Written by Kamayani R.
Edited by Romit C.
Machine learning is a field of study that has impacted almost every area of human life but can the algorithms of Machine Learning and Artificial intelligence help in solving the mysteries of physics?
Developments in human understanding of science are indeed based on formulating a set of mathematical equations which govern the natural phenomenon. Measurements have played a significant role in postulating theories and laws to understand nature. For example, through analysis of astronomical data and empirical observations of planetary motion (Riebeek, 2009), the Ptolemaic theory of motion was developed. This theory, followed later by Kepler's laws based on the elliptical motion of planets in a heliocentric system, contributed to the development of Newton's second law of motion, F = ma (Lucas, 2017). This universal and simple equation accurately explains physical dynamics. Many technological and scientific advancements have been made through such parsimonious models (simple models which can be understood and interpreted easily).
The fact that is largely unacknowledged in such developments is the intuitive leap that one has to take besides analyzing the necessary data. For example, today, we are aware that all objects, irrespective of their masses, fall with the same acceleration towards the center of the earth, a theory that was put forward by Galileo (Jones, 2019). Earlier, people believed that "heavier object[s] would want to fall more'" (Barth, 2020), in accordance with Aristotle's theory of gravity. Suppose one wishes to check the applicability of both theories by throwing two bodies from a height. In that case, one will come to a conclusion that Aristotle was right, ignoring the effect of air drag on the motion of freely falling objects. By just analyzing a given set of data and deducing physical laws, we wouldn't be able to unravel even the fundamental laws of nature.
Physics is based on human understanding of nature and the universe. Understanding the process behind scientific phenomena would rather be more helpful in unraveling the mysteries of the cosmos than just deducing a set of mathematical equations from data. Scientific understanding requires the analysis of data with intuition and intense experimentation. Developments in machine learning and artificial intelligence would help in analyzing data and fastening the process of scientific developments. Still, only data-driven laws by themselves rarely help in the advancement of physics.
References:
Barth, D. (2020). 6.3: Galileo's Falling Bodies. Libretexts. https://phys.libretexts.org/Bookshelves/Astronomy__Cosmology/Astronomy_for_Educators_(Barth)/06%3A_Exploring_Gravity/6.03%3A_Galileos_Falling_Bodies
Jones, A. (2019). The History of Gravity. Thoughtco. https://www.thoughtco.com/the-history-of-gravity-2698883
Lucas, J. (2017). Force, Mass & Acceleration: Newton's Second Law of Motion. Livescience. https://www.livescience.com/46560-newton-second-law.html
Riebeek, H. (2009). Planetary Motion: The History of an Idea That Launched the Scientific Revolution. Earthobservatory. https://earthobservatory.nasa.gov/features/OrbitsHistory
Comentarios