Quick Buzz Feed
Two tools have been selected from a pool of 48 submissions received in late 2025 to participate in Cochrane’s innovative platform study evaluating AI’s role in evidence synthesis.
Cochrane has officially announced Laser AI and Nested Knowledge as the two artificial intelligence (AI) tools chosen for its innovative platform study. This study aims to evaluate how AI can support and enhance various stages of evidence synthesis. These tools were selected from a total of 48 submissions received in late 2025, with the selection process strictly adhering to the Responsible AI in Evidence SynthEsis (RAISE) principles. Artur Nowak, Co-founder of Evidence Prime, expressed enthusiasm for Laser AI's participation, emphasizing Cochrane's methodological leadership in understanding AI's role in reshaping workflows. Similarly, Kevin Kallmes, CEO and Cofounder of Nested Knowledge, highlighted their alignment with Cochrane's mission and eagerness to learn from its experts regarding best practices for AI-enabled systems in evidence synthesis. Five other tools are on a reserve list for potential future inclusion. Cochrane has established formal, legally binding agreements with the developers, including a minimum financial contribution, to ensure compliance with standards such as data protection and intellectual property during the study.
The selection of these AI tools does not signify a formal endorsement by Cochrane. As per Cochrane's published position statement, authors are permitted to use AI tools provided they can demonstrate that the methodological rigor or integrity of their evidence synthesis will not be compromised. The current platform study itself is designed to evaluate this very aspect for the reviews it includes. For authors considering AI tools, Cochrane recommends adhering to the RAISE guidance, particularly RAISE 3, which offers advice on selecting and utilizing AI tools for evidence synthesis. New guidance, released in March 2026, further assists systematic reviewers by offering an overview of AI applications across different review stages and tool types, coupled with recommendations for their use. This guidance categorizes AI use into five recommendations: 'Acceptable for use' (outputs usable directly if limitations are noted), 'Human verification required' (outputs support tasks but need thorough human checking), 'Requires validation within the review' (outputs usable if performance is explicitly evaluated and adequate), 'Exploratory and supplementary use' (outputs for idea generation, requiring extensive human refinement or supplementary use), and 'Not acceptable for use' (current technology has serious limitations). The guidance also includes a table detailing the current state of AI tools for tasks like protocol writing (question formulation, drafting) and the search process (exploring literature, search strategy development), outlining tool classes, considerations, examples (e.g., ChatGPT, Carrot2, TERA WordFreq), and their respective recommendations. Additionally, Cochrane has simplified RAISE 3’s framework into key considerations to help systematic reviewers critically assess AI tools, encompassing ethical, legal, and regulatory aspects for tool selection.
Currently, the research team and two Cochrane review author teams have commenced training on both Laser AI and Nested Knowledge. Following this onboarding, a piloting phase for the protocol will be initiated with these two teams, after which additional Cochrane review author teams will be integrated into the study. Cochrane anticipates conducting interim analyses by mid-year 2026, with the full results of the study expected in the latter part of the same year. Beyond the study, Cochrane is actively working to enhance AI literacy and promote best practices for responsible AI use across its global community. This effort involves collaboration with the joint AI Methods Group and participation in partner projects such as Destiny, which focuses on Digital Evidence Synthesis Tool INnovation for Yielding Improvements in Climate & Health. Furthermore, Cochrane is dedicated to developing and disseminating comprehensive guidance, training materials, and resources specifically tailored for systematic reviewers, editors, and the broader evidence synthesis community. Acknowledgment is given to the Wellcome Trust (grant number 323143/Z/24/Z) for providing support to a part of the AI tool platform study.
For those interested in delving deeper, additional resources are available. Readers can access detailed information regarding the comprehensive selection process that led to the chosen AI tools and understand how Cochrane's core values were integrated throughout this rigorous evaluation. Furthermore, information is provided on Cochrane's initial launch of this innovative study designed to assess various AI tools specifically for evidence synthesis applications, offering a broader context to the current announcements.