Skip to content

SRC01-E01 — Elicit: Structured But Not Analytically Rigorous

Research R0049 — Landscape Scan
Run 2026-03-31-02
Query Q003
Source SRC01
Evidence E01

Description

Elicit implements the most structured research workflow among AI tools but lacks the five queried analytical rigor dimensions.

URL

https://elicit.com/

Extract

Elicit's Systematic Review workflow delivers human-level accuracy in evidence synthesis. Features: structured table-based data extraction (custom columns: "Intervention," "Outcome measures," "Sample size"); semantic search across 138M+ papers; 99.4% extraction accuracy (1,502/1,511 data points correct). Limitation: 39.5% average sensitivity in search — cannot serve as sole search tool. Cannot extract from figures.

Relevance to Hypotheses

Hypothesis Relevance Strength
H1 — Comprehensive frameworks exist Contradicts (structured but not analytically rigorous) Strong
H2 — No structured features Contradicts (highly structured extraction) Strong
H3 — Partial features exist Supports Strong

Feature checklist against queried dimensions:

Dimension Present? Notes
Calibrated probability language No No probability calibration
Formal bias assessment No No risk-of-bias scoring
Competing hypotheses No No ACH or equivalent
Search transparency logging Partial Shows papers found but not complete search log
Self-audit mechanisms No No methodological self-check

Context

FACT: Elicit is the most structured AI research tool evaluated. JUDGMENT: Its strength is in data organization and extraction, not in analytical reasoning or evidence quality assessment.