The Epistemic Turn
The longest sustained argument in the main series. What AI systems cannot see, what systems designed to see it would need to be, why the research infrastructure prevents the integration, and what research itself looks like when it stops decomposing what should not be decomposed. Six essays closing the diagnostic movement.
TAM_074
The Interrogator
What If the Most Important AI Is the One That Questions the Question?
The man with the notebook keeps a record not of wrong answers but of questions no system asks. He has scanned hundreds of AI job postings. None is for the role he describes: the …
TAM_075
The Epistemic Framework
A Design Specification for AI Systems That Interrogate Rather Than Optimize
Every optimizer has a blind spot defined by its objective function. This is the design specification for the system that makes the blind spot visible — not a better optimizer, but …
TAM_076
The Amplitude Problem
When AI Makes Everything Louder but Nothing Clearer
Priya has forty-seven papers open about maternal health interventions in districts like hers. She is not struggling with misinformation — what fills her screen is worse. …
TAM_077
The Injected Center
How Manufactured Consensus Becomes the Reasonable Position
Victor's student cited fourteen sources for an extreme monetary position. Three AI tools called it a legitimate minority view. The sources were real. The network was manufactured. …
TAM_078
The Missing Model
Why We Cannot Simulate What We Most Need to Understand
Ananya has the fiscal model, the nutritional model, the political model, the logistics model. She does not have a model for what happens to the neighborhoods where the ration shop …
TAM_079
The Reimagined Study
What Research Looks Like When You Stop Decomposing What Should Not Be Decomposed
Kavitha's third study improved hemoglobin levels by a statistically significant margin. The mechanisms that determined whether the finding mattered for these women's pregnancies …