Elicit

Elicit is an AI research assistant built specifically for academic and scientific literature review. It searches over 200 million papers indexed from Semantic Scholar, extracts data from documents, and summarizes findings across sources in structured formats designed to support evidence-based conclusions. The core claim is that it can compress hours of literature review into a faster, more organized workflow without sacrificing rigor.

The tool was developed by Ought, an AI safety research organization focused on building AI systems that are transparent and useful for reasoning-intensive tasks. That background shapes what Elicit is optimized for: not speed-generating content, but accurately surfacing and representing what the research literature actually says.

**How Elicit approaches literature review**

When you ask Elicit a research question, it searches its paper database, identifies the most relevant studies, and extracts specific data points from each: study design, sample size, measured outcomes, key findings, and limitations. These are presented in a structured table that allows side-by-side comparison across papers — a format that mirrors how systematic reviewers manually synthesize literature.

The extraction is noteworthy because it goes beyond summarization. Elicit can pull specific numerical results (effect sizes, confidence intervals, sample characteristics) from the body of papers, not just the abstract. This is the capability that makes it genuinely useful for systematic reviews and meta-analyses, where abstract-only review misses critical methodological detail.

**Built-in research workflow features**

Beyond search and extraction, Elicit supports several research workflow tasks: identifying concept trees (related search terms and their relationships), generating background literature for grant applications, and summarizing long documents. The Notebook feature allows you to save papers, organize findings, and annotate alongside Elicit's extractions.

Elicit is transparent about its uncertainty. When it is unsure about an extracted data point, it flags this rather than presenting the extraction with false confidence. This matters in research contexts where downstream decisions depend on data accuracy.

**Pricing**

Elicit offers a free tier with limited monthly queries — sufficient for occasional use or trial. The Plus plan ($10/month) or Basic plan increases query limits for more sustained research work. Full-featured access for intensive systematic review workflows requires a higher tier.

**Where Elicit falls short**

Elicit's database, while large, does not cover every journal or preprint server. Grey literature, conference proceedings, and emerging preprints may be missing or delayed. For highly specialized fields or recent publications, coverage gaps exist.

The AI extraction is not infallible — it can misattribute a finding or extract data from the wrong table. Experienced researchers treating Elicit as a primary source rather than a verification shortcut will need to spot-check extractions against the original papers, particularly for numerical claims.

**Who Elicit is designed for**

Elicit is most valuable for graduate students, academic researchers, science journalists, and research-heavy knowledge workers conducting literature reviews, systematic reviews, or evidence synthesis. It is less useful for casual question-answering or general research tasks where Wikipedia-style answers would suffice.