top of page
Search

The Sociology of Artificial Intelligence

INTRODUCTION  


Artificial intelligence (AI) has undergone an evolution unlike any other technology. Just in a couple of decades, it has transformed from a mere science-fiction imagination to a significant part of our reality. Now, students use AI to make their essays, employees use it to meet their deadlines and companies use it for recognition of customer likes and dislikes. It has become a rising force woven into the very fabric of all our lives. From the moment we unlock our phones that are powered by intelligent assistants, to the recommendations on our favorite products influence our access. AI nowadays also shapes our social interactions. 

 


John McCarthy in 1956 said, "Artificial Intelligence is the science and engineering of making intelligent machines, especially intelligent computer programs. “However, the sociological implications of AI extend far beyond its impressive technical capabilities. The impact of AI is on our social interactions. It can be looked at as a sociological phenomenon.  

This research article delves into the fascinating realm of the sociology of artificial intelligence, discussing how AI interacts with and reshapes the social world, fundamentally altering the landscape of human interaction. What this research deals with is relatively unexplored. It is quite a task to look at AI from the lens of sociology. But, the power of sociological imagination guides this research. The rise and working of AI in our  contemporary society will be analyzed in this study.  

 

This research shall look at AI in the context of the Theory of Social Learning by Albert Bandura and will further decide if AI can be said to be a Social Fact according to Émile Durkheim. We’ll also delve into the legal aspect a bit by observing if AI can be given personhood. By the end of the study, we will have a comprehensive understanding of the  sociological relevance of artificial intelligence. 

 

THE EMERGENCE OF AI IN SOCIETY AS A SOCIAL FACT 


Shaping the role of artificial intelligence (AI) in the social reality of human society turns out  to be an interesting phenomenon, which carries significant implications in various aspects of our lives. Besides people, experts also, have shown a certain level of both excitement and anxiety over the way AI keeps coming and becoming an integral part of society. 

 

French sociologist Emile Durkheim in The Rules of Sociological Method (1895) gave the theory of social facts. Durkheim's idea of "social fact" seems to have an implication in the AI domain. Durkheim defined social facts as "any way of acting, whether fixed or not, capable of exerting over the individual an external constraint; or: which is general over the whole of a given society whilst having an existence of its own, independent of its individual manifestations." Additionally, the impact of AI on human relationships, as explored by experts like Sherry Turkle, suggests the transformative influence of AI on social interactions and intimacy. This aligns with Durkheim's idea of social facts shaping human behaviour and societal norms.  So, to prove AI is a social fact we’ll have to check if it fits the 6 characteristics of a social fact: 


1. External to the individual: AI, as a phenomenon, exists independently of any single individual. It is a product of collective human effort, made of algorithms, hardware, and data, which are external to any individual user. 


2. Independent of their Individual Manifestation: AI systems operate based on predefined algorithms. While individual manifestations (such as specific applications or instances of AI use) may vary, the underlying principles and mechanisms of AI remain consistent across these manifestations. 


3. Coercive of Individual: AI can exert influence or constraint over individuals in various ways. For instance, recommendation algorithms on social media platforms can shape users' perceptions and behaviors, and automated decision-making systems in areas like hiring or loan approvals can significantly impact individuals' opportunities and outcomes. 


4. Explained by other Social Facts: AI's development and impact are intertwined with numerous social facts. We can explain AI using social facts like ‘technology’, ‘machinery’, and coding as a social fact. These all contribute to understanding the concept of Artificial Intelligence. 


5. Experienced through other Social Facts: Individuals experience AI through various social facts and contexts, such as education, employment, healthcare, entertainment, and governance. The usage of AI systems in these contexts shapes individuals' interactions, decisions, and experiences within society. 


6. Empirically Studied: AI is empirically studied across disciplines such as computer science, engineering, sociology, psychology, and ethics. Sociological empirical studies investigate its technical capabilities, societal implications, ethical considerations, and impacts on individuals and communities. 

 

Therefore, we can conclude that AI can be said to be a social fact. This conception will only increase over time, as it integrates into the society and becomes more coercive to a larger number of people. In summary, the concept of "social fact" as given by Emile Durkheim finds resonance in the domain of AI, as AI's pervasive influence on society aligns with the fundamental elements that exist beyond individuals' control and shape their behavior.  

 

DOES AI PERFORM SOCIAL LEARNING? 


Albert Bandura in his book ‘Social Learning Theory’ defines social learning as ‘a process that involves observing, imitating, and modelling the behaviors, attitudes, and emotional reactions of others.’ The 1961 Bobo doll experiment is one of the most well-known instances that Albert Bandura himself used to support his social learning hypothesis. Through observation and imitation of adults, children pick up aggressive behavior, as shown by the experiment conducted by Bandura and colleagues. Individuals who observed aggressive behavior were more likely to mimic it by hitting and punching the doll, but those who experienced non aggressive behavior showed less violence. This experiment showed that people pick up behaviors by watching and copying others, which supported Bandura's theory empirically and highlighted the role that social learning processes play in forming people's behavior. Further, Bandura proposed that learning can occur through direct experiences, but also through observing and imitating the actions of others, particularly within a social context. This theory emphasizes the importance of cognitive processes, such as attention, retention, reproduction, and motivation, in the learning process. 


Though generally applied in the context of social learning in humans, we thought it would be interesting to observe if it applies to the learning process of Artificial Intelligence. For this, we need to assume AI to be something that has ‘personhood’. This statement will be elaborated on in the third objective. But for now, let us first see how AI learns and then we’ll see if it could come under the theory given by Bandura. 

 

So, AI can engage in social learning through different mechanisms that are designed to mimic human social learning processes: 


1. Observational Learning: AI algorithms can learn by observing and imitating human behavior. For example, in ‘reinforcement learning’, AI can observe how humans perform certain tasks and then try to mimic their actions to achieve the similar outcomes. 


2. Collaborative Filtering: This technique is mostly used in recommendation systems. Here, AI algorithms analyze the behaviors and preferences of a group of users to make recommendations to individual users. By observing the preferences of similar users, the AI system then learns to make more accurate predictions about what a specific user might like. 


3. Social Network Analysis: AI systems sometimes analyze social networks to understand how information and influence flow through certain communities. By analyzing interactions between individuals and groups, AI can learn about trends, opinions, and behaviors prevalent in a society. This is how eco-chambers are formed by political parties’ IT Cells. 


4. Natural Language Processing (NLP): This mechanism got famous recently as here AI models trained on large corpora of text data learn language patterns of people, including social cues and norms. By analyzing conversations, social media posts, and other forms of communication, AI can infer social dynamics and learn from the language used by individuals! 


5. Human-AI Interaction: AI systems sometimes learn directly from direct interactions with humans. Through feedback mechanisms, AI agents can adapt their behavior based on the suggestions, responses and corrections provided by human users. This is seen in all the popular Ais like ChatGPT and Gemini, where it asks how satisfied you are with the answer given by the AI. All these methods point to one answer, which is that AI indeed performs social learning. Even though different from how humans perform this social learning, an argument can be made for including AI as an example of Bandura’s Social Learning Theory. 

 

Here is an experiment to conclude this objective: 


I asked ChatGPT to answer one question normally, 



But then I asked it to answer the same question but how a human would answer it. 


If you carefully observe, ChatGPT has added emotions and humane experiences when it comes to the second prompt. So, it has learnt using social learning that humans talk with emotions and subjectivity. Therefore, we can conclude that AI indeed performs social learning. 

 

AI AS A PERSON OF ITS OWN? 


The legal status of artificial intelligence (AI) has prompted a debate among experts and lawmakers worldwide. Can AI be considered a ‘person’ in the legal sense? There are two ways to satisfy the claim of calling AI a legal person. First way is to ‘logically’ prove as to why it should be a legal person. The second way is to ‘legally’ satisfy the criteria for a legal person given by previous precedents. 

 

To prove logically, let us look at the popular contemporary arguments. Professor Ryan Abbott, in his famous work ‘The Reasonable Robot’, advocated for AI to be recognized as a legal person. Abbott argues that granting legal personhood to AI could facilitate accountability and responsibility, especially in cases where AI systems are involved in decision-making processes. For example, imagine a case where an autonomous vehicle that runs on an AI system, gets into a crash accident. In this situation then, isn’t it the fault of the 


Artificial Intelligence system? However, since AI is not a legal person, the liability for this crash goes into unfortunate ambiguity. 


In addition to the side presented by Abbott, some concerns have been raised about the rights and responsibilities of AI if it were to be recognized as a legal person. This has led to diverse legislative responses across different countries. For example, the recent European Union's AI Act is a new approach to regulating AI to ensure the safety, transparency, and non

discriminatory nature of AI systems. But in the future, the legislation on this will need to be more comprehensive (especially in highly populated country like India). However, it is important to note that countries like Japan have adopted a more cautious approach. This approach avoids prescriptive regulations and instead allows for continued innovation in AI development. This juxtaposed with the European approach shows the diversity of approaches to AI regulation and the ongoing efforts to balance technological advancement with legal and ethical considerations. 


On the other hand, some sociologists and experts are opposing the pursuit to personhood for artificial intelligence. The overarching argument for this seems to be the fact that after all these systems are ‘artificial’. Dr. Brandeis Marshall, an American data scientist, argues against the premature discussions of granting moral consideration or civil rights to AIs. She feels this is a very early stage for personhood for a nascent concept like AI. Marshall supports focusing on first building a solid social framework for AI use that prioritizes protecting the civil rights of all humans impacted by AI. This perspective highlights the importance of considering the sociological impact of granting legal personhood to AIs. 

 

Now we move on to the legal aspect. We have seen legal personhood being given to nonhuman entities before. There were corporations (through Dartmouth College v. Woodward in 1819), and NGOs (in the Supreme Court case "Citizens United v. Federal Election Commission" of 2010). These implicitly establish a few criteria to give legal personhood: 


1. Autonomy and Decision-Making: Most experts believe AI has enough free will to act independently and make its own choices. But others believe it to simply be a tool controlled by humans. 


2. Distinct Legal Existence: It’s not clear if we can really separate AI from its creators and make it liable for its actions.  


These two criteria not being satisfied properly point us to the answer that AI cannot be given personhood, at least at this stage.  

 

It is evident how difficult it is to resolve AI's legal standing from the viewpoints of legal professionals and the legislative responses from various nations. This highlights how experts, legislators, and stakeholders must continue to communicate and work together to determine the legal implications of artificial intelligence in a changing society.  

 

CONCLUSION 


In conclusion, let us connect the three objectives together and what they have told us. So, all these 3 objectives were to look at AI from a very sociological lens. We have used our sociological imagination to use a daily use tool to extrapolate it to the discipline of sociology.  Thus firstly, we learned using the characteristics of social facts that AI indeed can be classified as a social fact under the definition given by Durkheim. Secondly, we understood how AI performs social learning to improve its responses and thus can be classified under the definition of social learning given by Bandura. Thirdly, we analyzed the scope of personhood for Artificial Intelligence. This was an intersection of law and sociology and was very interesting to observe. We found that it is too premature to consider AI as a legal person. Together, these three learnings can be connected to conclude that Artificial Intelligence is indeed a very important sociological phenomenon. It has a lot of applications in various sociological theories. It also has immense significance in the society, especially in terms of our future and how we will live. 

 

Keeping aside the conspiracy theories of AI taking over the human race, it is apparent how integral artificial intelligence is going to be in the coming years for our society to evolve. The  more and more AI develops, it adds to our quality of life and makes complex tasks very  simple. Looking from a sociological macro perspective, AI will be important for the entire  society’s development, mainly technological development. And looking from a sociological micro perspective, AI has immense values to individuals like students and workers. Therefore, we close the lens of sociology towards artificial intelligence.  

 

REFERENCES 


1. Bandura, A. (1977). Social Learning Theory. Englewood Cliffs: Prentice-Hall. 

2. Durkheim, É. (1982). The Rules of Sociological Method (S. Lukes, Ed.; W. D. Halls, 

Trans.). New York: Free Press. (Original work published 1895). 

3. Woolgar, S. (1985). WHY NOT A SOCIOLOGY OF MACHINES? THE CASE OF 

SOCIOLOGY AND ARTIFICIAL INTELLIGENCE. Sociology, 19(4), 557–572. 

4. Wolfe, A. (1991). Mind, Self, Society, and Computer: Artificial Intelligence and the 

Sociology of Mind. American Journal of Sociology, 96(5), 1073–1096. 

5. Nelson, L. K. (2023, November 30). The Sociological Take on AI: Unpacking Current 

unpacking-current-debates/. 

6. Schwartz, R. D. (1989). Artificial Intelligence as a Sociological Phenomenon. The 

Canadian Journal of Sociology / Cahiers Canadiens de Sociologie, 14(2), 179–202. 

7. ResearchGate. (2018, July). Towards a Sociological Conception of Artificial 

Intelligence. Retrieved from 

 ption_of_Artificial_Intelligence.  

8. Attwood, A. I. (2020). Changing social learning theory through reliance on the 

internet of things and artificial intelligence. Journal of Sustainable Social Change, 

9. OpenAI. (n.d.). How ChatGPT and our language models are developed. 

are-developed. 

10. Marshall, B. (2023, November 10). No legal personhood for AI. 

11. Abbott, R. (2020). The Reasonable Robot: Artificial Intelligence and the Law. 

Cambridge Publications. 

12. Kurki, V. A. J. (2019). The legal personhood of artificial intelligences. In A Theory of 

Legal Personhood. Oxford. Retrieved from 

13. Singh, S. (2017). Attribution of Legal Personhood to Artificially Intelligent Beings. 

Bharati Law Review, 194-201. 

F451330BFDB5.pdf.

 

This article is authored by Samyak Deshpande. He was among the Top 40 performers in the Contract Drafting Quiz Competition organized by Lets Learn Law.

 
 
 

Comments

bottom of page