Risk LabStrong architectural inferencev1.22.1

In plain English

This page provides local browser worksheets. They help plan reviews; they are not formal safety certifications.

  • Why this matters: AI risk can come from the whole arrangement, not one obvious model.
  • What to look for: data, memory, routes, adapters, tools, evaluators, updates, and rollback paths.
  • Technical version below: the expert terminology remains available and is linked through the glossary.

ModelBreeder Ecology Dashboard Review

Evidence levelStrong architectural inferenceTechnical label: Architectural inference

This worksheet checks whether a ModelBreeder-style dashboard helps reviewers see the system or merely hides risk behind a polished interface.

Review checklist

QuestionGood signalConcern
Does every candidate have a Genome record?Base, adapters, seed, A record of where a component or behavior came from. Open glossary definition, and mutation log are visible.Candidate appears as only a friendly name or score.
Does every candidate have a FitnessVector?Utility, cost, latency, memory, novelty, and A system that judges whether an AI output or candidate is acceptable. Open glossary definition disagreement are shown.A single composite score hides tradeoffs.
Is The decision not to change the system. Open glossary definition visible?The UI shows reject, archive, hold, retire, and no-op as normal outcomes.The UI pushes every candidate toward promotion.
Is novelty bounded?Novelty is tied to useful behavior descriptors and an archive.Weird or untested behavior is rewarded as inherently valuable.
Is Returning a system to an earlier known state. Open glossary definition complete?Rollback packet includes model, adapters, prompts, memory, router, evaluator, and deployment alias.Rollback means only replacing a model file.
Are source links visible?Preserved summaries and external source leads are linked.The dashboard makes claims without source boundaries.

Local worksheet use

Use this page before adding charts to the public Risk Lab or a future Evolution Lab. The goal is not to maximize animation. The goal is to make lineages, evidence, and release decisions legible.

v1.22.0 risk-side boundary

Evidence levelStrong architectural inferenceTechnical label: Strong architectural inference

This page preserves possibility-side source material. The A behavior pattern that can survive, move, or reappear across a changing AI system. Open glossary definition use is narrower: translate controlled-evolution vocabulary into risk review. For that translation, use ModelBreeder Risk Side, ModelBreeder Risk Translation, and Model Breeding Risk Boundaries.