Science HolicA Glimpse of the world through a Binary lense: AI and Machine Learning’s Role in Shaping Contemporar
Author: Ted Isidor
Editors: Ken Saito and Liane Xu
Artists: Doris Tan
The high-tech civilization we live in today would be possible without the contributions of great men and women of the past, such as Albert Einstein and Isaac Newton. Although our human minds have uncovered marvelous ideas and theories in the world of physics, physicists are turning towards the digitized world of programming to continue our hunt for knowledge and understanding.
According to a Forbes article, Artificial Intelligence (AI) is essentially a machine that can complete tasks in a way that we deem smart. Tesla cars, for example, utilize AI to drive along a given route in a way that is smart (abiding by road rules). Machine learning is when AI is used to give programs data and “learn” from it. Imagine a baby trying to learn how to walk. They take information like how their legs work, the floor, etc., and try multiple times to figure out how to get from one place to another. After many failed attempts, a baby can finally walk as they learn what they did wrong before and see a better way of doing it. In the ever-changing realm of science as a whole, machine learning has many perks, such as its ability to essentially run its own experiment. Recently, a computer program called MELVIN was able to solve an issue involving complex entangled states and photons, which physicists had struggled with for a long time. One major reason for integrating machine learning and AI as a whole in physics is the large quantity of data that physicists have to deal with, which can take days, weeks, and possibly months to sort out. Now, instead of pesky data and long waits on experimental results, physicists and the science community as a whole are given a chance to speed up the process, with only their creativity in creating programs being the only barrier to greatness.
It would be rather unfitting to discuss the applications of Artificial Intelligence in physics without noting the current drawbacks in its effectiveness. AI suffers from “quantifying uncertainties,” as stated in an article by Physics World. This means that some AI programs have issues understanding how they are not entirely sure if they are correctly labeling certain things. For instance, imagine a very bored data scientist creating an AI program that is able to identify and classify which animals are dogs and cats. On certain occasions, this program might fail and accidentally label a dog as a cat, for example, because it closely resembles a cat. Without knowing how many cases of dogs are being labeled as cats, the data scientist will not know that there is a severe issue in their code. For physicists, these mistakes can ruin an entire study, and when dealing with billion-dollar particle accelerators, this is quite a pain. Luckily, physicists are quite the experts when it comes to quantifying uncertainties, and as a result, more physicists are collaborating with the AI community to help fix this issue in AI programs. Another issue when it comes to the integration of AI in physics is the information that it provides. You see, in the same article mentioned earlier, the author also discusses how the information that AI can teach us about physics, such as how to turn 100% energy into mass, can impact the world as seen in the past. A particularly memorable one was the atomic bombs used in Hiroshima and Nagasaki.
With all this, now is a good time to discuss the future of AI in physics. Max Tegmark, a physicist who is known for discussing AI and physics, stated in an interview with Lex Fridman that AI can help distill complex physics in the sense that it helps make the results it calculates readable and human friendly. With this remark, it can be inferred that in the future, AI is highly likely to continue to be used by physicists to discover more and help us understand it. Technology has always been a tool for humanity since the dawn of its creation, yet through Artificial Intelligence and Machine Learning, our revolutionary asset has only just begun.
Citations:
Ananthaswamy, Anil. “AI Designs Quantum Physics Experiments beyond What Any Human Has Conceived.” LiveScience, Purch, 14 July 2021,
www.livescience.com/ai-designs-quantum-physics-experiments.html.
Commissariat, Tushna. “AI and Particle Physics: a Powerful Partnership.” Physics World, Physics World, 13 May 2021, physicsworld.com/a/ai-and-particle-physics-a-powerful-partnership/.
Falk, Dan, and substantive Quanta Magazine moderates comments to facilitate an informed. “How Artificial Intelligence Is Changing Science.” Quanta Magazine, Quanta Magazine, 21 Jan. 2021,
www.quantamagazine.org/how-artificial-intelligence-is-changing-science-20190311/.
Fridman, Lex, director. Max Tegmark: AI and Physics | Lex Fridman Podcast #155. Max Tegmark: AI and Physics | Lex Fridman Podcast , YouTube, 17 Jan. 2021,
www.youtube.com/watch?v=RL4j4KPwNGM.
Marr, Bernard. “What Is The Difference Between Artificial Intelligence And Machine Learning?” Forbes, Forbes Magazine, 13 May 2019,
www.forbes.com/sites/bernardmarr/2016/12/06/what-is-the-difference-between-artificial-intelligence-and-machine-learning/?sh=7b21fdb92742.