AI algorithms use similar processes to identify and classify celestial objects from vast amounts of observational data. Let’s imagine you’re a space detective, and the Universe has sent you a mystery file full of unknown objects. Some are stars, some are galaxies, but they all look pretty similar from here on Earth. Could you figure out which is which? And could you teach a computer to do it for you?

Why do we care?
We study stars and galaxies because they are the building blocks of the Universe. Stars forge nearly all the elements that make up planets and (our) life itself, while galaxies show us how the universe is organized on the largest scales. By understanding how stars form, evolve, and die—and how galaxies grow and interact—we can trace the history of the cosmos and our place within it.

Modern telescopes collect staggering amounts of data about stars and galaxies, far more than humans can examine alone. AI tools allow us to sift through this flood of information, spot patterns we might miss, and make new discoveries faster. By combining human insight with AI’s speed, we can unlock hidden stories of the cosmos and deepen our connection to the Universe we all live in.

The Science of Stars and Galaxies
Our Milky Way galaxy contains several hundred billion stars of various ages, sizes and masses. A star forms when a dense cloud of gas collapses due to gravity until nuclear reactions begin deep in the interior of the cloud and provide enough energy to stop the collapse.

Many factors influence a star’s life and final form, but the most important is its initial mass. A star with a mass similar to the Sun’s will become a white dwarf. A more massive star may explode as a supernova, leaving a neutron star.

Galaxies are gravitationally-bound systems of stars, gas, dust and dark matter. A typical large spiral galaxy such as our home galaxy, the Milky Way, consists of hundreds of billions of stars, enough gas, and dusto make billions more stars, and at least ten times as much dark matter as all the stars, gas and dust put together. This type of galaxy is identified by its distinctive spiral shape, with visible “arms” of dust and gas orbiting around the center.

An elliptical galaxy has stars distributed in an elliptical shape, ranging from elongated to nearly circular in appearance. The light is smooth, with the apparent brightness decreasing as you go farther out from the center.

An irregular galaxy is a strangely shaped galaxy, often rich in interstellar matter, but not a member of either of the major classes of spiral or elliptical galaxies.

Introduction Activity
on How AI is Trained

Imagine you don’t know what Skittles are. Your friend wants Skittles, and you have a pile of Halloween candy to look through. You lift up pieces of candy one by one, and your friend tell you whether each one is Skittles or not. After going through a chunk of the pile, you start being able to tell when you’re holding Skittles.

Even after getting trained, sometimes you still make mistakes. Sometimes you pick up a pack of M&Ms, and believe they’re Skittles. This is how AI learns.

Materials:
Images of different celestial objects printed & cut out (one facilitator’s version, and as many of the learners’ version as needed).

Grades: 5-8

Learners will:
• Classify the images into categories based on their features

• Discuss how they made their classification decisions, highlighting the visual cues and patterns they used.

• Explain that AI algorithms, like machine learning, use similar processes to identify and classify celestial objects from vast amounts of observational data.

Get materials ready:
Print the included set of real images from public archives (Chandra X-ray Observatory, Hubble telescope, James Webb telescope and others) that include a mix of stars and galaxies. Print labeled images for your own use as an answer key. Finally, print the observation worksheet for participants to fill out as they sort (example at right; downloads below.)

Facilitate the Classification Challenge
• Spread out the printed images (or provide a digital grid).

• Ask learners to sort the images into two piles: Stars vs Galaxies.

• Encourage users to write or talk through their reasoning: What made them choose one category over the other? Have them think one step further — if they were designing an AI model, what attributes of stars/galaxies would they provide as rules?

• Possible traits used for classification: “roundness”, point-like vs extended, diffraction spikes, color, visible structure. Some images might contain both galaxies and stars of course, but the object is to identify the primary case.

STEAM extension: Provide each learner with poster paper, and with their own set of star/galaxy images. After the sorting activity, have them arrange and glue all their star images to create their very own "galaxy" or all their galaxy images to create their very own "galaxy cluster" — or both. Have them name their galaxy/galaxy cluster, and share with the group. For further extension, learners can write a short story, poem or other piece about their cosmic creation.

download FACILITATOR IMAGE CARDS (Printable Download)

download WORKSHEET (Printable Download)

download LEARNER IMAGE CARDS (Printable Download)

download FULL ACTIVITY PACKET (Printable Download)