Level 1 — Absolute Beginner
Scientists found more than 100 new planets in space. They used a computer program called RAVEN to find them. RAVEN is a type of artificial intelligence. It can look at data very fast and find things that people cannot see.
The planets were hiding in old data from NASA. NASA has two space telescopes called Kepler and TESS. These telescopes watch stars and look for small dips in starlight. A dip happens when a planet moves in front of a star. This is called a transit.
Human scientists looked at the same data before, but they did not find these planets. The dips in starlight were too small or too hard to see. RAVEN studied the light curves and found patterns that humans missed. A light curve is a chart that shows how bright a star is over time.
Some of the new planets are in very interesting places. A few are in the habitable zone, which is the area around a star where water can exist. This discovery shows that artificial intelligence can help us learn more about the universe.
- planets
- Large round objects in space that move around a star.
- artificial intelligence
- A computer program that can think and learn like a person.
- data
- Information collected by computers or machines.
- telescopes
- Tools that help people see things that are very far away.
- starlight
- The light that comes from a star.
- transit
- When a planet moves in front of a star and blocks some light.
- light curves
- Charts that show how bright a star is over time.
- habitable zone
- The area around a star where water could exist on a planet.
Level 2 — Elementary
A team of astronomers has used an artificial intelligence tool called RAVEN to discover more than 100 hidden planets in existing NASA data. RAVEN stands for Rapid Automated Vetting of Exoplanet candidates using Neural networks. The tool analyzed information that had already been collected by NASA's Kepler and TESS space telescopes.
Kepler and TESS search for exoplanets by watching thousands of stars at once. When a planet passes in front of its star, it causes a tiny dip in the star's brightness. Scientists call this the transit method. RAVEN was designed to detect these dips by studying light curves, which are detailed graphs of a star's brightness over time.
What makes this discovery remarkable is that human astronomers had already reviewed the same data. They missed these planets because the signals were too faint or too irregular to notice. RAVEN's neural network was able to recognize subtle patterns in the data that the human eye could not detect, proving that AI can find what people overlook.
Among the newly discovered planets, several orbit in the habitable zone of their stars, where conditions might allow liquid water to exist. Others follow unusual orbital paths that surprised researchers. This breakthrough demonstrates how artificial intelligence can accelerate the pace of scientific discovery by processing massive datasets far more efficiently than humans.
- astronomers
- Scientists who study stars, planets, and space.
- exoplanets
- Planets that orbit stars outside our solar system.
- neural networks
- Computer systems designed to learn and find patterns, inspired by the human brain.
- brightness
- How much light something gives off; how bright it is.
- transit method
- A way to find planets by watching for dips in a star's light.
- detect
- To find or notice something that is hard to see.
- faint
- Very weak or hard to see or hear.
- subtle
- Not obvious; difficult to notice.
- orbital
- Related to the path a planet takes around a star.
- datasets
- Large collections of organized information.
Level 3 — Intermediate
In a remarkable feat of computational discovery, astronomers have deployed an artificial intelligence tool called RAVEN — Rapid Automated Vetting of Exoplanet candidates using Neural networks — to uncover more than 100 previously hidden planets lurking in existing NASA data. The tool sifted through years of observations collected by NASA's Kepler and TESS space telescopes, missions originally designed to survey the sky for exoplanets using the transit method.
The transit method works by monitoring thousands of stars simultaneously and flagging tiny dips in their brightness that occur when a planet crosses in front of its host star. These periodic dips create a signature pattern in the star's light curve, a detailed record of brightness over time. However, not every signal is straightforward. Background noise, instrument glitches, and overlapping stellar signals can obscure genuine planetary transits, making them extremely challenging for human reviewers to confirm.
RAVEN's deep learning neural network was trained on thousands of confirmed and false planetary signals, enabling it to distinguish real transits from false positives with remarkable precision. When applied to archival Kepler and TESS data, the algorithm identified over 100 credible planet candidates that had eluded human vetting teams. Many of these missed detections involved planets with faint signals, long orbital periods, or configurations that produced irregular transit patterns.
The implications of this discovery extend well beyond the planet count itself. Several of the newly confirmed worlds reside in the habitable zone of their respective stars, where surface temperatures could theoretically permit liquid water. Others occupy eccentric orbits or exist in multi-planet systems with fascinating gravitational dynamics. This achievement underscores a broader trend in modern astrophysics: artificial intelligence is no longer merely an assistant to human researchers but an indispensable tool capable of extracting insights from data at a scale and depth that human analysis alone cannot match.
- computational
- Related to using computers to process information or solve problems.
- deployed
- Put into active use for a particular purpose.
- sifted
- Carefully examined large amounts of information to find what is useful or important.
- survey
- To systematically observe and record data about a large area or group.
- simultaneously
- At the same time; happening together.
- periodic
- Happening at regular intervals of time.
- obscure
- To make something difficult to see, understand, or detect.
- false positives
- Results that incorrectly indicate something is present when it is not.
- archival
- Relating to stored records or data from the past.
Level 4 — Advanced
In what may constitute one of the most consequential applications of machine learning in observational astronomy, a research team has leveraged an artificial intelligence framework called RAVEN — Rapid Automated Vetting of Exoplanet candidates using Neural networks — to unearth more than 100 previously undetected exoplanets concealed within archival data from NASA's Kepler and TESS missions. These space telescopes, which have collectively monitored the luminosity of millions of stars over the past decade, employ the transit photometry method: detecting the infinitesimal dimming that occurs when an orbiting planet traverses the face of its host star.
Despite the extraordinary volume of data amassed by these missions, the process of vetting potential planetary candidates has long been a bottleneck. Each transit signal must be scrutinized against a litany of astrophysical false positives — eclipsing binary stars, instrumental artifacts, and systematic noise — before a candidate can be elevated to confirmed planet status. Human reviewers, constrained by cognitive limitations and the sheer magnitude of the data, inevitably overlook signals that fall below their perceptual threshold, particularly those with low signal-to-noise ratios or anomalous periodicity.
RAVEN's deep learning architecture, a convolutional neural network meticulously trained on thousands of labeled transit signals, circumvents these human limitations. By ingesting raw light curve data and autonomously extracting hierarchical features — from broad flux variations down to microscopic transit signatures — the algorithm can disambiguate genuine planetary transits from spurious signals with a classification accuracy that surpasses traditional vetting pipelines. When retrospectively applied to the Kepler and TESS archives, RAVEN flagged over 100 viable exoplanet candidates that had slipped through every preceding round of human and algorithmic review.
The ramifications of this discovery are profound and multifaceted. Among the newly catalogued worlds, a notable subset resides within the circumstellar habitable zone, where equilibrium temperatures could sustain liquid water on a rocky surface. Others exhibit highly eccentric orbits or participate in resonant multi-planet configurations that offer invaluable laboratories for testing theories of planetary formation and gravitational dynamics. More broadly, RAVEN's success epitomizes a paradigm shift in astrophysics: the transition from hypothesis-driven inquiry to data-driven exploration, where machine intelligence systematically interrogates datasets of unprecedented scale to surface discoveries that would otherwise remain perpetually obscured from human cognition.
- consequential
- Having significant and far-reaching importance or effects.
- leveraged
- Used something to maximum advantage to achieve a result.
- luminosity
- The total amount of light emitted by a star or celestial object.