Each year, researchers across the globe run hundreds of thousands of studies to help improve our understanding of everything from traumatic brain injuries to biodiversity loss. They generate vast amounts of data — the raw material for all scientific discovery. Yet according to Frontiers CEO Dr. Kamila Markram, an alarming 90% of that data gets lost.
After the papers are written and the headlines fade, the raw datasets behind most studies get forgotten. They lie buried on lab computers, stuck in unreadable formats or simply never shared or stored. When that happens, science loses time, momentum, and the chance to build on what has already been discovered. This lost opportunity has slowed down scientific progress. However, that might be about to change.
A new initiative called FAIR² (pronounced “fair squared”) is trying to change that. It was launched in October by Frontiers, the open science publisher, as an attempt to save the research data we keep losing. FAIR² applies AI to assist scientists in preparing, reviewing, and making their datasets available so others can actually find and use them.
It turns static spreadsheets into interactive research repositories. It takes care of the more challenging aspects of scientific research such as formatting, quality control, metadata, peer review, and even visualizations. The goal with this tool is to speed up the entire process while ensuring
(NicoElNino/Shutterstock)
Frontiers, the organization behind FAIR², is a familiar name in the world of open-access research. Based in Switzerland, it manages a portfolio of peer-reviewed journals across disciplines like neuroscience, health, and environmental science. In recent years, it has become increasingly focused on the infrastructure of research. Instead of simply publishing findings, the company now emphasizes preserving and sharing the data that underpins them. FAIR² is part of that shift, an effort to close one of the most stubborn gaps in modern science.
“We’ve never seen a marketplace for data collaboration at this level, and until now, the space has been ignored for too long,” said Dr. Sean Hill, co-founder and CEO of Senscience, the AI venture powering FAIR².
“Science puts billions of dollars into creating data, and most of it in the long run is not useful, nor retrievable, and researchers hardly ever get credit. Now, with Frontiers FAIR², every dataset is cited, each scientist recognized at last for the crucial work of creating a dataset. It’s how cures will cure, climate solutions solve, and new devices get discovered — it’s how we unleash science.”
So what does this look like in practice? FAIR² is designed to make scientific data reusable, trusted, and properly credited. The process starts when a researcher uploads their dataset — it could be anything from genomic sequences to climate measurements.
The platform’s AI Data Steward then steps in to curate and prepare the data for reuse. It runs technical checks, inserts standardized metadata, and organizes everything according to FAIR²’s open specification. This turns a static set of numbers or images into something that can be understood, shared, and cited.
Four key outputs come from this process. The first is a certified and documented data package, ready for long-term use. The second is a peer-reviewed article explaining the dataset’s value, coverage, and limitations. The third is an interactive interface that lets others explore the data directly, using charts, summaries, and AI-assisted Q&A. The fourth is a certificate verifying that the dataset meets FAIR²’s technical and ethical standards.
(cono0430/Shutterstock)
Scientists who have tested FAIR² say it fills a gap that has slowed research down for years.
Dr. Vincent Woon Kok Sin, a researcher in climate and sustainability at HKUST, said the platform helped make his team’s global waste dataset more visible and accessible. That kind of visibility can make a big difference for researchers working in places where reliable data is hard to come by.
Maryann Martone, editor of the Open Data Commons, put it plainly. “Every PI would like their data to be findable and reusable,” she said. “The real bottleneck has always been the length of time and amount of effort this takes.” FAIR² helps cut through that.
Across disciplines — whether environmental policy or health research — people are beginning to notice the same thing. This is more than a storage platform. It is a way to take data off the shelf and put it to work. For researchers. For AI models. And for people with questions who cannot afford to wait ten years.
FAIR² is not the only project working to fix the way science handles data. Similar efforts are gaining traction around the world. The Allen Institute’s Brain Map is building large-scale open datasets in neuroscience that anyone can explore. The Human Cell Atlas is mapping every cell type in the human body and making the data freely available and standardized. NASA’s Earthdata platform is another example, offering environmental and climate datasets that are already cleaned and ready for analysis.
What sets FAIR² apart is how early it steps in. It helps scientists shape their data before it is shared, not after. That small shift makes a big difference. It means the data is ready to be understood, reused, and trusted from day one. If more science worked this way, less of it would be lost. And more of it might finally move the world forward.
Related Items
Data is at the Center of Scientific Discovery Inside MIT’s New AI-Powered Platform
The History of Data Science: From Cave Paintings to Big Data
How Scientists Are Teaching AI to Understand Materials Data
The post Science Loses 90% of Its Data. A New AI Approach Could Change That appeared first on BigDATAwire.
