Node & Norm develops methods, structured records, and reference implementations for examining AI-assisted decisions. The technical question is whether a reviewer can reconstruct the evidence, authority, intervention, and resulting action from the preserved record.
The website develops these questions through essays and narratives. This GitHub organization holds the specifications, assessment rules, source records, code, tests, and version histories needed to examine the work.
| Project | Technical materials to inspect | Research boundary |
|---|---|---|
| Trust, Autonomy, and Evidence (TAE) | Evidence architecture, claim-to-source records, deterministic assessment logic, historical cases, and mutation controls. Start with the reproducibility guide. | Working research. Internal checks and selected historical cases leave independent reliability and contemporary transfer open. |
| Human Influence Telemetry (HIT) | Documentary assessment contract, scoring engine, conformance fixtures, reviewer materials, and claim-evidence gates. Read the research status. | Research in progress. Contract stability, internal conformance, and external-rater replication are separate claims; current-contract replication remains unresolved. |
| Governed Decision Intelligence (GDI) | Governed Decision Record schema, gate classification, worked examples, and reference implementation. Start with validation and reproducibility. | Open specification and reference implementation. Independent field validation and external conformance acceptance remain open. |
| Catholic Doctrinal Fidelity Index (CDFI) | Domain-specific evaluation governance, scoring configuration, cap rules, judge-reliability protocols, and literature traceability. | Evolving. Scoring-contract reconciliation remains open. SAICRED retains its separate project leadership and publication authority. |
| Regulated Gate Decision Support (RGDS) | Decision-log schema, illustrative records, structural validation, and selected semantic checks for regulated phase gates. | Reference implementation. Reconstruction speed, field effectiveness, and regulatory outcomes remain unestablished. |
The RGDS independent study provides the accompanying exploratory account; its repository preserves the study materials.
- Read the project's research-status and limitations statements before interpreting its outputs.
- Select a tagged release or commit. Record the manuscript, schema, software, and evidence-packet versions separately where the project distinguishes them.
- Follow that repository's installation and validation instructions. Retain the command, environment, inputs, and output needed to reproduce a finding.
- Inspect negative controls, excluded material, missing evidence, and open exceptions alongside successful checks.
- Cite the exact artifact and version. Report a discrepancy through the project's issue or contribution process, with a minimal reproducible example and the affected claim.
Passing software checks establishes the behavior tested under the specified inputs. Claims about source accuracy, reviewer reliability, institutional adoption, or practical benefit require evidence suited to those questions.
AI Assistance Governance develops a working profile for bounded AI participation, assigned human authority, and dependency assessment. Applied AI Research Translator provides a governed workflow from research inputs to traceable claims, bounded tasks, and human-reviewed decision records.
Each repository maintains its own methods, releases, attribution, license, and correction process. The organization directory is a route into those records; project-level status statements define the scope of their claims.
Researcher: Mark Julius Banasihan.
TAE remains research in progress. The manuscript From Formal Authority to Practical Human Control is in progress with arXiv.org; this states the author's publication plans and does not claim arXiv acceptance. Current archived paper and software versions are linked from the TAE repository.
The About page describes the research purpose and intellectual influences, including Daniel Kahneman's work on judgment and Stuart Russell's work on human control. Those influences frame the questions; the repositories carry the methods and evidence used to examine them.
Essays · Research overview · Research integrity · Brand assets