• Sign-up for Consensus
• Consensus Overview
• Searching with Consensus
• Accessing Full-Text
• Limitations and Cautions
• Guide Attribution

Join us for an online workshop on how to use Consensus. Register below:

November 4, 2025 @ 3:00 PM

December 1, 2025 @ 3:00 PM

Consensus is an AI search tool specifically focused on academic research content. Use it to search for peer-reviewed literature, not to ask general purpose questions.

Consensus searches content from Semantic Scholar, OpenAlex, and their own "crawl of the scholarly web." Consensus is not searching subscription-based content available through other OSU Libraries resources like Scopus, but can connect OSU users to paywalled content it finds through its own search algorithm.

The default setting with the Consensus Pro account is to retrieve and analyze 20 papers based on your prompt.

You also have access to 50 "deep search" searches per month. Deep searches are a more intensive (and somewhat lengthier) search that results in a report of the literature on your research question. The report includes a summary, an overview of research gaps, indications of the consensus on the topic across the field, and clickable links to the literature included in the report.

Pro Search

Use the Pro Search to do a broader search of the literature, which will ultimately result in an analysis of up to 20 papers Consensus determines to be most relevant to your research question. You can ask your question using keywords, open-ended questions, or requests for specific output styles, for example, "provide a bulleted list" or "create a comparison table." (Note that a "quick search" can be performed, but it only analyzes 10 papers - with our trial access, you can run unlimited "pro" searches).

A strength of Consensus is the ability to filter your search. Choose from options like journal rank, study design, or field of study.

Deep Search

Consensus Deep Search attempts to provide a detailed report on your topic by breaking down your question, running multiple searches, ranking and reviewing up to 50 papers based on relevance, and structuring a review that provides interactive visuals related to top authors, evidence strength, and research gaps. Since this search is more comprehensive and complex, it often takes a few minutes to run, so be sure to be patient! 

With our trial access to Consensus, OSU users will be able to run 50 Deep Searches per month. 

See an example of a Deep Search on the topic of information inoculation and misinformation sharing.

Additionally, with Deep Search, Consensus attempts to provide some transparency and explanation for its search strategy, so you can better understand how different searches resulted in new papers, potentially providing you with ideas on how to further adjust or adapt your own searches.

Search Features and Visualizations

Based on your query or prompt, Consensus can include visual figures in the analysis of any search, such as the Consensus Meter, Research Gaps, Top Contributors, and the Results Timeline.

Consensus Meter

Enter a "yes/no" question to get an overview of the "consensus" on your research question. For example, "do people really prefer robots with human features?" If at least 5 papers are available that directly answer your question, the tool will provide an overview of the findings, including a Consensus Meter illustrating the number of papers in agreement or disagreement with your question. Note that while this feature can provide a quick sense of which way the research has trended, it is limited to the 20 most relevant results and the model may sometimes misclassify or miss nuance in results.

In addition to the Consensus Meter display, there is a narrative summary of the papers included in the search as well as a tabular summary.

Results Timeline

The Results Timeline attempts to plot the top search results chronologically; each point represents a paper, with larger points indicating more citations. This allows you to trace the progression of ideas and spot influential works quickly. Simple keyword-style queries will often show the Results Timeline, or you can specifically ask Consensus to show it by asking for a timeline or how the research has evolved over time on any query.

Research Gaps

Consensus analyzes existing studies, and attempts to identify areas that need further exploration, guiding future research efforts.

• For Open Access papers, users can directly download the full-text PDF through Consensus. Click the PDF button for that article in order to download. 

• For paywalled articles, Consensus provides a direct link to the publisher’s page along with metadata and abstracts to help users determine relevance, without violating any copyright laws. Click the "Full text" button to go to the publisher's page.  If OSU has subscription access and if you are either on a campus network or have logged in through "Anywhere Library Access", then you should be able to access the full text.

• For articles that we do not have access to, you may request through Interlibrary Services.

While Consensus can be a useful addition to your research toolkit, users should approach any and all AI-generated summaries critically. While the sources returned in Consensus results should not be "hallucinated", the generated summary or analysis of those articles can be flawed. The platform attempts to mitigate common AI hallucination risks by searching real peer-reviewed papers before synthesis, but potential misinterpretations can still occur. Specifically, hallucinations may manifest as misread sources where the AI summarizes a real paper incorrectly, potentially introducing inaccuracies or misrepresentations of research findings. Users should always independently verify key claims and dive into the original source materials to develop comprehensive understanding.

The platform's database, though extensive with over 220 million peer-reviewed papers, does not encompass all research, and summaries represent only a portion of available literature on any given topic (as is the case with any database or research literature search tool). Additionally, due to publisher licensing and copyright limitations, for many articles only the abstract, and not the full-text, is available for analysis by the tool. As such, Consensus should be viewed as a research enhancement tool that accelerates initial exploration, not as a definitive or comprehensive research solution.

Thanks to Librarian Hannah Rempel at Oregon State University for providing permission to reuse and remix this guide.