What the models cannot see.
A model can sharpen our view of reality. It cannot become reality.
My work starts with a practical question: how can evidence help a person decide, without hiding the uncertainty that comes with it?
What statistics taught me
Statistical methods are useful precisely because they make their limitations discussable. An estimate depends on observations, assumptions, a model, and the conditions under which it was validated. A forecast can be useful while remaining uncertain. An apparent signal can disappear when the data, population, or question changes.
I learned this working with public-health surveillance, where an alert does not decide what caused a pattern, or what should happen next. Those judgments require context and expertise. I try to make that distinction visible in the systems I build.
What philosophy keeps reminding me
Data are projections of reality. Models are projections shaped by data and assumptions. Reality is larger than both.
Daoist thought reminds me that naming and describing something does not exhaust it: 道可道，非常道。 Buddhist thought provides another useful reminder—to avoid confusing a convincing appearance with a complete account of what is real.
I use these as intellectual influences, not as claims that classical texts were about modern statistics or AI. They encourage a habit I value in scientific work: hold an explanation firmly enough to test it, but loosely enough to revise it.
What AI changes
AI makes it easier to work with evidence that does not fit into a dataset: papers, guidelines, reviews, documents, code, and competing interpretations. It also makes it easier to produce plausible answers without showing where they came from.
That is why I care about provenance, calibration, validation, abstention, privacy, and meaningful human review. These are not decorations around an AI system; they are part of what makes its output useful.
What I build toward
My aim is not to replace judgment with automation. I want to help people see a decision more clearly: what the evidence supports, what remains unknown, which consequences matter, when the system should stop, and who is accountable for the next step.
Observe carefully. Model humbly. Decide consciously.