Elicit

Learning Objectives

• Understand what Elicit does and how it differs from Consensus and standard academic search
• Identify Elicit's key features: automated paper finding, column-based data extraction, and literature review tables
• Evaluate when Elicit is the right tool for academic research workflows

What Is Elicit?

Elicit is an AI research assistant built by Ought, a research organization focused on AI-assisted reasoning. Launched in 2022 and continuously updated since, Elicit is designed specifically for academic researchers, graduate students, evidence synthesis professionals, and systematic reviewers who need to work through large bodies of scientific literature.

While Consensus focuses on providing a clear empirical answer to a question by synthesizing the consensus, Elicit takes a different approach: it helps researchers build structured tables from paper data, automatically extracting specific fields from each paper — population, intervention, outcomes, sample size, statistical significance — into a browsable comparison table that dramatically accelerates literature reviews.

How Elicit Works

Paper Discovery

Elicit searches its database of over 125 million academic papers (primarily from Semantic Scholar) to find papers relevant to your research question:

• Enter a research question in natural language
• Elicit finds 4–8 most relevant papers by default (expandable to 100+)
• Papers are ranked by relevance and quality signals
• Results include full citation, abstract, year, and journal

Column-Based Data Extraction

The central Elicit feature: define columns of data you want extracted from each paper, and Elicit reads each paper and populates a table automatically.

Standard pre-built columns include:

• Study type (RCT, observational, meta-analysis, case study)
• Sample size
• Intervention
• Outcome
• Main finding
• Limitations
• Statistical significance

Custom columns (Plus): define any field you want extracted — "What population was studied?", "What dosage was used?", "What were the adverse effects?"

This creates a structured comparison table across dozens of papers without manually reading and coding each one — the most time-consuming part of a traditional systematic review.

Paper Summaries

For each paper, Elicit generates:

• A one-paragraph plain-language summary of the paper's key findings
• An assessment of the study's strengths and limitations
• A "Claims" extraction — the specific empirical claims made by the paper

Upload Your Own Papers

Elicit allows you to upload your own PDFs — useful when you already have a paper set and want Elicit's extraction and comparison features applied to documents not in its database.

Literature Review Export

Export your Elicit table as:

• CSV for Excel / Google Sheets
• BibTeX for citation managers
• Formatted table for inclusion in academic documents

Pricing

The Plus plan is the practical entry point for serious academic use — custom columns and up to 100 papers per search are essential for meaningful literature reviews.

Strengths

• Structured extraction: Column-based data extraction across many papers is unique and genuinely time-saving
• Custom columns: Define exactly the fields you need extracted — population, dosage, outcome measure — rather than being limited to pre-built fields
• Paper upload: Apply Elicit's extraction to documents you already have, not just those in its database
• Designed for researchers: The workflow (table view, column definitions, export) matches how academic researchers already think about literature review
• No hallucinated citations: Every paper reference is real and retrievable
• Honest about uncertainty: Elicit flags when extraction is uncertain rather than guessing

Limitations & Considerations

• Database size: 125 million papers vs. PubMed, Scopus, or Web of Science — may miss papers in niche databases or behind paywalls
• Extraction errors: AI extraction of specific data fields from complex papers occasionally makes errors — validate important extractions against original papers
• Free tier limited to 8 papers: Not sufficient for real literature reviews
• Not real-time: Newly published research takes time to appear in the database
• Academic focus only: Not useful for industry reports, news, or non-academic research
• Requires research skills: Elicit assists the research process — it doesn't replace knowing what to look for or how to evaluate evidence quality

Comparison: Academic Research Tools

Best Use Cases

When to choose alternatives:

• Quick empirical consensus answer → Consensus
• Broad web + academic mixed research → Perplexity or ChatGPT Deep Research
• Research against your own uploaded documents → NotebookLM
• Comprehensive journal database access → PubMed, Scopus, or Web of Science directly
• Developer search API → Tavily

Getting Started

1. Create an account at elicit.com — free tier with 8 papers per search
2. Enter a research question: "What are the effects of mindfulness-based stress reduction on anxiety?"
3. Review the default paper set and columns Elicit extracts
4. On Plus: add custom columns (e.g., "What was the sample size?" "What outcome measure was used?")
5. Export the table as CSV and continue analysis in a spreadsheet

Key Takeaways

• Elicit automates the data extraction step of literature review — building structured comparison tables from research papers without manual coding
• Custom extraction columns allow you to define exactly what data you need from each paper and extract it across an entire paper set
• Best for systematic reviews, graduate research, and evidence synthesis where structured comparison of many studies is needed
• Complements Consensus (quick consensus answers) and NotebookLM (analysis against your own uploaded documents)
• Free tier is limited to 8 papers; the Plus plan ($12/month) is the practical minimum for real literature review work