Artificial intelligence (AI) is changing evidence synthesis. It offers clear opportunities to accelerate and streamline the way systematic literature reviews (SLRs) are conducted. However, efficiency alone is not enough, evidence syntheses need to be carried out in a manner that is robust and reproducible and should stand up to critical scrutiny. So the question is, can AI improve the SLR process without compromising the necessary rigour, transparency, or scientific credibility?

At Broadstreet, rigour and reproducibility have always been central to our SLR work. For the last several years, we have been watching the growth of AI with cautious interest. We did not want to adopt AI simply because it was available or even just because the market was moving in that direction. We took our time and deliberately evaluated available solutions that have demonstrated validity before incorporating AI into our service offering.

With our many years of experience conducting evidence syntheses, we knew what we wanted in our AI-enabled SLR service offering. Our solution offers flexible human levels of involvement; being able to vary the degree of AI autonomy and researcher verification by project and task allows us to maintain full control of how we apply AI assistance and ensure customization to meet our clients’ needs. Further, we ensure full traceability and auditability of decisions and outputs. With data extraction in particular, structured outputs that can be readily verified against source publications are a necessity.

In our view, our work should be AI-enhanced, not AI-replaced. AI is most valuable when it handles repetitive, high-volume tasks like screening large citation sets, locating information, extracting data, and facilitating quality control, efficiently. Human expertise remains essential for interpretation, ambiguity resolution, methodological decisions, and scientific judgment. The balance of AI and human involvement needs to be tailored to each review depending on how the evidence will be used.

For our clients this means potentially faster and more efficient SLR execution. It will allow us to shift researcher time from repetitive processing toward higher-value scientific and methodological work while keeping transparency and traceability intact. The objective is to promote efficient use of expert researcher time without sacrificing quality.

Over the next few years, we expect evidence synthesis will continue to evolve rapidly under the influence of AI as its capabilities continue to evolve. The key differentiator will not simply be access to AI, but how responsibly and thoughtfully it is embedded into rigorous evidence-synthesis workflows. Successful AI-enabled SLRs require the right combination of technology, process, oversight, and methodological expertise.

We see our new AI-enhanced SLR offering as a natural evolution of Broadstreet’s evidence-synthesis capabilities. Our goal is not to automate the systematic review; rather, it is to maximize systematic reviewers’ efficiency while preserving the rigour, transparency, and scientific judgment that high-quality evidence synthesis requires.

Please get in touch to discuss how AI can enhance your evidence synthesis project.