Skip to Main Content

Consensus

An overview of the Consensus.app tool

Specific Guardrails Against AI Misuse

Consensus outlines the following safeguards that it's built to reduce the risk of AI misuse:

  • Search before synthesis: Every response starts with a literature search. This ensures all citations are real papers, never hallucinated or invented sources.
  • Summarization, not speculation: Consensus's AI tools are strictly limited to summarizing content from the papers retrieved. They don’t “fill in the blanks” with outside knowledge.
  • Transparent sourcing and attribution: Every claim is cited, and every citation is clickable. Consensus's interface enables users to inspect, verify, and reference the original source content with one click.
  • AI “checkers” for relevance: Before summarizing, Consensus uses separate models to verify that the source actually contains information relevant to the query. If a paper doesn’t meet its threshold for relevance, it won’t be used in the response. 

Consensus is built to enhance – not replace – the research process.