data environments with database access tied to roles and qualifications.
A conventional report follows defined query logic and a generative model produces responses that can vary.
Both can nonetheless restrict an investigation through the information and assumptions made available to the user.
The National Institute of Standards and Technology identifies systemic, computational, statistical, and human cognitive sources of AI bias.
Training data and the choice of prediction target and the way people interpret results all matter.
And so today I'm not going to talk about AI killing everyone, but I am going to talk about when AI can get something wrong, can make everyone wrong in the same way.
So if you ask five people to analyze a hospital problem, any agreement can be reassuring.
As five people use the same artificial intelligence model, the same source material and the same assumption, that agreement may mean something different.
Users of Epic and other hospital systems may frame their questions around what they can retrieve.
AI can extend that boundary, but can also make the boundary harder to see.
So EPIC has substantial reporting capabilities, such as its slicer, dicer tools, self-reporting exploration.
EPIC describes Sidekiq as an AI assistant that builds reports for employing language questions.
UC Davis Health also describes reporting a reporting workbench and the clarity and the caboodle data environments with database access tied to roles and qualifications.
A conventional report follows defined query logic and a generative model produces responses that can vary.
Both can nonetheless restrict an investigation through the information and assumptions made available to the user.
The National Institute of Standards and Technology identifies systemic, computational, statistical, and human cognitive sources of AI bias.
Training data and the choice of prediction target and the way people interpret results all matter.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.
