AI Tracks a Star’s Final Moments

Astronomers have captured the violent death of a massive star locked in a deadly orbit with a black hole. This breakthrough was made possible by an artificial intelligence system developed by a team at the University of California, Santa Cruz, which specializes in monitoring stars immediately after they explode as supernovae.

Figure 1. AI Detects Rare Supernova That Shined Twice.

In July 2023, the event—designated SN 2023zkd—was detected by this AI-powered system, which scans the sky in real time for unusual stellar explosions. The rapid alert allowed researchers to begin detailed follow-up observations almost immediately, a critical step in reconstructing the full sequence of the star’s catastrophic demise. Figure 1 shows AI Detects Rare Supernova That Shined Twice.

Unmatched Observations from Multiple Telescopes

By the conclusion of the event, SN 2023zkd had been tracked by a wide network of telescopes, both on Earth and in space. This included two instruments at Hawaiʻi’s Haleakalā Observatory, which are part of the Young Supernova Experiment (YSE) led by UC Santa Cruz.

“Something exactly like this supernova has not been seen before, so it may be extremely rare,” said Ryan Foley, associate professor of astronomy and astrophysics at UC Santa Cruz. “Humans are reasonably good at spotting things that ‘aren’t like the others,’ but AI can flag unusual events earlier than a human might notice. This early detection is critical for time-sensitive observations.”

Surveying the Skies for Supernovae

Foley’s team runs the Young Supernova Experiment (YSE), which scans a region of the sky roughly 6,000 times the size of the full moon—about 4% of the night sky—every three days. The survey has uncovered thousands of new cosmic explosions and other astrophysical transients, often within days or even hours of their initial eruption.

The scientists interpreting SN 2023zkd suggest that a collision between the massive star and its black hole companion was inevitable. As energy was lost from the orbit, the star moved closer to the black hole until gravitational stress triggered the supernova. The discovery was published on August 13 in The Astrophysical Journal. “Our analysis shows the blast was caused by a catastrophic encounter with a black hole companion, providing the strongest evidence yet that such close interactions can actually detonate a star,” said lead author Alexander Gagliano, a fellow at the NSF Institute for Artificial Intelligence and Fundamental Interactions.

An alternative hypothesis is that the black hole tore the star apart before it could explode naturally. In this case, the debris from the star collided with surrounding gas, producing the bright light observed. Either scenario leaves behind a single, heavier black hole.

A Supernova That Glowed Twice

Located about 730 million light-years from Earth, SN 2023zkd initially appeared to be a typical supernova, producing a single burst of light. However, over several months, it unexpectedly brightened again. Archival data revealed that the system had been gradually brightening for more than four years before the explosion—a rare prelude for supernovae.

Detailed analysis at UC Santa Cruz showed that the supernova’s light was shaped by material the star had shed in its final years. The initial brightening came from the blast wave hitting low-density gas, while a second, delayed peak resulted from a slower collision with a dense, disk-like cloud. The star’s erratic pre-explosion behavior suggests it was under extreme gravitational stress from a nearby compact companion, likely a black hole.

From Skepticism to Confirmation

Foley recalled that the idea of a black hole triggering a supernova initially sounded almost like science fiction. Through careful analysis of spectral data and discussions with Gagliano, the team confirmed the binary system interpretation. Gagliano led the effort on spectral analysis, while Foley served as a spectroscopy expert and critical sounding board.

Power of AI and Collaboration

“Our team built the software platform that consolidates data and manages observations, integrating AI tools for studies like this,” Foley said. The research collaboration combined diverse expertise, including theory (led by Enrico Ramirez-Ruiz at UC Santa Cruz) and AI analysis (V. Ashley Villar at Harvard). The study was conducted under YSE, coordinated with the Center for Astrophysics | Harvard & Smithsonian and MIT, and funded by NSF, NASA, the Moore Foundation, and the Packard Foundation.

Funding Uncertainty Threatens Future Discoveries

Foley expressed concern over uncertain funding, which is forcing the collaboration to take fewer risks and reducing overall scientific output. “This uncertainty is shrinking our program, limiting graduate admissions, and pushing students to leave the field or work outside the U.S.,” he said.

Despite these challenges, Foley emphasized the broader potential of AI-driven anomaly detection. “Techniques like this could be applied to screening for diseases, detecting security threats, monitoring mental health, or preventing financial fraud. Anywhere real-time anomaly detection is needed; these methods will likely play a role.”

Source: SciTECHDaily

Cite this article:

Priyadharshini S (2025), AI Spots Rare Supernova That Flared Twice in a Lifetime, AnaTechMaz, pp.490