Cognotik | EthicalReasoningTask

⚙️ TaskConfig.json

{
  "task_type": "EthicalReasoning",
  "ethical_dilemma": "Should we deploy an unvetted AI safety patch to prevent a theoretical breach?",
  "input_files": ["docs/security_policy.md", "logs/threat_model.json"],
  "stakeholders": [
    "End Users",
    "Security Team",
    "Company Shareholders"
  ],
  "ethical_frameworks": [
    "utilitarianism",
    "deontology"
  ],
  "context": "The breach is estimated at 5% probability but catastrophic impact."
}

→

👁️ Session UI Output

Overview

Context

Dilemma Analysis

Synthesis

Synthesis & Recommendation

Conflict: Utilitarianism favors deployment to minimize aggregate risk, while Deontology warns against violating testing protocols.

Final Recommendation: Perform a 'Canary' deployment to 1% of users. This balances the duty of care (Deontology) with risk mitigation (Utilitarianism).

📥 Download: Report.pdf | Report.md

Live Results Showcase

Explore actual artifacts generated by the EthicalReasoningTask, including markdown reports and analysis transcripts.

Execution Configuration

Field	Type	Description
`ethical_dilemma`*	`String`	A clear description of the ethical problem or decision to be made.
`input_files`	`List<String>`	Optional glob patterns for files providing context for the analysis.
`stakeholders`*	`List<String>`	Individuals, groups, or entities affected by the decision.
`ethical_frameworks`	`List<String>`	Frameworks to apply. Options: `utilitarianism`, `deontology`, `virtue_ethics`, `care_ethics`, `rights_based`.
`context`	`String`	Optional background information or constraints.

* Required fields

Task Process Lifecycle

Initialization: Validates configuration and ensures dilemma and stakeholders are defined.
Context Loading: Aggregates context from previous tasks and specified input files.
Dilemma Analysis: An expert agent deconstructs the core conflict and maps stakeholder interests.
Framework Application: Sequential analysis of the dilemma through each selected ethical lens (Utilitarianism, Deontology, etc.).
Synthesis: A "Master Ethicist" agent compares framework recommendations and identifies critical trade-offs.
Reporting: Generates a comprehensive multi-format report (Markdown, HTML, PDF).

Kotlin Boilerplate

// Direct Task Execution
val task = EthicalReasoningTask(
    orchestrationConfig = config,
    planTask = EthicalReasoningTaskExecutionConfigData(
        ethical_dilemma = "Use of user data for training",
        stakeholders = listOf("Users", "Developers"),
        ethical_frameworks = listOf("utilitarianism", "rights_based")
    )
)

Embedded Execution (Headless)

Invoke via UnifiedHarness for CI/CD or automated pipelines:

val harness = UnifiedHarness(serverless = true, ...)
harness.start()
harness.runTask(
    taskType = EthicalReasoningTask.EthicalReasoning,
    typeConfig = TaskTypeConfig(),
    executionConfig = EthicalReasoningTaskExecutionConfigData(
        ethical_dilemma = "Automated deployment of breaking changes",
        stakeholders = listOf("Customers", "DevOps Team"),
        ethical_frameworks = listOf("utilitarianism", "deontology")
    ),
    workspace = File("./workspace"),
    autoFix = true
)

Prompt Segment

The following logic is injected into the LLM context:

EthicalReasoning - Analyze a dilemma through multiple ethical frameworks
  ** Optionally specify input files (supports glob patterns) to provide context
  ** Files will be read and included in the analysis
  ** Specify the ethical dilemma and stakeholders
  ** Provides analysis from each framework's perspective
  ** Synthesizes findings into a balanced recommendation
 ** Highlights ethical trade-offs and points of conflict
  ** Useful for:
     - AI safety and alignment
     - Product and policy ethics
     - Corporate governance