During Friday (November 11, 2016)'s lecture, the idea of cargo cult science was introduced. The idea itself is very interesting and agreeable: directly replicating other people’s method without understanding it will not obtain the same results. However, the main example of this idea that most people seem to know about when asked, is very poor.
This example can be spotted in the speech given by Feynman in 1974, which is quoted in multiple different articles and books [1]. Feynman stated: “… During the war they saw airplanes land with lots of good materials, and they want the same thing to happen now. So they… make things like runways … put fires along the sides … wooden hut … two wooden pieces on his head like headphones and bars of bamboo sticking out like antennas…” [2], If I may paraphrase that, we can consider Feynman’s claim to be: “The indigenous islanders replicated the setup of an airplane runway in order to obtain goods.” However, similar to how we cannot simply make an assumption for intentionality, and claim something like “The coffeemaker was designed for safety because it uses a ‛light up’ switch” [3], Feynman’s claim lacks evidence to be backed up. Not at any point of the speech did he even mention evidences of any sort that proves the islanders’ true intention to be obtaining goods. As a matter of fact, it is quite possible that the islanders are imitating the outside people for other reasons. Research has been conducted to show that it is part of human nature to imitate other people [4], for example, “People playing the schoolyard game rock-paper-scissors are far more likely to match their opponent's gesture when they can see each other.” This clearly shows that there is a lot more than just one reason why the islanders could have replicated the airplane runway. We can even read deeper into Feynman’s claim and easily interpret: “the indigenous islanders are not very intelligent as they believe what they are doing have no difference in what the outsiders are doing.” This is simply not true as more and more researches has shown that isolated indigenous people can potentially be very intelligent [5]. For example, the Sentinelese people located in the Indian ocean had no problem surviving the 2004 tsunami on their own, which might not have been possible if there aren’t intelligent people among them. Finally, if we take a look at the most famous picture of Cargo Cult Science and compare it to a modern rocket launch, who are we to judge what the people in the upper picture are doing, as we ourselves do the exact same thing. In the end, even though the main idea of Cargo Cult Science is mostly agreeable and true, the primary example that most people seems to know is not backed up with sufficient evidence and should be considered very poor. [1] Cargo Cult Software Engineering page 11 [2] The article published by Richard P Feynman: http://calteches.library.caltech.edu/51/2/CargoCult.pdf [3] ESC101 Lecture 05: https://design.engsci.utoronto.ca/courses/esc101/20169/ESC101%2020169%20Lecture%2005%20Actual.pdf [4] As part of human nature, people like to imitate others http://www.dailymail.co.uk/sciencetech/article-2016708/Born-copycats-Why-just-fight-subconscious-impulse-imitate-others.html [5] Sentinelese had survived the 2004 tsunami. http://www.survivalinternational.org/campaigns/mostisolated
0 Comments
During the Engineering Science Education Conference that took place on January 20th, the idea of machine learning was introduced and elaborated. The concept is incredibly eye-opening and clearly demonstrates how much technologies have advanced for the past decade. However, being powerful as it is, the machine learning technology is unlikely to predict human behaviors, which many have believed to be true [1].
Machine learning is “a method of data analysis … Using algorithms that iteratively learn from data… allows computers to find hidden insights without being explicitly programmed where to look.” [2] Essentially, machine learns what to do by analyzing a large amount of data and imitating the successful attempts. A great example shown by the lecturer was one of her research projects, the self-driven drones and ground vehicles, these vehicles always fails to arrive at their destination in the first attempts, but after dozens of tries, they can complete the task easily as they have learned what to and not to do by analyzing increasingly more data. [3] Machine learning technology has really taken off for the past decade, it is gradually starting to be implemented in our everyday life. [2] An example that shocked the world not very long ago is the google DeepMind product AlphaGo [4], which defeated one of the world’s best Go players, Lee Sedol with a convincing score of 4-1. [5] It is clearly shown that machine-learning can complete many things that human beings cannot achieve, by recording and analyzing data to predict the right action. The question was then raised, with enough data, can machine predict human behavior? Many people have considered this possible [1] and are looking to implement that in use, for example, stopping crime before it happens. Despite the powerfulness that machine learning has already shown, it isn’t possible for it to predict human behaviors. Humans have instincts, decisions might not be the most logical when analyzed. In the AlphaGo vs Lee match, Lee won the fourth game by making an amazing play that no one could’ve thought of. These types of decisions are made not only based on calculations, but also the instinct coming from the player’s experience or, “guts”. [6] The core of machine learning is data analysis, but these data never demonstrates a person’s mind, only what they have done, the research by MIT [1] used an example of predicting the possibility of a student dropping out, the data analyzed consist of, time the student spent on the course website, how long ahead did the student start a problem set etc. What the data doesn’t show, is the student’s family and friends influences, their mental status, etc. Which are all imperative when it making decision. In the end, even though the technology of machine learning is developing unbelievably fast, and we are likely to get products such as self-driving vehicles in no time. It is unlikely that this technology will be able to analyze and predict human behaviors due to human’s instinct and personalities. References: [1] O. Goldhill, "An algorithm can predict human behavior better than humans," Quartz, 2015. [Online]. Available: https://qz.com/527008/an-algorithm-can-predict-human-behavior-better-than-humans/. [2] SAS Institute, "Machine learning: What it is and why it matters," 2017. [Online]. Available: https://www.sas.com/en_us/insights/analytics/machine-learning.html# [3] A. Schoellig, "Learning Control Theory and Foundations," 2014. [Online]. Available: http://www.dynsyslab.org/portfolio/learning-control/. [4] DeepMind AlphaGo, 2016. [Online]. Available: https://deepmind.com/research/alphago/. [5] S. Borowiec, "AlphaGo seals 4-1 victory over go grandmaster Lee Sedol," The Guardian, 2016. [Online]. Available: https://www.theguardian.com/technology/2016/mar/15/googles-alphago-seals-4-1-victory-over-grandmaster-lee-sedol. [6] C. Metz, "In Two Moves, AlphaGo and Lee Sedol Redefined the Future," WIRED, 2016. [Online] https://www.wired.com/2016/03/two-moves-alphago-lee-sedol-redefined-future/. |
|