myofficehours.ai

doc

Social Science Computer Review
Social Science Computer Review

Can AI Replace Your Second Reviewer? A Proof of Concept

Hilkenmeier and colleagues test Elicit against human reviewers on 602 data points from a real systematic review, reporting 81.4 percent accuracy against 86.7 percent for humans, and propose a hybrid extraction workflow. You get a defensible, evidence-based protocol for where AI can and cannot substitute for a second human reviewer.

Task
Research & literature
Time
1.5 hr
Level
advanced
Cost
Free tier — There is a free tier that covers this, with paid limits above it.
Tools
Elicit
Published
2025-12-03