R0049/2026-03-31/Q003-H2¶
Statement¶
No AI-assisted research tools implement any of the five target features (calibrated probability language, formal bias assessment, competing hypotheses, search transparency logging, self-audit mechanisms).
Status¶
Eliminated. Several tools implement individual features: scite implements citation-context analysis (a form of evidence quality assessment), Open Synthesis implements ACH (competing hypotheses), Microsoft Copilot Critique implements a form of cross-model audit, and deep research tools provide partial search transparency.
Supporting Evidence¶
| Evidence | Summary |
|---|---|
| SRC01-E01 | PaperQA2 lacks all five target features (supports H2 for this specific tool) |
| SRC02-E01 | STORM lacks all five target features (supports H2 for this specific tool) |
Contradicting Evidence¶
| Evidence | Summary |
|---|---|
| SRC04-E01 | Scite implements Smart Citations (supporting/contrasting/mentioning) — a form of evidence quality assessment |
| SRC05-E01 | Microsoft Copilot Critique uses cross-model verification — a form of audit mechanism |
| SRC06-E01 | Open Synthesis implements ACH for competing hypotheses analysis |
Reasoning¶
While no single tool implements a comprehensive framework, individual features exist in isolation across different tools. The five target features map onto the tools landscape as follows: scite touches evidence quality, Open Synthesis touches competing hypotheses, and Microsoft Critique touches audit mechanisms. Calibrated probability language and search transparency logging have the weakest representation.
Relationship to Other Hypotheses¶
Eliminated by evidence supporting H3. The landscape contains scattered individual features, not comprehensive frameworks (supporting H3) and not zero implementation (eliminating H2).