We study how intelligent systems distribute agency, authority, understanding, and responsibility.
Buildcraft Research turns patterns from products and practice into evidence that teams can use. The work is institutional: methods are disclosed, claims are supported, limitations are visible, and findings belong to the Buildcraft research program rather than Carl’s personal essay archive.
A recurring study of the relationship designed between people and AI.
The Human-AI Index evaluates products and systems across the six dimensions of the Buildcraft Model: Agency, Authority, Legibility, Trust, Correction, and Accountability.
It does not rank which product has the most AI. It examines which products create the clearest, most capable, and most responsible relationship between human and machine.
Explore the Human-AI Index
A method before a benchmark.
The first release is in development. No products have been scored or ranked for publication.
Each release will disclose its research question and inclusion criteria, evaluation method and rubric, sample and product versions, dimension-level findings, product examples, limitations and conflicts, and publication and revision dates.
Reports and datasets will be linked when available. Until the method and evidence are ready, there are no index scores to report.
Explore the Buildcraft ModelReports, methods, and patterns for Human-AI Systems Design.
Research reports, Index releases, methods, and patterns will appear here as they are published. Each publication will identify its authors, date, citations, methodology, and limitations.
Research informs our work. Practice gives the research better questions.
Start a project