Elicit: What is it?
Elicit is an AI research assistant. Elicit uses language models to help you automate research workflows, such as undertaking parts of literature reviews.
Elicit can find relevant papers without perfect keyword match, summarize takeaways from the paper[s] specific to your question, and extract key information from the papers. [@elicit-research].
Get started with Elicit:
- Go to: elicit.com
- Click on 'Get Started'
- Create a Basic* (free) account - add your email address and create a password
- Claim free Basic credits (5,000).
*Elicit Basic: essentially a free trial. You can't export results or transfer unused credits. Credits are deducted when you search, run workflows, add columns to tables, etc. For example, a search uses approx. 200 credits. You can only add more credits if you subscribe.
Using Elicit to analyse research papers
- Use: Find papers to search across the Elicit database
- Use: Extract data from PDFs to search across papers you already have in PDF format (useful for screening papers you've already retrieved from another source, e.g. databases)
- PDFs uploaded* will be added to your own 'Library' on Elicit - these are only visible to you and will not be added to the Elicit database
- Use options to add columns and retrieve a range of data** (e.g. intervention effects, study design, duration, length to follow-up, dose, participant count, participant age, etc.) from your results
- You can continue to query results using additional natural language searches
- You can do a citation network search (also known as chaining, citation trailing, snowballing, etc.)
*Note: File size limited to approx. 20MB per paper.
**Note: Pre-set columns currently have a biomedical focus, but you can create custom columns in Elicit.
Using Elicit to explore concepts
- Use: List of concepts to summarise concepts across papers
- Elicit attempts to identify concepts and group these together (e.g. when a specific effect is discussed across multiple papers, Elicit will show the effect, then show all the papers and quotes that support this effect)
- Organises information in a more 'decision relevant' or 'query relevant' way and displays a summary of supporting academic literature alongside items retrieved*
- List of concepts combines workflows:
- Found n papers = Find
- Found n concepts in n papers = Extract
- Found n unique concepts = Deduplicate
- Final answer = Results**
- Workflows are saved to your Elicit history for retrieval at a later stage if required
- Ranking of results is 'mostly semantic'.
*Note: May include results from language models (such as ChatGPT) if supporting literature can't be found in existing sources - be wary of possible hallucinations, check sources.
**Note: Options to extract results only available with Elicit Plus subscription.
Elicit - Advantages:
- Helps to speed up the process of conducting a literature review
- Can help automate systematic reviews and meta-analyses
- "Elicit tends to work best for empirical domains that involve experiments and concrete results. This type of research is common in biomedicine and machine learning." [Elicit Help Center].
Elicit - Disadvantages:
- Elicit Basic (Free) = 5,000 credits, one-time use only (free trial); no export option
- Elicit Plus ($10 per month / $120 per year) = 12,000 credits per month (unused credits cannot be carried on to the next month, but you can add more credits if required)
- Credits are deducted when you run workflows and add columns to tables in Elicit
- Elicit is not so good for answering questions or retrieving information that has not been included in academic journal articles
- Elicit does not work well in helping to identify facts in theoretical or non-empirical domains
- Elicit searches within a limited set of full-text results (Semantic Scholar), although if a DOI is provided in results, you can retrieve full-text from external sources
- Elicit currently has a biomedical focus.
Support resources:
