I’m often frustrated that I don’t have time to read the torrent of cool papers coming faster and faster from amazing people in relevant fields. Other scientists say the same—they’re underwater, too. And, if I’m honest, many colleagues probably don’t have time for my lengthy conceptual papers either. So whom are we writing these papers for?
That question feels urgent. Science is supposed to be a great, ongoing conversation—across disciplines, generations, and continents. But today, we’re generating data and knowledge at a pace unmatched in history, and so much of it risks being unread, untested, or unused.
Lately, I find myself thinking about the audience for our work. Today, it’s mostly our peers and students, doing their best to keep up. But soon, our readers may not be human in the familiar sense. The only entities with the “bandwidth” to read everything—synthesizing the mass of literature and data—could soon be AIs: research copilots, lab automation platforms, hybrid systems, and augmented researchers. Not just today’s language models, but whatever comes next—capable of interrogating results, connecting distant dots, and accelerating discovery in ways we haven’t seen before.
If that’s even partly true, our scientific communication needs to evolve. If we only write for present-day humans, we’re leaving vast potential on the table. But how do we publish for a world where tomorrow’s readers could be algorithms or systems we haven’t yet imagined?
This is where it’s crucial to distinguish between FAIR², the open specification, and the new Frontiers FAIR² Data Management service.
FAIR² is an open, community-driven specification—a shared set of standards for what truly usable, machine-actionable research data and methods look like. The aim is to make data (and all the nuanced details that make it valuable) as discoverable, interrogable, and reusable as possible—not just for other scientists today, but for whatever forms of intelligence science brings tomorrow. FAIR² is designed to be publisher-agnostic, discipline-spanning, and future-proof. And importantly, it is not a product or a platform. Instead, it’s a specification that will be stewarded by a new nonprofit association—a neutral body that we are in the process of setting up now. That nonprofit, and the broader community, will govern FAIR²’s evolution, keeping it open and truly in service to science.
Alongside that, there’s the Frontiers FAIR² Data Management service, which is the practical, researcher-facing platform my team at Senscience has built to help bring FAIR² to life. This service, which is currently in its pilot stage, operationalizes the FAIR² specification—making it possible for scientists to easily create, document, and publish FAIR²-compliant data packages and methods. At the heart of this service is the AI data steward: a digital collaborator that helps you upload not just your data, but your detailed protocols and context. It prompts you for what’s missing, helps standardize and structure your methods, and ensures that your work is transparent, machine-actionable, and ready to be reused by both humans and AI. You remain the expert; the AI just lightens the load.
The distinction is important. FAIR² as a specification is about defining the rules of the road—open to all, designed for everyone. The Frontiers FAIR² Data Management service is one way (not the only way) to actually travel that road, supporting researchers who want to make their work reusable, discoverable, and part of science’s living memory.
Here’s what I want to be clear about: I hope and expect that humans will always be at the heart of scientific discovery. The intuition, creativity, and collaborative spirit of real scientists—asking questions, making leaps, seeing the world differently—are irreplaceable. But the acceleration possible with AI tools is undeniable. When data and methods are truly FAIR²—open, transparent, and machine-actionable—AIs don’t replace us; they amplify us. They become partners in finding patterns, surfacing contradictions, and connecting dots across a landscape too vast for any individual or team to map alone.
This is the promise I see in FAIR² and in Senscience’s practical implementation of it. It’s a chance to reimagine what science can be: not just a static record, but a living, growing web of knowledge. The easier, more rewarding, and more routine it is to leave a trail that others (human or AI) can follow, the stronger the whole scientific ecosystem becomes.
Of course, not every dataset will be perfect, and not every method will be described flawlessly. There will always be practical challenges and some fear in opening up your work to scrutiny. But with the right support, tools, and community standards, sharing becomes less a leap of faith and more an investment in the future of discovery.
So, who will read our science? Maybe not everyone. Maybe not even most people. But if we build and use the right open infrastructure—FAIR² as a shared, evolving foundation, and platforms like the Frontiers FAIR² Data Management service as enabling tools—then our work has a chance to matter far beyond the present moment. Not just now, but for whatever future science brings: for the next generation of humans, and for the intelligent systems we’ll collaborate with.
The future of science will be written for many audiences. Let’s make sure we’re writing for all of them.
Sean Hill is a neuroscientist, founder of Senscience, and advocate for collaborative, living science.