Science Beside U: How KRICT Turned Complex AI Research Into Clear, Credible Storytelling
Science communication works when it's honest, specific, and useful. That's the spirit behind Science Alive 2025 (Dec 11, IBS Science and Culture Center, Daejeon) and the PR Awards, which recognize press releases that translate research into public value.
At the center of one standout story: Na Kyung-seok, senior researcher at KRICT's Data Chemistry Research Center, and Kang Ga-ram, a senior administrator leading press outreach. We met them at KRICT's Daejeon headquarters on the 19th of last month to unpack how they built a press release that was both accessible and accurate-and why that balance matters even more with AI.
The research: AI to shortlist synthesizable materials
Last December, Na and collaborators led by KAIST's Professor Park Chan-young presented at NeurIPS, a top venue in the AI field. Their model streamlines the step scientists often stall on: which material candidates are actually worth trying to make.
Here's how they explained it to a general audience: diamonds (carbon-based) are stable but hard to produce because the process needs extreme heat and pressure. Water is the opposite-hydrogen and oxygen react easily to form it. The AI estimates thermodynamic stability and likely reactivity to flag candidates with a higher chance of successful synthesis.
The PR craft: clarity without distortion
Kang saw the news value and started the first draft-with help from AI-to speed up backgrounding. Na then reviewed the draft multiple times to fix subtle errors that came from oversimplification. That push-and-pull created something the public could understand without bending the science.
Kang put it plainly: AI now handles early knowledge gathering, so conversations with researchers move faster and deeper. The result is a release that keeps nuance where it counts and cuts the noise everywhere else.
Reality beats hype-especially with AI
Na's stance is direct: communicate the limits as clearly as the promise. "There are still many areas of AI where accuracy is low," he said. Media outreach should reflect that reality so expectations don't outpace what the tools can do.
That editorial posture pays off. Five months after the release, a medical school team reached out for a collaboration, expanding Na's group-traditionally focused on inorganic materials-into the bio field.
Building an AI-forward program with credibility
Kang is focused on turning accurate communication into a talent magnet. Earlier this year, KRICT launched the Digital Chemistry Research Center to build an autonomous chemical research system using data, simulations, and AI. She plans to highlight what's actually being invested and delivered-so applicants and partners know the ground truth.
What PR and research teams can copy tomorrow
- Start with a one-sentence "how it works," then add a single concrete example (like diamond vs water). Avoid analogies that create new misunderstandings.
- Use AI for prep, not final claims. Let it draft backgrounders and questions; let experts validate the science line by line.
- State limitations upfront: known failure modes, current accuracy, and where human judgment is still required.
- Quantify usefulness over grandeur (e.g., "cuts candidate screening time" vs vague superlatives).
- Invite cross-disciplinary contact with a clear call: who should reach out, about what, and why now.
- Create a revision loop: comms drafts, researcher redlines, comms reframe for clarity, final scientific check.
Why this matters now
AI claims are everywhere. Trust isn't. Teams like Na and Kang show that precision and accessibility aren't opposites-they're the standard. Get the science right, and the story spreads itself.
If your team is upskilling on AI workflows for comms and research enablement, these curated resources can help: AI courses by job function.
Your membership also unlocks:
