AI Discovers Supernova Without Human Intervention for the First Time

In a significant leap in astronomical research, an international collaboration led by Northwestern University has unveiled a novel artificial intelligence (AI) tool that autonomously discovered, identified, and classified its first supernova.

Named the Bright Transient Survey Bot (BTSbot), this cutting-edge system marks a pivotal shift towards the full automation of supernova detection and analysis, effectively eliminating human intervention and mitigating potential errors.

AI Confirms and Detects a Supernova

Over the past six years, humans have dedicated approximately 2,200 hours to visually inspecting and classifying supernova candidates. With the BTSbot now in operation, this invaluable time can be redirected towards accelerating the pace of scientific discovery.

Adam Miller, leading the effort at Northwestern, emphasized the historic nature of this achievement, stating that a series of robots and AI algorithms have successfully identified and confirmed a supernova, a feat never before accomplished.

This development signals a crucial step towards refining models, enabling robots to discern specific subtypes of stellar explosions.

"This represents an important step forward as further refinement of models will allow the robots to isolate specific subtypes of stellar explosions. Ultimately, removing humans from the loop provides more time for the research team to analyze their observations and develop new hypotheses to explain the origin of the cosmic explosions that we observe," Miller said in a press statement.

Nabeel Rehemtulla, co-leader of the technology development, noted that this achievement streamlines extensive studies of supernovae, deepening our understanding of star life cycles and the genesis of elements produced by supernovae, including carbon, iron, and gold.

Bypassing Manual Step

Currently, humans collaborate with robotic systems to detect and analyze supernovae. Robotic telescopes survey sections of the night sky, searching for new sources.

When a new source is detected, humans take over, using automated software to verify and observe candidates, particularly obtaining spectroscopic observations to confirm a supernova.

The BTSbot was engineered to bypass this manual step. Trained on a machine-learning algorithm with over 1.4 million historical images from a wide range of sources, including confirmed supernovae, flaring stars, variable stars, and galaxies, the BTSbot is designed to autonomously identify and classify supernova candidates.

In a test scenario, the BTSbot identified a newly discovered supernova candidate named SN2023tyk, detected by the Zwicky Transient Facility (ZTF) on October 3.

The BTSbot promptly requested the potential supernova's spectrum from the Palomar Observatory, where another robotic telescope (SED Machine) conducted comprehensive observations.

The resulting spectrum was sent to Caltech's SNIascore, which determined the supernova's type. After confirming it as a Type Ia supernova, characterized by the complete explosion of a white dwarf in a binary star system, the BTSbot publicly disclosed the discovery to the astronomical community.

The successful deployment of BTSbot exemplifies a momentous achievement in automated supernova detection, paving the way for more efficient and comprehensive exploration of cosmic phenomena.