AI engineering research
Probabilistic Systems Engineering
Gregory Tomlinson · About · Research
An evolving research corpus on AI-assisted software work: omitted scope, blended inference, default fill-in, VDG, preserved truth, and where authority actually does and does not live.
Current anchors
The live center is failure visibility in one response or across change, plus the narrower research boundary that survived after the stronger contract-authority claim failed.
Why AI Prompting Fails v1.2
Rename a field. Tighten a validation rule. Update a serializer. Adjust an error path. Generate a test. Ask for one bounded change in one visible place and the result can feel almost unnervingly effective.
PaperWhen AI Collapses Fact and Assumption v1.4.1
Blended inference is the baseline response mode of LLMs. Smooth prose is the goal. In software, that smoothness can hide the boundary between grounded analysis and inferred assumptions.
PaperContract Authority Under AI v0.7.3
The strongest contract-authority claim, the research boundary, and the narrower result that survived.
Protocols & Contracts
Explicit source-of-truth artifacts and response-structure documents that support the papers without pretending to be the final live authority surface.