Dinosaurs as Living Birds, AI in Science, and US Science Funding Debates

Episode snapshot

This episode surveys the idea that birds are dinosaurs, explores AI as a potential partner in scientific discovery, and examines proposals to change how the US funds science. It also includes discussions on space age laundry research, geoengineering field trials, and New Zealand amny survivors of the dinosaur era.

• Birds as dinosaurs and the fossil evidence linking modern avians to their reptilian ancestors
• AI as a computational co scientist and the challenge of hallucinations in large language models
• US science funding policy changes and the tension between independence and political oversight
• Geoscience and policy updates including Arctic sea ice geoengineering experiments and space related research

Overview

Inside Science presents a multifaceted look at science in the current moment. The episode begins with a deep dive into the long standing idea that birds are dinosaurs, framing this as a living connection rather than a distant fossil. It then moves to the cutting edge of AI assisted science, followed by a discussion of science funding policy in the United States. Throughout, the program blends interviews, expert commentary, and recent research to illuminate how scientific knowledge is built, contested, and applied.

Birds as living dinosaurs the paleobiology segment

The centerpiece of the episode features a conversation with Steve Brusarti, a respected palaeontologist, about the enduring legacy of dinosaurs in today’s birds. The host traces the historical arc from early Darwin era debates to Thomas Huxley, who championed the link between birds and dinosaurs using evidence from reptilian like features in modern bird anatomy and the then new Archaeopteryx fossils. The narrative then moves to key turning points: Ostrom’s 1973 Nature paper reigniting the dinosaur bird hypothesis after Deinonychus revealed a feathered dinosaur sister lineage, and the late 1990s discoveries of feathered Dinosaurs in China that provided the decisive feathered fossil evidence ornithologists had demanded. The discussion expands into how this fossil record supports the idea that dinosaurs did not vanish but evolved into birds, and how isolated ecosystems like New Zealand preserved dinosaur like niches long after their mainland counterparts had disappeared. The kiwi and other endemic birds are described as living echoes of a lost world, offering a lens into how evolutionary transformation produced the modern avian fauna we observe today.

AI in science the potential co researcher

Another major thread centers on artificial intelligence as a research partner. Claire Bryant from the University of Cambridge describes her experience with Google’s latest AI system and the CO scientists framework that pairs with Gemini, a general large language model. The guest explains that these systems enable researchers to parse vast literature at scales beyond any single human capability, helping to form and refine hypotheses. The discussion includes a concrete example of how the AI helped identify specific molecular switches implicated in cross species disease dynamics, and how researchers used human oversight to quality check AI outputs, iterate with preliminary lab data, and narrow down to precise protein and amino acid differences between humans and birds. The interview also touches the tension around hallucinations, with the host and expert noting that current AI tools can produce made up quotes or facts, and that the problem is actively being worked on. Pushmeet Kohli from Google DeepMind adds depth by describing the role of AI as a lens rather than a replacement for human thinking, emphasizing that the technology is accelerating science by shifting the human role from solving puzzles to architecting the right questions. The segment closes with reflections on how scientists should respond to these advances, stressing that the technology aims to expand scientific activity rather than reduce it.

Science policy and funding in the United States

The program shifts to a policy themed discussion with Liz Ginexi and Roland Pease about proposed changes to how federal science funding is allocated. The Federal Register, the White House Office of Management and Budget, and advisor Russell Vogt are noted as central actors in the plan to require discretionary grants to undergo political pre issuance review, effectively moving some grant selection decisions away from independent experts and civil servants toward the White House political apparatus. The conversation unpacks what “gold standard science” could mean in this context, highlighting the lack of precise definitions in the proposal and the potential for politicization of funding decisions. A key point is that despite this rhetoric, Congress maintains the constitutional power of the purse and determines funding levels, complicating the President’s ability to unilaterally direct research priorities. The discussion also touches on concerns raised by the scientific community about maintaining independence and the role of peer review in funding allocation, and how this could impact U S leadership in global science collaborations.

Field tests and geoengineering updates

Caroline Steele reports on recent geoengineering experiments aimed at thickening Arctic sea ice. Field trials in Canada and Svalbard Norway tested the concept of covering sea ice with seawater to promote rapid refreezing. The results were mixed: Canada observed a modest delay in sea ice melt, extending the summer duration by about seven to ten days, while the Norwegian study did not show longer lasting ice. The discussion emphasizes the scale of the experiments and the many factors involved, including warming effects of the act of pumping water and the energy use of the equipment. A separate caution from scientists who argued that geoengineering efforts could distract from aggressive emissions reductions is noted, underscoring the policy and ethical complexities of large scale climate intervention. The segment frames geoengineering as an area requiring careful cost-benefit analysis and rigorous governance before broader deployment.

Space laundry and microbial control

The program concludes with a practical, quirky look at how astronauts manage washing on the International Space Station. With limited water, there are no traditional washing machines; clothes are worn for long periods and then burned up upon reentry. Researchers at the Astrobiology conference present a plasma based laundry gun concept that uses ionizing electricity to generate reactive ions that kill microbes on fabrics, potentially reducing odors and extending wearing time in microgravity. The current prototype works on a small patch, but researchers are exploring scaling up toward a plasma washing machine suited for space missions.

New Zealand’s last outposts of the age of dinosaurs

Lastly, the episode revisits the idea that New Zealand represented a dinosaur dominated ecosystem long after the era of giant land dinosaurs ended elsewhere. The kiwi and a few remaining sublime avian species are framed as remnants of that age, with the moa and Haast’s eagle serving as dramatic reminders of how isolation produced unique ecological niches until human arrival dramatically altered the landscape. The conversation closes by inviting listeners to reflect on how present day fauna carry traces of deep evolutionary history.

Closing

Listeners are encouraged to share ideas about science topics they want investigated next, emphasizing the program’s mission to explore, explain, and connect science across disciplines.