Elicit and Consensus: research-grade AI

Elicit and Consensus are research tools built for academic-grade retrieval. Different slot than Perplexity — purpose-built for literature review.

What Elicit is

A research assistant focused on academic literature:

• Search across millions of papers.
• Summarize findings per paper.
• Extract data from papers (sample sizes, effect sizes, study design).
• Compare studies in a structured matrix.
• Notebooks for ongoing research.

Designed for serious researchers — scientists, analysts, students writing papers.

What Consensus is

Similar space, different angle:

• Consensus-oriented — emphasizes synthesis of what the literature actually says.
• Simpler UX — quick questions, grounded answers.
• "Yes / No / Possibly / Mixed" labels on questions where the research is mixed.

Good for quickly answering "what does the research say about X?"

Where they shine

• Literature reviews. Covering dozens of papers on a narrow topic.
• Meta-analysis prep. Pulling structured data across many studies.
• Evidence-based decisions. Questions like "does X intervention work?"
• Academic writing. Finding the relevant cites for a claim.

Where they don't

• Non-academic topics. Industry reports, news, blog posts aren't their domain.
• Very recent research. Preprints are indexed but with lag.
• Synthesis quality. Good, but a specialist human outperforms on technical depth.

Using Elicit well

Query shape matters:

• Research question format. "Does intermittent fasting improve metabolic health in overweight adults?"
• Not keyword search. "intermittent fasting metabolic" gives worse results than a question.

Workflow:

1. Ask the question.
2. Review the paper summaries Elicit surfaces.
3. Use "Extract data" to pull structured info (population, interventions, results) from relevant papers.
4. Compare in the matrix view.
5. Read the top 3-5 papers in full for your synthesis.

Using Consensus well

Good for:

• Quick fact-checks on scientific claims.
• Understanding the distribution of research opinion.
• Surfacing key recent papers.

Not as strong for:

• Deep investigation requiring data extraction.
• Building a comprehensive literature review.

Free vs. paid tiers

Both tools have free tiers and paid tiers. Paid adds:

• Higher query limits.
• Data extraction features.
• Export to reference managers.
• Team features.

For occasional use, free tier covers you. For frequent research, paid is cheap relative to hours saved.

Limitations to know

• English-language bias. Coverage is strongest in English literature.
• Open-access bias. Papers freely available are over-represented; some paywalled literature underrepresented.
• Field coverage varies. Medicine, psychology, CS well-covered. Niche fields thinner.
• Recency. Preprints included but with varying indexing delay.

Integration with your workflow

• Reference manager integration (Zotero, Mendeley). Saves hours.
• Export to CSV/BibTeX for inclusion in your writing.
• Notebook features for ongoing research projects.

Comparing with general tools

Perplexity handles "what's the state of research on X?" adequately but surface-level. Elicit goes deeper: specific papers, extracted data, comparative views.

ChatGPT Deep Research is broader; Elicit is deeper on academic sources specifically.

For serious literature work, Elicit + reading the actual papers beats general tools. For quick answers, general tools are faster.

Verification remains essential

AI-extracted paper summaries are 85-95% accurate. That 5-15% matters:

• Always check effect sizes, p-values, sample sizes against the source.
• Read the methods section of any paper you'll cite.
• Be skeptical of papers you haven't read.

Elicit accelerates discovery; it doesn't substitute for reading.

Ethical use

• Attribution. Cite papers you use, not the tool that surfaced them.
• Don't plagiarize. AI summaries are starting points, not text to copy.
• Be transparent about your process if relevant to your audience.

The honest assessment

For literature-heavy research (academia, medicine, policy), Elicit is a legitimate productivity tool. Not a replacement for expertise or reading, but a strong accelerator.

For everyone else, Perplexity and ChatGPT Deep Research handle most research needs; Elicit is specialized to a narrower use case.