Introduction
With the development of information technology and the Internet, human beings have switched from the era of information scarcity to the era of information overload. Information overload refers to the situation when the brain is overwhelmed with more information data than it can handle and process (Daiz, 2022). In order to cope with the information overload, a recommendation system supported by big data and intelligent algorithms was born (Meng, 2021) . Recommendation algorithms can extract data from a large amount of information, thereby improving user efficiency and reducing waste of time and energy. Nowadays, recommendation algorithms have penetrated deeply into all levels of society, around 75% of the content or information people got comes from kinds of recommendation algorithm (Kasula, 2020). However, while the recommendation algorithm is helping users in lots of area, it seems that it is also conducting some secret operations in private. This Blog will start with a basic introduction to recommendation algorithms, and then conduct a critical analysis of recommendation algorithms with examples to more fully illustrate whether recommendation algorithms is as “pure” as it shows.
What is Recommendation Algorithm?
A recommender system is a type of artificial intelligence or AI algorithm related to machine learning（NVIDIA, 2021）. The recommendation algorithm analyzes and predicts the interests and desires of consumers at a highly personalized level through the interaction data received from mobile phones, including impressions, clicks, likes, purchases, etc., so as to recommend relevant content to users. Whether the user’s data is obtained will have a huge impact on the calculation results of the recommendation algorithm. Bilibili is a well-known video website in China. When the user is not logged in, the content pushed by the website is messy and includes various types. video. However, when the user has logged in, the push content of the video will be more unified in some types.
Recommendation algorithms can be a good helper for users
Recommendation algorithms can tease out very subtle human behaviors and provide users with personalized content recommendations. It is widely used to filter out “invalid” items and information based on user characteristics, therefore recommend items and information that meet user preferences in a more detailed and targeted stage, also reduce the time wasted by users browsing a huge amount of irrelevant information, thereby optimizing improved user experience (Hallinan, Striphas, 2016). Through analyzing the similarity between items, Netflix uses algorithms to find and determine other content similar to the content that members have seen, and further matches and screens based on the records of member consumption, and finally displays it (Simplilearn, 2023). Recommendations based on “personalization” meet people’s individual needs, just like customizing a “private list” for each user. The algorithm recommendation is a personal portrait, and the system provides users with interesting content based on the data information in the network, which greatly saves people’s time and energy in obtaining content that meets their personal needs. Personalized recommendations start from the individual and are centered on the individual. Individuals are no longer limited by information content and time and space in the mass media era, but focus on personal interests and meet individual needs.. RED is a social software. Before I entered “pets”, I found that there was no information about pets on the interface, because this was a recommendation based on my previous search results (Picture 1). However, when I search for “pets” and return to the discovery interface, the algorithm has already started matching pictures/content related to pets and displayed them on the interface (Picture 2).
Picture 1
Picture 2
In addition to showing users information with a high degree of similarity based on previous data, recommendation algorithms can also help users find additional preferences. Steve Jobs said: “A lot of times, people don’t know what they want until you show it to them” (Edson.J, 2012) The recommendation algorithm can also make new matches based on the analysis data, showing users what they didn’t want before. Items that have been touched but may be of interest (muquido, 2020). China’s Taobao.com is an online shopping website under the Alibaba Group in mainland China. It is the largest online retail platform in the Asia-Pacific region. When a user clicks on Taobao, there are recommendations related to the user’s previous search information on the Taobao page. In addition, there will be a small number of other types of recommendations. These few recommendations, which seem to be less relevant to the previously searched information, can also arouse the user’s interest and even start a new hobby
However, recommendation algorithm can also be a spy and accomplice
Although the recommendation algorithm seems to exist entirely to optimize the user experience, and it has indeed achieved certain results. However, users are monitored and controlled invisibly through recommendation algorithm. Recommendation algorithms shape the practice of surveillance (Kate, 2021). Zobuff calls surveillance capitalism the monetization of data captured by monitoring people’s activities and behaviors online and in the real world (Barney, 2023). She believes that surveillance capitalism uses the data presented by users on the data platform as raw materials, and after processing these data, sells them to the behavioral futures market. Therefore, these markets can use these users data to capture users’ interests, and make commercial products and advertisement based on these information (Zobuff, 2020). Google has enabled this behavior to obtain more accurate data on the relationship between the two through the user’s click and search behavior, and make advertisements and purchase recommendations based on this data (Terry, 2022). In this case, user privacy is in a gray area, and users may unknowingly, through simple clicks or searches on digital platforms, their data becomes raw material and processed and sold, their preferences are broadcast and sold as commodities. In a sense, what the user sees is not necessarily what the user “want to see”, but may also be what the algorithm “wants the user to see”. The recommendation algorithm becomes the role of the salesperson, recommending information and products based on this data. In this case, users gradually give up actively exploring and obtaining information by themselves, and passively accepting information has become a habit. And in the process of passively accepting information, the thinking of the brain becomes slow, gradually producing an addictive behavior (KC, 2022). Users are monitored, controlled, and used unconsciously, and become pawns in business competition. Facebook uses this well to control users. It combines the recommendation algorithm with the infinite scrolling method to encourage users to continuously swipe to browse new content, enhance user addiction, and improve user retention (Newberry, 2023). TikTok also adopts a similar method. When the user clicks on Tiktok, a large amount of filtered information passively and coherently enters the user’s brain through simple sliding and short videos, and the user’s thinking is gradually influenced by these large amounts of information. which causing users to spend a lot of time unconsciously browsing these videos.
Moreover, due to the solidification of accepted information types, recommendation algorithms create an information cocoon for users, which may lead to users gradually becoming withdrawn and losing the ability to see things comprehensively, even socially isolated. The information cocoon refers to the fact that users are trapped in a fixed information field due to the active selection of users and the passive recommendation of algorithms(Han, 2022). Users who have been in the “information cocoon room” for a long time lack the opportunity to contact heterogeneous people or viewpoints. They only focus on the fields they are familiar with, and have almost no understanding of things in other fields. If a user clicks on a dancing video on Tiktok, then Tiktok will push him more and more dancing videos until his recommendation page is flooded with dancing videos and he cannot see other types of videos. For a long time, the user cannot receive other information, and his life becomes monotonous and boring. He will not participate in the discussion of other topics, and gradually breaks away from other richer lifestyles that he might have participated in.
Repeated recommendation of conceptual content will strengthen the user’s perception of a certain opinion, and even the polarization of a certain opinion. In 2022, there was an incident in Hebei, China, where a car-hailing driver continued to deviate from the route and caused a female passenger to jump off the car, which aroused a lot of attention. Because the driver continued to deviate from the route during the ride, the passenger asked the situation many times, but the driver ignored it, causing the passenger to jump out of the car because of fear. As soon as the incident came out, it immediately aroused huge public opinion. Since there have been many similar negative incidents of online car-hailing before, in most people’s impressions, online car-hailing has been associated with negative news, forming people’s opinion. inherent cognition. Therefore, when the driver ignored the passenger’s questions and the passenger still chose to jump off the car when there was no danger for the time being, most people believed that this was another vicious incident before the police had reached the conclusion of the investigation(Yifang, 2022). It is not difficult to see that people have a relatively fixed understanding of events related to online car-hailing, which interferes with their judgment on things.
Conclusion
All in all, the recommendation algorithm does optimize the user experience, helps users find their own preference information more efficiently and accurately, and even helps users dig out their own currently unknown preferences. But there is no doubt that although it is difficult to be detected, the disadvantages of the recommendation algorithm are also obvious. Its function is not as pure as it seems on the surface. And the potential connection with surveillance capitalism and recommendation algorithm may cause huge problems. However, we currently have no way to cut off these signs as there is no good solution exist. A clear solution may need to wait for users to become more aware of these problems in the future. Ultimately, it’s hard to tell whether the referral code is a helper or an accomplice, and so far it appears to be both. Therefore, while enjoying the convenience brought by the recommendation code, users also need to be vigilant and not fall into the hotbed of the recommendation algorithm
Bibliography
Crawford, K. (2021). Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. New Haven: Yale University Press. https://doi-org.ezproxy.library.sydney.edu.au/10.12987/9780300252392
Take online courses. earn college credit. Research Schools, Degrees & Careers. (n.d.). Retrieved April 17, 2023, from https://study.com/learn/lesson/information-overload-overview-examples.html
Feng, S., Meng, J., & Zhang, J. (n.d.). Journal of Web Engineering. Retrieved April 17, 2023, from https://journals.riverpublishers.com/index.php/JWE/article/view/5969
Kasula, C. (2020, June 28). Netflix Recommender system - A big data case study. Retrieved April 17, 2023, from https://towardsdatascience.com/netflix-recommender-system-a-big-data-case-study-19cfa6d56ff5
What is a recommendation system? (n.d.). Retrieved April 17, 2023, from https://www.nvidia.com/en-us/glossary/data-science/recommendation-system/
Edson, J., 2012. Design Like Apple: Seven Principles for Creating Insanely Great Products, Services, and Experiences. John Wiley & Sons. pp. 47
Hallinan, B., & Striphas, T. (2016). Recommended for you: The Netflix Prize and the production of algorithmic culture. New media & society, 18(1), 117-137.
Simplilearn. (2023, February 16). Netflix recommendations: How netflix uses AI, Data Science, and ML: Simplilearn. Retrieved April 17, 2023, from https://www.simplilearn.com/how-netflix-uses-ai-data-science-and-ml-article
Miquido, & Holewa, K. (2023, April 14). Recommendation systems: Benefits, Types & Examples – Miquido blog. Retrieved April 17, 2023, from https://www.miquido.com/blog/perks-of-recommendation-systems-in-business/
Flew, Terry (2021) Regulating Platforms. Cambridge: Polity, pp. 79-86.
Crawford, Kate (2021) The Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. New Haven, CT: Yale University Press, pp. 1-21.
Simplilearn. (2023, February 16). Netflix recommendations: How netflix uses AI, Data Science, and ML: Simplilearn. Retrieved April 17, 2023, from https://www.simplilearn.com/how-netflix-uses-ai-data-science-and-ml-article
Kasula, C. (2020, June 28). Netflix Recommender system - A big data case study. Retrieved April 17, 2023, from https://towardsdatascience.com/netflix-recommender-system-a-big-data-case-study-19cfa6d56ff5
Mayigongfang. Retrieved April 17, 2023, from https://m.eefung.com/daily-report/20220105162114270
Sima, Y., & Han, J. (2022). Online carnival and offline solitude: “information cocoon” effect in the age of algorithms. Advances in Social Science, Education and Humanities Research. doi:10.2991/assehr.k.220504.445
Be the first to comment