In the search for knowledge, answers to multiple questions have been sought, among them, the composition of the physical and chemical world that surrounds us, with great advances produced by the work of the Nobel Prize winner, Marie Curie, and in the theoretical field, the theory of everything has also been sought, proposed by physicists such as Albert Einstein (Mann, 2019) and at the time, Stephen Hawking.

In these advances, physics has faced the challenge of performing complex mathematical calculations that allow the creation of hypotheses and their subsequent experimentation in laboratories, accompanying the human race in the understanding of the greatest question of all: where do we come from and where are we going?

In the face of these challenges, physicists are increasingly training and using Artificial Intelligence (AI) and machine learning techniques to advance our understanding of the physical world, but there is a growing concern: bias in the profession, systems and understanding of the broader impact on society.

In this regard, the journal Physics World, in its May 2021 issue, published the work of Dr. Julianna Photopoulos, where she explores the problems of racial and gender bias in AI and what physicists can do to recognize and address the problem (Durrani, 2021), with the goal of making technology-enhanced physics a fairer, more inclusive and intelligent science.

In the latest work in physics, for example in the understanding of muons, which originate from cosmic rays and have unlocked the secrets of large structures such as the Giza pyramids of Egypt (Research and Science, n.d.), AI allows us to point to new physics: a measurement of the magnetic moment of the muon is at odds with the standard model, potentially hinting at new forces or particles, bringing us ever closer to understanding the origin of the universe and the composition of matter.

While this area of AI application in physics seems to have no ethical connotation, it is important to mention that more and more of this type of experimentation is being transferred to medical physics, and materials design, which will eventually be used by society at large and where deep problems remain hidden:

  • The lack of gender and racial diversity that exists in physics affects both the work being done and the systems being created.
  • AI systems are dependent on the humans who program it, so the recognition of individual biases is still little explored terrain, thus causing their replication and amplification.
  • There is not enough education for physicists, regarding the ethics and risks involved in dealing with data extracted from human beings. By computing large amounts of data, the line between the human-social factor and the digital is blurred.

New research work in science and physics, for example to improve accuracy and reliability in AI, led by Dr. Payel Das, senior manager of research staff at IBM’s Thomas J Watson Research Center, work in recognition of the biases and risks of AI thus increasing precision, reliability and explainability.

An application of AI that has the potential to shed light on physics, and that by understanding the biases that could be part of the calculations, points to the construction of a future that learns from the past for a better future in the field of science, innovation and in general, humanity, so that addressing potential biases is an obligation for all organizations and professionals involved in its development and implementation.

Undoubtedly, 2021 will be the year in which we will begin to see great advances in physics and the ethical, responsible and transparent use of AI, and it is up to us, as a society, to participate in the explainability of these systems, demanding the identification and elimination of biases that could divert us from the common good, for all.

––––––––––––––––––––––––––––––––––––––––––––––––––––––––

References

Durrani, M. (2021, 4 mayo). Artificial intelligence: towards better, smarter and fairer physics. Physics World. https://physicsworld.com/a/artificial-intelligence-towards-better-smarter-and-fairer-physics/

Mann, A. (2019, 29 agosto). What Is the Theory of Everything? Space.Com.

https://www.space.com/theory-of-everything-definition.html

Muones: las partículas poco conocidas que permiten sondear lo impenetrable. (s. f.). Investigación y Ciencia. https://www.investigacionyciencia.es/

Photopoulos, J. (2021, 28 mayo). Fighting algorithmic bias in artificial intelligence. Physics World. https://physicsworld.com/a/fighting-algorithmic-bias-in-artificial-intelligence/