Recently one of my fellow members of the Comal County Friends of the Night Sky asked how Artificial Intelligence (AI) might be used in Astronomy. AI/ChatGPT is a hot topic these days so OK, let’s see.
Astronomy is a multi-faceted discipline incorporating physics, art, cosmology, even philosophy. These facets each are composed of sub-facets and specialties. The one discipline that is common to all is mathematics.
Early astronomers observed the natural structure and dynamics of celestial body movement. This was painstaking, tedious work, documenting the position and apparent motion of everything that moved up there. They were collecting data! Then, using this data to construct the motion of the Sun, Moon, and planets for example, a picture of how our solar system worked started to emerge. From these data came the calendar, and a way to predict seasons. Early astronomers also predicted future outcomes by the position and movement of heavenly bodies. Yes, they were astrologers too. Early cultures associated the heavens with various gods and lived according to predictions. Funny thing, astrology uses some complex math formulas to arrive at a prediction or explanation.
Data collection and mathematics, that’s the ticket, eh? However, as history shows, that ticket doesn’t necessarily get you to the correct destination. We’ve had many fits and starts in our quest for understanding this universe and that’s where philosophy comes into play. An astronomer’s philosophic leanings can and have affected their use of mathematics on data. Tycho Brahe and earlier astronomers saw our universe as Earth centered. They then created sophisticated mathematical models of it to conform. Albert Einstein would not accept his own work showing our universe is expanding. He created a “fudge factor” to make the universe fit his static universe beliefs.
It can be argued that AI in astronomy came about in the 1990s to process the terabytes of data produced by digital sky surveys. Such mountains of data have necessitated creating algorithms (mathematics) to analyze the data. These algorithms were astronomers DIY AI, specifically developed to look for certain patterns, signals, etc. in the data. This form of AI is specialized however, and generally not able to adjust itself to changing data patterns.
Today’s AI is a different machine. It’s a little scary, as it can and will adapt to changes, like a human. AI is a learning machine, and it remembers way better than any of us. AI doesn’t just analyze, it will suggest things, it will see anomalies, it will make corrections.
It will argue.
That’s a little scary too, but it might point out a philosophic error in logic, perhaps preventing astronomers from going down the dreaded data “rabbit hole”.
All of this should be a boon for processing the big astronomical data produced. The dark side is…might AI put some astronomers out of their jobs? That’s a fear.
What’s in the Sky?
June 3: Full, Strawberry Moon.
June 4; before sunrise; east: Jupiter, with Mercury and Uranus to its lower left hang low in the sky. Use binoculars if you have them.