Changes for page Automation
Last modified by Robert Schaub on 2026/02/08 21:22
From version 2.2
edited by Robert Schaub
on 2025/12/17 18:07
on 2025/12/17 18:07
Change comment:
Update document after refactoring.
Summary
-
Page properties (2 modified, 0 added, 0 removed)
Details
- Page properties
-
- Parent
-
... ... @@ -1,1 +1,1 @@ 1 - FactHarbor.Archive.FactHarbor 0\.9\.40.Specification.WebHome1 +Archive.FactHarbor 0\.9\.40.Specification.WebHome - Content
-
... ... @@ -3,6 +3,7 @@ 3 3 Automation in FactHarbor amplifies human capability while implementing risk-based oversight. 4 4 5 5 This chapter defines: 6 + 6 6 * Risk-based publication model 7 7 * Quality gates for AI-generated content 8 8 * What must remain human-only ... ... @@ -18,14 +18,18 @@ 18 18 19 19 1. **Input**: User pastes a block of raw text. 20 20 1. **Deep Analysis (Background)**: The system autonomously performs the full pipeline **before** displaying the text: 22 + 21 21 * Extraction & Normalisation 22 22 * Scenario & Sub-query generation 23 23 * Evidence retrieval with **contradiction search** 24 24 * Quality gate validation 25 25 * Verdict computation 28 + 26 26 1. **Visualisation (Extraction & Marking)**: The system displays the text with claims extracted and marked. 30 + 27 27 * **Verdict-Based Coloring**: The extraction highlights (e.g. Orange/Green) are chosen **according to the computed verdict** for each claim. 28 28 * **AI-Generated Label**: Clear indication that content is AI-produced 33 + 29 29 1. **Inspection**: User clicks a highlighted claim to see the **Reasoning Trail**, showing exactly which evidence and sub-queries led to that verdict. 30 30 31 31 === 1.2 Technical Scope === ... ... @@ -37,12 +37,12 @@ 37 37 * **No Human Approval Gate**: Demonstrates scalable AI publication 38 38 * **Structured Sub-Queries**: Logic generated by decomposing claims into the FactHarbor data model 39 39 40 - 41 41 == 2. Publication Model == 42 42 43 43 FactHarbor implements a risk-based publication model with three modes: 44 44 45 45 === 2.1 Mode 1: Draft-Only === 50 + 46 46 **Mode 1 (Draft-Only)**: Failed quality gates or high-risk content pending expert review. Internal review queue only. 47 47 48 48 See [[AKEL Publication Modes>>FactHarbor.Specification.AI Knowledge Extraction Layer (AKEL).WebHome#HPublicationModes]] for detailed mode specifications. ... ... @@ -49,6 +49,7 @@ 49 49 50 50 51 51 === 2.2 Mode 2: AI-Generated (Public) === 57 + 52 52 **Mode 2 (AI-Generated, Published)**: Passed all quality gates, risk tier B or C, clearly labeled as AI-generated. Users can request human review. 53 53 54 54 See [[AKEL Publication Modes>>FactHarbor.Specification.AI Knowledge Extraction Layer (AKEL).WebHome#HPublicationModes]] for detailed requirements. ... ... @@ -55,6 +55,7 @@ 55 55 56 56 57 57 === 2.3 Mode 3: Human-Reviewed === 64 + 58 58 **Mode 3 (Human-Reviewed, Published)**: Validated by human reviewers or experts, highest trust level. Required for Tier A content publication. 59 59 60 60 See [[AKEL Publication Modes>>FactHarbor.Specification.AI Knowledge Extraction Layer (AKEL).WebHome#HPublicationModes]] for detailed requirements. ... ... @@ -64,7 +64,9 @@ 64 64 65 65 66 66 Risk tiers determine review requirements and automation levels. See [[Governance>>FactHarbor.Organisation.Governance.WebHome#HRiskTierPolicyGovernance]] for tier policy governance. 74 + 67 67 === 3.1 Tier A (High Risk) === 76 + 68 68 * **Domains**: Medical, legal, elections, safety, security 69 69 * **Automation**: AI can draft, human review required for "Human-Reviewed" status 70 70 * **AI publication**: Allowed with prominent disclaimers and warnings ... ... @@ -71,6 +71,7 @@ 71 71 * **Audit rate**: Recommendation: 30-50% 72 72 73 73 === 3.2 Tier B (Medium Risk) === 83 + 74 74 * **Domains**: Complex policy, science, causality claims 75 75 * **Automation**: AI can draft and publish (Mode 2) 76 76 * **Human review**: Optional, audit-based ... ... @@ -77,12 +77,12 @@ 77 77 * **Audit rate**: Recommendation: 10-20% 78 78 79 79 === 3.3 Tier C (Low Risk) === 90 + 80 80 * **Domains**: Definitions, established facts, historical data 81 81 * **Automation**: AI publication default 82 82 * **Human review**: On request or via sampling 83 83 * **Audit rate**: Recommendation: 5-10% 84 84 85 - 86 86 == 4. Human-Only Tasks == 87 87 88 88 These require human judgment and cannot be automated: ... ... @@ -94,7 +94,6 @@ 94 94 * **Audit system oversight** and quality standard definition 95 95 * **Risk tier policy** adjustments based on societal context 96 96 97 - 98 98 == 5. AI-Draft with Audit (Semi-Automated) == 99 99 100 100 AKEL drafts these; humans validate via sampling audits: ... ... @@ -109,12 +109,12 @@ 109 109 * **Public summaries** and accessibility text 110 110 111 111 Most Tier B and C content remains in AI-draft status unless: 121 + 112 112 * Users request human review 113 113 * Audits identify errors 114 114 * High engagement triggers review 115 115 * Community flags issues 116 116 117 - 118 118 == 6. Fully Automated Structural Tasks == 119 119 120 120 These require no human interpretation: ... ... @@ -128,7 +128,6 @@ 128 128 * **Layout generation** (diagrams, summaries, UI presentation) 129 129 * **Federation integrity checks** (cross-node data validation) 130 130 131 - 132 132 == 7. Quality Gates (Automated) == 133 133 134 134 ... ... @@ -147,6 +147,7 @@ 147 147 Instead of reviewing all AI output, systematic sampling audits ensure quality: 148 148 149 149 === 8.1 Stratified Sampling === 158 + 150 150 * Risk tier (A > B > C sampling rates) 151 151 * Confidence scores (low confidence → more audits) 152 152 * Traffic/engagement (popular content audited more) ... ... @@ -154,7 +154,9 @@ 154 154 * User flags and disagreement signals 155 155 156 156 === 8.2 Continuous Improvement Loop === 166 + 157 157 Audit findings improve: 168 + 158 158 * Query templates 159 159 * Source reliability weights 160 160 * Contradiction detection algorithms ... ... @@ -162,11 +162,11 @@ 162 162 * Bubble detection heuristics 163 163 164 164 === 8.3 Transparency === 176 + 165 165 * Audit statistics published 166 166 * Accuracy rates by tier reported 167 167 * System improvements documented 168 168 169 - 170 170 == 9. Automation Roadmap == 171 171 172 172 Automation capabilities increase with system maturity while maintaining quality oversight. ... ... @@ -174,6 +174,7 @@ 174 174 === 9.1 POC (Current Focus) === 175 175 176 176 **Automated:** 188 + 177 177 * Claim normalization 178 178 * Scenario template generation 179 179 * Evidence metadata extraction ... ... @@ -183,6 +183,7 @@ 183 183 * **Risk tier assignment** 184 184 185 185 **Human:** 198 + 186 186 * High-risk content validation (Tier A) 187 187 * Sampling audits across all tiers 188 188 * Quality standard refinement ... ... @@ -191,6 +191,7 @@ 191 191 === 9.2 Beta 0 (Enhanced Automation) === 192 192 193 193 **Automated:** 207 + 194 194 * Detailed scenario generation 195 195 * Advanced evidence reliability scoring 196 196 * Cross-scenario comparisons ... ... @@ -199,6 +199,7 @@ 199 199 * **Increased AI-draft coverage** (more Tier B content) 200 200 201 201 **Human:** 216 + 202 202 * Tier A final approval 203 203 * Audit sampling (continued) 204 204 * Expert validation of complex domains ... ... @@ -207,6 +207,7 @@ 207 207 === 9.3 Release 1.0 (High Automation) === 208 208 209 209 **Automated:** 225 + 210 210 * Full scenario generation (comprehensive) 211 211 * Bayesian verdict scoring across scenarios 212 212 * Multi-scenario summary generation ... ... @@ -215,12 +215,12 @@ 215 215 * **Most Tier B and all Tier C** auto-published 216 216 217 217 **Human:** 234 + 218 218 * Tier A oversight (still required) 219 219 * Strategic audits (lower sampling rates, higher value) 220 220 * Ethical decisions and policy 221 221 * Conflict resolution 222 222 223 - 224 224 == 10. Automation Levels Diagram == 225 225 226 226 {{include reference="FactHarbor.Specification.Diagrams.Automation Level.WebHome"/}} ... ... @@ -241,5 +241,4 @@ 241 241 * [[AKEL (AI Knowledge Extraction Layer)>>FactHarbor.Specification.AI Knowledge Extraction Layer (AKEL).WebHome]] 242 242 * [[Requirements (Roles)>>FactHarbor.Specification.Requirements.WebHome]] 243 243 * [[Workflows>>FactHarbor.Specification.Workflows.WebHome]] 244 -* [[Governance>>FactHarbor.Organisation.Governance]] 245 - 260 +* [[Governance>>Archive.FactHarbor.Organisation.Governance.WebHome]]