The 95th Joseph Henry Lecture
Agentic AI and the Search for Extraterrestrial Life
The SETI-NVIDIA Collaboration at the Allen Telescope Array
Michael Garrett (U. of Manchester) & Adam Thompson (NVIDIA)
Michael Garrett
Sir Bernard Lovell Chair of Astrophysics
University of Manchester
Chair, SETI Permanent Committee of the International Academy of Astronautics
Adam Thompson
Head of Product - Edge Supercomputing
NVIDIA
In Memory of PSW Science Member Dick Garwin
Video
About the Lecture
For millennia, humanity has looked to the stars and asked, “Are we alone?” Today, motivated acutely by exciting discoveries across astronomy, the Search for Extraterrestrial Intelligence (SETI) has transformed this profound philosophical question into a rigorous, data-driven observational science. As we survey the cosmos for technosignatures—remotely detectable indicators of advanced technology beyond Earth—our observational scope is expanding exponentially, from upgraded facilities like the Hat Creek Radio Observatory, the Very Large Array (VLA), and MeerKAT in South Africa, to next-generation giants like the Square Kilometre Array (SKA) and ngVLA. These efforts culminate in conceptual missions to the lunar farside, where the Moon will shield our instruments from terrestrial radio interference.
However, the sheer scale of the universe presents an unprecedented data challenge. The cosmos is vast, and the radio spectrum is incredibly dense. Isolating a faint, engineered anomaly from a background of cosmic noise and terrestrial interference presents profound challenges. Modern radio astronomy instrumentation generates staggering volumes of raw data, demanding real-time, high-speed anomaly detection across billions of frequency channels simultaneously.
The bottleneck in modern astrophysics is no longer exclusively our ability to collect photons and radio waves; it is our capacity to process them. To push the boundaries of the search for technosignatures, we are fundamentally rethinking our approach to signal processing, observatory operations and ultimately observational astronomy.
Additional Information
https://blogs.nvidia.com/blog/seti-institute-ai-fast-radio-bursts/
https://www.aanda.org/10.1051/0004-6361/202555217
About the Speaker
Michael Garrett
Michael (Mike) Garrett is the inaugural Sir Bernard Lovell Chair of Astrophysics at the University of Manchester. He is the Chair of the SETI Permanent Committee of the International Academy of Astronautics and is a Visiting Professor at Leiden University. In addition, he is chair of the International Academy of Astronautics (IAA) SETI Committee and serves on the scientific advisory boards of both the SETI Institute and Breakthrough Listen. Previously Mike was Director of Jodrell Bank Centre for Astrophysics, General and Scientific Director of the National Institute for Radio Astronomy in the Netherlands (ASTRON) and Director of the Joint Institute for Very Long Baseline Interferometry (VLBI) in Europe.
Mike developed the technique of wide-field Very Long Baseline Interferometry and spearheaded the roll-out of real-time VLBI (e-VLBI) across the European VLBI Network and beyond. He was instrumental in finalizing the original design concept for the Square Kilometre Array (SKA) telescope. While his scientific interests in astrophysics are eclectic, his main scientific focus currently is in searching for anomalies in large astronomical data sets.
Among other honors and awards received, Mike is an elected Fellow of the Royal Society, and he has served on many international scientific committees and advisory groups.
He earned a BS in Astronomy at the University of Glasgow and a PhD in Radio Astronomy at the University of Manchester.
LinkedIn Profile: https://www.linkedin.com/in/prof-michael-a-garrett/
Adam Thompson
Adam Thompson is Head of Product for Edge Supercomputing at NVIDIA. In that role, he works at the intersection of GPU computing, real-time sensor pipelines, and AI-ready scientific and industrial systems.
Adam’s technical focus is edge and high-performance computing for real-time AI, signal processing, smart sensors, and high-speed sensor I/O connected to GPU-accelerated compute, applications of deep learning to RF data, data compression, and collaborative edge-to-cloud or edge-to-datacenter workflows. More recent work emphasizes digital twins of instruments, online AI training and fine-tuning, and streaming scientific data pipelines.
Adam is best known for helping build GPU-based platforms for real-time scientific and sensor computing and for creating cuSignal, a GPU-accelerated signal-processing library written in Python widely used in sensor-processing communities integrated in CuPy. Recently he has worked to enable real-time AI pipelines for scientific instruments and collaborations linked to radio astronomy and the Allen Telescope Array.
Adam earned a BS in Electrical Engineering at Clemson University and an MS in Electrical and Computer Engineering at the Georgia Institute of Technology.
LinkedIn Profile: https://www.linkedin.com/in/awthomp/