Orca Athalyx Intelligence markOrca AthalyxBack to Othalyx
Working Methodology · v0.1 · 2026

The Othalyx
Five-Lens Model

A structured methodology for examining what an AI-generated recommendation may be missing before it influences consequential human action.

Status Active / Working Model
Abstract

Between output and action.

The Five-Lens Model provides a common sequence for examining an AI-generated recommendation before it influences a consequential human decision. It does not produce a numerical score or substitute for domain expertise. It makes relevant questions visible so a human decision-maker can proceed, adapt, escalate, defer, or stop with greater clarity.

Model Overview

Five connected questions.

  1. 01Trust

    Can this output be relied upon?

  2. 02Context

    What could change the answer?

  3. 03Culture

    Which human context matters?

  4. 04Behavior

    How might the output influence the decision-maker?

  5. 05Consequence

    What happens if someone acts?

Lens Definitions

What each lens examines.

01

Epistemic Reliability

Trust

Can this output be relied upon?

Evaluates
Evidence · Source quality · Freshness · Uncertainty · Traceability · Completeness · Scope
Key questions
  • What evidence supports this?
  • What relevant information may be missing?
  • Does the output sound more certain than the evidence warrants?
  • What should be independently verified?
02

Context

What could change the answer?

Evaluates
Objective · Timing · Constraints · Incentives · History · Relationships · Dependencies · Reversibility
Key questions
  • What circumstances are missing?
  • What assumptions must be true?
  • Could timing or environment change the recommendation?
  • What alternative explanation should be considered?
03

Culture

Which human context matters?

Evaluates
Language · Values · Religion · Social norms · History · Geography · Authority · Representation
Key questions
  • Is cultural context actually material here?
  • Could language or tone change meaning?
  • Is one social norm being treated as universal?
  • Would embedded local expertise change interpretation?

Culture should contextualize judgment, not stereotype individuals.

04

Behavior

How might the output influence the decision-maker?

Evaluates
Automation bias · Anchoring · Authority · Framing · Emotional response · Dependency · Agency · Friction
Key questions
  • Could the user defer merely because the recommendation came from AI?
  • Does the presentation imply more authority than warranted?
  • Are reasonable alternatives visible?
  • Should more friction exist before action?
05

Consequence

What happens if someone acts?

Evaluates
Direct effects · Indirect effects · Stakeholders · Severity · Reversibility · Distribution · Feedback loops · Scale
Key questions
  • Who benefits?
  • Who bears risk?
  • How reversible is the action?
  • What happens if this decision is repeated at scale?
Action States

Judgment remains human.

The model supports a considered action state rather than an automated verdict.

  • Proceed
  • Adapt
  • Escalate
  • Defer
  • Stop
Limitations

A working model, not a universal answer.

The Five-Lens Model does not establish factual truth, predict individual behavior, replace legal or domain-specific review, or guarantee a good outcome. Materiality depends on the decision, available evidence, affected people, and operating environment.

Validation Roadmap

Built to be tested.

  • Apply the model to varied consequential decision scenarios.
  • Examine whether the prompts reveal missing information and hidden assumptions.
  • Test where specific lenses are material, immaterial, or require domain expertise.
  • Refine language, decision states, and practical worksheet use through applied learning.
Download the Evaluation Worksheet All Research