Description
Astronomy research has seen a near exponential growth in artificial intelligence being used over the past four decades. In this presentation, we will present a review on what AI is being used for and doing for astronomy. We build a set of MNRAS publications from 2023 that contains artificial intelligence in the full text of the publication for our sample. From this, we analyse the titles and abstract to determine (a) what sub-domain of astronomy is being investigated; (b) what algorithm/s are being utilised; and (c) what the goal of those algorithms are. We find that galaxy morphology (>20%), cosmological simulations (>15%), and stellar astrophysics (>13%) are the most common sub-domains where artificial intelligence is being applied. Although neural networks and deep learning feature heavily among the algorithms mentioned, the most popular is simply "machine learning" within the abstracts (17%). We suggest to authors to be more explicit in their abstracts about what methods are being applied. Finally we discuss what these approaches are being used for and find that making new predictions (>20%), object detection (>15%), classification of galaxies (>15%), and parameter estimations (>13%) are the most frequent goals, although there are many others which we discuss. Given the advent of new surveys such as Euclid & the Square Kilometre Array, we assume the application of AI to the field will only increase further with time until these algorithms embed themselves as tools available for all to use.
