Whether we like it or not, artificial intelligence will change the way we interact with the universe. Astronomy as a science has a long tradition of finding patterns by sifting through vast amounts of data, serendipitous discoveries, and deep connections between theory and observation. These are all areas where AI systems can make astronomy faster and more powerful than ever before.

However, it is important to note that “artificial intelligence” is a very broad term that covers a wide range of semi-connected software tools and techniques. Astronomers often resort to neural networks, in which software learns all the connections in a training data set and then applies the information about those connections to a real data set.

Take data processing for example. The beautiful images posted online from the Hubble Space Telescope or James Webb Space Telescope are far from the first time these instruments have made a pass over this particular patch of sky.

Raw astronomical images are full of errors, dirty foregrounds, contamination, artifacts, and noise. Processing and cleaning these images into something presentable, let alone useful for scientific research, requires a large amount of input, often done partly manually and partly by automated systems.

Astronomers are increasingly turning to artificial intelligence to process data and remove unnecessary parts of images to produce a clean result. For example, the image of the supermassive black hole at the center of the Messier 87 (M87) galaxy, first published in 2019, was updated in April 2023 using machine learning, revealing a much clearer picture of the structure of the black hole.

In another example, some astronomers will feed images of galaxies into a neural network algorithm and report to the algorithm a classification scheme for the detected galaxies. Current classifications were created manually, either by the researchers themselves or through the scientific efforts of volunteers. Once trained, the neutral network can be applied to real data and automatically classify galaxies; This is a much faster and less error-prone process than manual classification.

Astronomers can also use artificial intelligence to remove optical noise created by Earth’s atmosphere from space images taken by ground-based telescopes.

AI is even proposed to help us detect signs of life on Mars, understand why the Sun’s corona is so hot, or determine the age of stars.

Astronomers are also using neural networks to delve deeper into the universe than ever before. Cosmologists are beginning to use artificial intelligence to understand the fundamental nature of the universe. Two of the greatest cosmic mysteries are the identity of dark matter and dark energy, which are beyond our current knowledge of physics, accounting for more than 95% of all energy in the universe.

To help identify these strange substances, cosmologists are now trying to measure their properties: how much dark matter and dark energy there are, and how they have changed over the history of the universe. Small changes in the properties of dark matter and dark energy have profound effects on the ultimate history of the cosmos, affecting everything from the location of galaxies to the rate of star formation in galaxies like the Milky Way.

Neural networks help cosmologists distinguish many effects of dark matter and dark energy. In this case, the training data comes from complex computer simulations. In these simulations, cosmologists change the properties of dark matter and dark energy and see what happens. They then feed these results into a neural network, so the network can discover all the interesting ways the universe is changing. Although cosmologists aren’t quite ready for prime time, we hope that cosmologists will then be able to train the neural network to make real observations and tell us what the universe is made of.

Such approaches are becoming increasingly critical as modern astronomical observatories produce enormous amounts of data. The Vera K. Rubin Observatory, a state-of-the-art facility being built in Chile, will be tasked with providing more than 60 petabytes (where one petabyte equals a thousand terabytes) of raw data in the form of high-resolution images. of the sky. Analyzing such a large amount of data is beyond the power of even the most determined graduate student. This task can only be accomplished by computers with the help of artificial intelligence.

The search for the unexpected will be of particular interest to future observatories. For example, astronomer William Herschel accidentally discovered the planet Uranus while doing a routine survey of the night sky. AI can be used to flag and report objects of potential interest, identifying anything that doesn’t fit an established pattern. And in fact, astronomers have already used artificial intelligence to detect a potentially hazardous asteroid using an algorithm developed specifically for the Vera C. Rubin Observatory. Who knows what discoveries we will incorporate into the machine in the future? Source