Bowen Wang designed Aftershock Oracle to anticipate the strength, timing, and locations of seismic events following an initial quake. The system relies on decades of tectonic and seismic data compiled by the United States Geological Survey from 1950 to 2025. After testing several machine learning algorithms, he determined that the XGBoost model yielded the most accurate forecasts. Users can enter an earthquake's epicenter, depth, and magnitude into a custom interface to generate these hazard assessments.
Key points
The Aftershock Oracle tool calculates the probability, magnitude, and locations of second…: The Aftershock Oracle tool calculates the probability, magnitude, and locations of secondary quakes based on initial mainshock parameters.
Thirty national finalists were evaluated on scientific rigor, communication skills, and p…: Thirty national finalists were evaluated on scientific rigor, communication skills, and problem-solving capabilities by professional judges.
Wang aims to apply his technical experience toward a future career in civil and structura…: Wang aims to apply his technical experience toward a future career in civil and structural engineering with his own firm.
Sources checked by our analysis pipeline: 1
A 12-year-old student from California has won a top national technology prize. He designed an artificial intelligence tool that can forecast earthquake aftershocks. Bowen Wang is a seventh-grade pupil at Central Middle School in San Carlos. He built a program called Aftershock Oracle. It predicts where secondary quakes will strike, how often they will happen, and how strong they will be after a major earthquake. His research won him the First Place Technology Award and a $3,500 prize at the 2026 Thermo Fisher Scientific Junior Innovators Challenge. The Society for Science organized the contest in Washington, D.C. It is the top national science, technology, engineering, and mathematics (STEM) contest for middle school pupils across the United States.
Bowen lives in Northern California. His family’s closeness to active fault lines inspired his research. “My house is only a few miles away from the San Andreas Fault,” Bowen stated in his competition profile published by the Society for Science. “Over the past year, there have been several earthquakes that woke me up at night.” Scientists have always found main earthquakes very hard…
Led by CEO Pranav Mulgund, the procurement startup aims to tackle longstanding operational bottlenecks in the industrial sector. By deploying artificial intelligence directly across complex manufacturing datasets, the platform helps businesses automate and reconcile problematic purchasing and payment issues. The major capital injection marks the enterprise tech firm's formal public launch out of stealth mode.
Background agent status updates running roughly every 30 seconds repeatedly transmit full transcripts back to the primary model, quietly draining user rate limits and budgets. Enabled by default since August 2026 without an official toggle, the mechanism occasionally costs more than the primary conversation despite having zero influence on ongoing model context. Flamerix AI has released an open-source workaround to disable the recurring timer and eliminate the associated overhead.
A Chinese artificial intelligence model has demonstrated the ability to carry out complex cyberattacks, according to researchers. The operations were conducted within controlled testing environments.