To find out more about the podcast go to What’s the evidence for vaccines?.
Below is a short summary and detailed review of this podcast written by FutureFactual:
Inside Science (BBC) August 7, 2025: Vaccine Funding Fallout, UK Isambard-AI Supercomputer, and Breakthroughs in AI and Space
BBC Inside Science's August 7, 2025 episode surveys science funding debates, cutting-edge computing, and bold research across medicine, robotics, and space. Host Marnie Chesterton and Caroline Steele examine the US decision to pause half a billion dollars in RNA vaccine funding, with expert Anne Willis clarifying vaccine effectiveness and safety. They then tour the UK's Isambard AI supercomputer, and hear how researchers are using it to study cancer detection across skin tones and to enable imitation learning for robotics. The episode also revisits DNA analysis of Napoleon's troops challenging long-held beliefs, and explores plans to study an interstellar object with space missions. It closes with a playful nod to science culture and audience contributions.
RNA vaccines: funding decisions and the science behind efficacy
The episode opens with the US Department of Health and Human Services halting a substantial portion of funding for RNA vaccine research, prompting questions about safety and long-term strategy. Prof. Anne Willis, director of the MRC toxicology Unit at the University of Cambridge, lays out the scientific reality: the COVID-19 RNA vaccines were highly effective against severe infection and remained a valuable tool against evolving variants. She explains that while mutations necessitate updates for diseases like flu and COVID, this does not undermine the safety or overall effectiveness of RNA vaccine platforms. The conversation clarifies vaccine platforms—mRNA, traditional virus-based, and inactivated vaccines—and argues that the UK remains committed to advanced therapeutics and cancer vaccines that also rely on RNA technology.
"The COVID RNA vaccines were highly effective against severe infection." - Prof. Anne Willis
The discussion highlights the difference between emergency-era funding and ongoing scientific standards, and what a shift away from RNA technologies could mean for future pandemics and broader medical innovation.
Isambard AI: Britain's new high-performance compute for research
Next, Inside Science visits Isambard AI, the UK’s most powerful supercomputer, housed in Bristol and funded by the government. The system aggregates nearly 55,000 Nvidia chips and a 200-gigabit network to deliver about 23 exaflops of performance when combined with other UK resources. Professor Simon McIntosh Smith explains why this scale matters: the ability to train large language models from scratch in the UK and to develop sovereign AI solutions with transparent data provenance. The project is framed as a strategic move to place the UK at the forefront of AI-enabled discovery and to reduce dependence on foreign compute capabilities.
"We can create UK versions of these things where we know exactly what data went into them." - Prof. Simon McIntosh Smith
AI for cancer detection and addressing bias in data
Researchers at the University of Bristol are using Isambard AI to probe whether AI-based cancer detection tools perform equally across skin tones. Dr James Pope describes a process that revealed bias when training data favored lighter skin tones, even after addressing dataset balance. The goal is to understand the sources of bias through explainable AI techniques and to mitigate them so that diagnostic tools serve diverse populations reliably. The work illustrates how high-performance computing accelerates iterative model development and robust evaluation in medical AI.
"There was bias with three data sets from commonly used data sets" - Dr James Pope
From wearables to imitation learning: analysing daily tasks with AI
In Bristol, researchers are also using Isambard to analyse footage from wearable cameras to understand everyday human actions. Prof. Dima Daman explains that the project aims to interpret behavior, assist wearers, and advance imitation learning for robotics. By studying how people perform tasks in real life—from kitchen routines to daily errands—the team hopes to teach robots to replicate natural human actions, improving automation and safety in real-world environments.
"There are two things we're trying to do one, to basically interpret this and assist the Human, the person wearing the camera" - Prof. Dima Daman
Historical science and space ambitions: Napoleon's army and comet Interceptor
The show moves to two science-frontier topics: historical DNA and space exploration. DNA analysis of 13 soldiers from the 1812 Russian campaign challenges classic accounts of disease-driven deaths, revealing the presence of pathogens such as Salmonella enterica and Borrelia recurrentis. While peer-reviewed publication is pending, the work demonstrates how modern genomics can revise history. On the space front, attention shifts to the interstellar visitor 3 Eye Atlas and plans for ESA’s Comet Interceptor, a mission concept slated for 2029 that could wait for an interstellar object to approach, enabling close study and potential discoveries about the origins of other star systems.
"DNA analysis could be used to fill in the blanks in historical records" - Caroline Steele
Closing notes: science culture, discovery, and public engagement
The episode closes with light segments on science storytelling—updates to Tom Lehrer’s elements song and viewer submissions—and a recap of how high-performance computing, AI, and interdisciplinary research are shaping the future of trusted science media. The voices in the studio celebrate curiosity, rigorous science, and the ongoing effort to bring credible science to a broad audience.