Building Trustworthy Scientific Infrastructure

How CERN Earns Trust at Scale

There’s a 54-kilometer-long bike path on the border between France and Switzerland, meandering in a rough circle through the foothills of the Jura Mountains, low-lying farmland, and quiet suburban towns outside Geneva. Every four kilometers or so, an exhibition platform shows visitors that they’re cycling above the most powerful particle accelerator in the world. Three hundred feet below the bike path, powerful electromagnets accelerate beams of protons in opposite directions to nearly the speed of light before the beams collide head-on, recreating the universe’s earliest moments.

Why does the Large Hadron Collider, operated by the European Organization for Nuclear Research, or CERN, have a bike path above it? It’s one of many deliberate steps the organization has taken to engage the public, writes Kyriaki Papageorgiou. With visibility treated as “a central institutional task,” along with other practices and standards that build credibility with the public, CERN has “made large-scale collaboration functional, productive, and credible, … offering unusually rich accounts of how trust is built in practice.”

As Europe seeks to build an infrastructure for the use of artificial intelligence in science, Papageorgiouargues “there is much to learn from CERN’s governance and institutional design, as well as the everyday practices through which its experiments are made visible and collectively monitored, claims are slowed down, uncertainty is explained, and trust is produced.”

FUTURE TENSE FICTION

Future Tense Fiction, our collaboration with ASU’s Center for Science and the Imagination, features short speculative fiction paired with original artwork and a response essay from an expert who connects the story to real-world policy issues. This week we’re publishing a brand-new story by Troy L. Wiggins.

Illustration by Rey Velasquez Sagcal

What happens to abandoned technology after conflict? This month’s Future Tense Fiction story, “Critical Functions” by Troy L. Wiggins, follows two scrappy cousins battling to get by in a postwar bartering economy. When they encounter an abandoned autonomous military system, they get pulled into a dilemma that reveals some dark secrets about their divided world.

Plus: Military studies scholar Lena Trabucco describes the challenges of designing artificial intelligence systems that human operators can truly rely on.

THREE QUOTES AND A PROVOCATION

Some highlights from last week’s Open Editorial Meeting on biosecurity and AI:

Sonia Ben Ouagrham-Gormley: In past bioweapons programs, projects fail—even when the expertise is available—in the absence of proper integration and coordination of the work.... AI, for now, is not changing the challenge of bioweapons development. But the other problem we are facing is that we really don’t know what AI is doing, or can do, because most of the analysis is focused on the first stage of development: the design, the acquisition of material or data. What happens after is not analyzed at all. Will the uplift of AI have repercussions on the other stages of weapons development, or is it just limited to the design stage?

Edward You: The critical nodes within the federal government, especially when it comes to law enforcement intelligence, to address these bio-threat matters, don’t exist anymore, or they’ve been drastically reduced.... If you’re very concerned about AI biotech, who the hell do you notify if you do detect something suspicious?

Yong Bee Lim: The big elephant in the room that no one ever talks about is trust and transparency from ... the frontier AI companies themselves. And so, basically, we’re getting a lot of really second- or third-hand information, because there’s so much filtering that happens from the frontier companies in regards to their products. And then the question becomes: How can we, as biosecurity experts, assess the information in a way that is good, and reliable, and applicable to the policy process?

Provocation: [paraphrased from Ed You] Today there’s a lot of focus on the potential “zombie apocalypse” aspect of AI biotechnology, but the problem may be less states or terrorists than criminal networks who are trying to make narcotics. We may be going from Breaking Bad to Brewing Bad. We should be funding education in this space because it’s no longer biosafety or biosecurity as we know it, and we need to equip everyone—from high schoolers to policymakers—to understand these risks. We need to become more and more responsible, but we’re not training on that.

(Stay tuned for details about our next Open Editorial Meeting in December, topic TBD.)

There are few relationships more complex to narrate than the relationship between a mother and a daughter. Rachel Aviv’s You Won't Get Free of It: Stories of Mothers and Daughters mines those complexities with her characteristic depth and curiosity; she goes deep on a series of compelling and often contradictory characters whose specific thoughts, dilemmas, mistakes, and motivations make the reader question their own. Aviv’s book, a collection of seven essays originally published in The New Yorker, follows mothers searching for daughters and daughters searching for mothers in a way that made me reflect on belonging, lineage, gender, and what we expect from our families of origin. (For an equally deep and nuanced exploration of the bonds between fathers and sons, check out the Reaching Out podcast from KCRW and Futuro Media, in which host Sayre Quevedo embarks on a journey to meet his birth father.)

—Mia Armstrong-López, managing editor, ASU Media Enterprise

New print subscribers will receive the Winter 2027 issue and get a year’s worth of insights into today’s most essential topics—for as little as $25.

Know someone who appreciates Big Science projects like CERN? Send them this newsletter so they can sign up and join the science policy conversation.

Issues in Science and Technology is a publication of Arizona State University and the National Academy of Sciences.

National Academy of Sciences, 500 5th St NW, Washington D.C., 20001