Authors: Neshat Parvin, Arshmeet Kaur
Abstract
This paper examines how algorithmic systems and digital surveillance interact to shape user behaviour on contemporary digital platforms. Drawing on Shoshana Zuboff’s theory of surveillance capitalism, David Lyon’s concept of the culture of surveillance, Frank Pasquale’s analysis of algorithmic opacity, and Theodor Adorno and Max Horkheimer’s critique of instrumental rationality, the study explores the relationship between data extraction, behavioural prediction, and user autonomy. Using a qualitative methodology based on semi-structured interviews with students at Delhi University, the research investigates perceptions of data collection, algorithmic influence, and digital privacy. The findings suggest that participants are generally aware that platforms collect personal data and use algorithms to curate content, yet they feel limited in their ability to understand, control, or challenge these processes. Interview responses reveal experiences of behavioural conditioning, content personalisation, self-censorship, and a sense of resignation toward pervasive digital surveillance. The paper argues that algorithmic systems function not merely as technological tools but as mechanisms of behavioural governance embedded within the broader structures of surveillance capitalism. These developments raise significant concerns regarding individual autonomy, democratic accountability, and the growing concentration of informational power in the hands of digital corporations.
Introduction
Over the past two decades, the rapid expansion of digital platforms has transformed the relationship between technology, power, and everyday human behaviour. What began as tools designed to connect people and simplify information access has evolved into something far more complex: a vast infrastructure for collecting, analysing, and acting upon human behavioural data. This paper asks a focused central question: in what ways do algorithmic systems and digital surveillance work together to modify user behaviour, and what does this mean for individual autonomy and democratic accountability?
This is not simply a question about technology. It is a question about power. When a platform like Meta curates a user’s news feed, or when TikTok’s recommendation engine determines which videos appear next, these are not neutral acts. They reflect deliberate design choices rooted in commercial objectives. As Shoshana Zuboff argues, digital platforms have pioneered a new economic logic-one that transforms human experience itself into raw material for behavioural prediction and modification. Google’s advertising ecosystem, for example, does not merely sell attention, it sells predictions about what users will do next.
To understand these dynamics, this paper draws on a carefully selected set of theoretical frameworks. Zuboff’s concept of surveillance capitalism provides the economic foundation; David Lyon’s work on the culture of surveillance extends this into social and cultural life; Frank Pasquale’s analysis of algorithmic opacity raises urgent questions about accountability; and Theodor Adorno and Max Horkheimer’s critique of instrumental rationality offers a deeper historical lens through which to examine how reason itself can become a tool of domination. Each framework addresses a distinct dimension of the problem, and together they allow for a more complete picture than any single theory could provide.
The paper proceeds as follows. The literature review critically examines each theoretical tradition and identifies tensions between them. The theoretical framework then applies these ideas directly to contemporary algorithmic systems. The methodology explains the qualitative approach used to ground these abstractions in lived experience. Results and analysis present findings from student interviews, and the conclusion draws out the broader implications of the argument.
Literature Review
The scholarly literature on digital surveillance and algorithmic power has grown considerably in recent years, but it remains fragmented across disciplines. Three bodies of work are particularly relevant here, though they approach the problem from different angles and with different emphases. Rather than treating them as a unified school of thought, this review traces where they converge, where they diverge, and what gaps remain.
Surveillance Capitalism and Behavioural Data Extraction
Shoshana Zuboff’s The Age of Surveillance Capitalism (2019) provides the most comprehensive account of how digital platforms extract and monetise human behavioural data. Zuboff’s central argument is that companies like Google and Facebook do not simply sell products; they sell predictions about human behaviour. This is possible because every interaction a user has with a digital platform, every search query, every click, every moment of hesitation before a purchase-generates what Zuboff calls ‘behavioural surplus’: data that exceeds what is needed to improve the service and becomes the raw material for predictive models sold to advertisers and other third parties.
It is worth pausing on the mechanism here, because the paper’s central argument depends on understanding exactly how this shaping occurs. Algorithms do not modify behaviour directly through coercion. Instead, they work through what Zuboff describes as ‘instrumentarian power’: the structuring of information environments in ways that nudge users toward particular choices. When Instagram’s algorithm surfaces content that provokes strong emotional responses, it is not simply predicting engagement; it is actively engineering it. When Spotify generates a personalised playlist, it is not merely reflecting taste; it is gradually shaping it by exposing users to new music within carefully calibrated parameters.
Zuboff also introduces the concept of the ‘Big Other’: a distributed digital infrastructure that monitors and shapes behaviour without the visible coercive apparatus of traditional surveillance. Unlike a police state, the Big Other operates through convenience and personalisation. Users do not feel watched; they feel served. This asymmetry-in which corporations accumulate extraordinary knowledge about individuals who remain largely ignorant of how that knowledge is used- is central to Zuboff’s critique.
One limitation of Zuboff’s framework, however, is that it can read as technologically deterministic. By framing surveillance capitalism as an almost irresistible systemic logic, it risks understating the agency of users and the role of regulatory and political contestation. Lyon’s work provides a useful corrective here.
Surveillance and the Transformation of Social Life
David Lyon’s The Culture of Surveillance (2018) shifts attention from the economic logic of data extraction to its social and cultural consequences. Where Zuboff focuses on corporations as the primary agents of surveillance, Lyon emphasises how surveillance has become woven into the fabric of everyday social interaction. People do not simply submit to surveillance; they participate in it, often willingly.
Lyon’s concept of a ‘culture of surveillance’ captures something important that Zuboff’s account sometimes misses: the extent to which users have internalised surveillance as normal. Sharing location data with apps, posting personal milestones on social media, consenting to cookies without reading them- these are not acts of ignorance alone. They reflect a broader cultural shift in which visibility has become associated with social participation. Lyon draws on Foucault’s concept of the panopticon, but updates it for an era in which the watchtower is not a prison structure but a smartphone.
However, Lyon’s account diverges from Zuboff’s in an important way. While Zuboff treats behavioural modification as primarily a corporate project — driven by the profit motive and enabled by data- Lyon is more attentive to the role of state surveillance and the ways in which commercial and governmental data collection increasingly intersect. The post-Snowden revelations demonstrated that platforms like Google and Facebook were not only mining data for commercial purposes; they were also sharing it with intelligence agencies. This blurring of commercial and state surveillance is a dimension that Zuboff’s framework does not fully address.
Lyon also draws attention to how surveillance produces social sorting: the categorisation of individuals into profiles that determine access to services, opportunities, and information. A person categorised as a high credit risk may be shown different financial products; a user whose browsing history marks them as politically conservative may be fed a different information diet. These sorting processes are rarely transparent to those they affect, which brings us to Pasquale’s contribution.
Algorithmic Power and the Black Box Society
Frank Pasquale’s The Black Box Society (2015) focuses on a problem that both Zuboff and Lyon acknowledge but do not fully explore: the opacity of algorithmic systems and its consequences for accountability. Pasquale argues that the algorithms used by search engines, social media platforms, financial institutions, and government agencies are deliberately kept hidden from public scrutiny. This opacity is not incidental; it is structural. Transparency would invite regulatory oversight, competitive imitation, and user resistance.
Pasquale’s analysis is particularly illuminating when applied to concrete cases. Google’s PageRank algorithm determines which information users encounter and which remains effectively invisible. Facebook’s content moderation algorithms decide what counts as acceptable speech on a platform used by billions of people. Amazon’s pricing algorithms adjust prices in real time based on user behaviour, location, and purchasing history. In each case, decisions with profound social consequences are made by systems that are accountable to no one outside the corporation that owns them.
The relationship between opacity and democratic accountability is a critical one. When users cannot understand why they are being shown particular content, recommended specific products, or assigned a particular credit score, they cannot meaningfully contest those decisions. This is not merely a consumer rights issue. It is a political one. Algorithmic systems that determine what information people encounter shape political opinions, influence elections, and frame what counts as credible knowledge. The inability to scrutinise these systems represents a structural threat to the conditions of democratic deliberation.
Where Pasquale diverges most sharply from Zuboff and Lyon is in his emphasis on legal and regulatory solutions. Zuboff tends toward a broader critique of the economic system that produces surveillance capitalism; Lyon is more sociologically descriptive; but Pasquale is insistently reformist, arguing that transparency requirements, algorithmic auditing, and stronger antitrust enforcement could meaningfully constrain corporate power. This is a genuine tension within the literature, and one that this paper will return to in the conclusion.
Theoretical Framework
The theoretical framework developed in this paper draws primarily on Adorno and Horkheimer’s Dialectic of Enlightenment, but situates their argument in relation to the contemporary scholarship reviewed above. The aim is not simply to apply a classical text to a new context, but to show how the historical process they described continues to unfold in the specific form of algorithmic capitalism.
Adorno and Horkheimer’s central argument, developed in the context of fascism and mass culture in the 1940s, is that the Enlightenment promise of emancipation through reason had become its opposite. Reason, which was supposed to free human beings from myth and domination, had been instrumentalised: reduced to a means of calculation, efficiency, and control. Science and technology, rather than serving human flourishing, had become tools for managing and manipulating human populations.
This concept of instrumental rationality is directly relevant to contemporary algorithmic systems. When a platform like TikTok uses machine learning to maximise time-on-app, it is not making a cultural or ethical judgement about what content is valuable. It is performing a calculation: what sequence of videos, presented at what intervals, will keep this particular user engaged for the longest time? The output of this calculation -the endlessly scrolling feed, the algorithmically timed notification, the autoplay feature -is an expression of pure instrumental rationality, stripped of any concern for what the user actually wants or needs.
Adorno and Horkheimer also developed the concept of the culture industry to describe how mass media under capitalism standardises cultural output, not in order to fulfil human needs, but to generate compliance and consumption. Contemporary algorithmic platforms represent a sophisticated extension of this dynamic. Netflix’s recommendation algorithm does not simply reflect viewer preferences; it actively shapes them by determining which content receives prominence and which is buried. Spotify’s Discover Weekly playlist does not emerge from some neutral assessment of musical quality; it is engineered to maximise streaming hours. In both cases, the appearance of personalisation conceals what is, in practice, a highly managed cultural environment.
The connection between Adorno and Horkheimer’s framework and Zuboff’s surveillance capitalism is particularly instructive. Both identify a historical process in which a system ostensibly designed to serve human beings ends up serving itself -or more precisely, serving the interests of those who own and control it. Zuboff’s ‘instrumentarian power’ is, in many ways, instrumental rationality applied to human behaviour itself: the reduction of human experience to data points that can be analysed, predicted, and ultimately steered.
Lyon’s culture of surveillance fits within this framework as the social manifestation of these processes. If instrumental rationality describes the logic of algorithmic systems at the level of design, Lyon describes what it feels like to live inside those systems: the normalisation of monitoring, the voluntary surrender of privacy, the gradual restructuring of social life around visibility. Together, these frameworks suggest that algorithmic behavioural control is not simply a corporate practice. It is a feature of a broader social formation in which the logic of calculation and prediction has become hegemonic.
Pasquale’s contribution, in this theoretical context, is to focus attention on the institutional mechanisms through which this logic is sustained and protected. Algorithmic opacity is not just a technical feature; it is a political one. It protects the systems of instrumental rationality from external accountability, making them harder to contest and easier to entrench. The ‘black box’ is, in this sense, the legal and institutional expression of what Adorno and Horkheimer described as the structural tendency of instrumental reason to conceal its own operations
Methodology
This study adopts a qualitative approach, grounded in semi-structured interviews with students at Delhi University. The choice of a qualitative method is not incidental. The theoretical framework developed above is concerned with lived experience- with how people actually feel and think about algorithmic influence, data collection, and surveillance. Quantitative methods could measure awareness or attitudes at scale, but they cannot capture the texture of how students navigate these experiences in their daily lives, or how they articulate the relationship between their online behaviour and the platforms that shape it.
Semi-structured interviews were chosen because the topic requires both focus and flexibility. Fixed questions ensured that key themes -data awareness, algorithmic influence, the experience of addiction or compulsion on platforms, were addressed consistently across participants. The open-ended format allowed for follow-up questions when participants raised unexpected or particularly illuminating points. The interviews lasted between fifteen and twenty minutes each and were conducted in person, a choice made deliberately to encourage candour and reduce the discomfort that participants sometimes feel in recorded or digitally mediated settings.
Participants were selected using convenience sampling, targeting students who use social media platforms such as Instagram, YouTube, and TikTok on a daily basis. This sampling strategy was appropriate for the exploratory aims of the research, though it introduces limitations that are acknowledged below. The final sample consisted of fifteen to twenty participants, recruited initially through the researcher’s existing networks and extended through snowball sampling where necessary.
Prior to fieldwork, the interview schedule was piloted with three participants. This process revealed that questions about data awareness worked better when they preceded questions about algorithmic influence, allowing participants to move from more concrete and familiar concerns toward more abstract analytical ones. The pilot also allowed for refinement of note-taking protocols, as interviews were documented through detailed handwritten notes rather than audio recording, again to minimise participant self-consciousness.
Research Objectives
- To examine how digital platforms, use algorithmic systems and targeted content to shape and influence user behaviour.
- To assess the awareness and perceptions of users regarding data collection and surveillance practices on digital platforms.
Central Hypothesis
Adorno and Horkheimer’s Dialectic of Enlightenment reveals how reason, which promised liberation during the Enlightenment, has devolved into new systems of domination. This study applies their analytical framework to surveillance capitalism, proposing that digital platform users recognise that companies collect their data and that algorithms shape their online behaviour, but nevertheless feel unable to meaningfully control or contest these processes.
Interview Questions
- Are you aware that companies collect your data through apps, websites, and devices?
- Do you feel you have control over your personal data online?
- To what extent do you agree that companies collect more data than necessary?
- To what extent do you agree that AI algorithms influence what content you see online?
- Have you ever felt addicted to a platform due to constant engagement features?
- Can you recall a time when a recommendation led you to watch, read, or buy something you had not planned to?
- What do you find most useful about AI-driven features on the platforms you use?
When participants gave brief or monosyllabic answers, follow-up prompts such as “Why do you think that?” or “Can you tell me more about that experience?” were used to draw out more substantive responses. This approach helped to connect individual accounts to the broader theoretical concerns of the research.
Analysis
Interview notes were analysed iteratively. Initial coding identified recurring themes across participants; subsequent passes looked for patterns within and between themes. The analysis was guided by the theoretical framework, but remained attentive to unexpected findings. Three to four main themes were identified, and representative examples were selected to illustrate each. This approach stays close to what participants actually said while connecting their experiences to the theoretical argument about instrumental rationality and behavioural control.
Ethics
Informed verbal consent was obtained from all participants before interviews began. Participants were told that the research was being conducted for academic purposes, that participation was voluntary, and that they could withdraw at any time. Anonymity was maintained by referring to participants as Student A, Student B, and so on throughout the analysis.
Limitations
The study has several important limitations. The sample is small and drawn from a single university, which limits the generalisability of the findings. Convenience sampling may have produced a sample that is more digitally aware than the broader student population. The reliance on handwritten notes rather than recordings introduces the possibility of transcription error and limits the precision of quotation. These limitations are inherent to the exploratory, qualitative design of the study and should be borne in mind when interpreting the results.
Results and Analysis
The study was conducted using a mixed convenience and network sampling approach. Participants were contacted through existing social networks and invited to share the study with eligible peers. The final sample consisted primarily of students, all of whom used social media platforms daily and had some awareness of data privacy as a concept, though the depth of that awareness varied considerably.
Three themes emerged consistently across the interviews, each of which connects directly to the theoretical framework developed above.
Web Cookies as Tools of Behavioural Tracking and Data Exploitation
Participants demonstrated a basic awareness that websites use cookies to track their activity, but most had only a vague understanding of what this meant in practice. Several described accepting cookie notifications automatically, without reading them, because refusal seemed to impede access to content they wanted. This reflects precisely the dynamic Lyon describes in his account of voluntary surveillance: participation is not freely chosen so much as structurally compelled by platform design.
When participants were asked what they understood cookies to actually do, the most common response was that they ‘remember’ user preferences. Very few were aware that third-party cookies track browsing behaviour across different websites, building profiles that are shared with advertisers and, in some cases, with data brokers whose practices are almost entirely opaque to users. This gap between surface awareness and substantive understanding is significant: it suggests that the ‘informed consent’ implied by cookie banners is largely fictional. Users consent to something they do not understand, and the platforms that design those consent interfaces have no obvious incentive to make them clearer.
Digital Footprints and Instagram’s Curation of User Experience
The most animated responses in the interviews came when participants discussed their experiences of Instagram’s algorithm. Several described a sense of unease at the precision with which the platform seemed to know their interests, moods, and preoccupations. One participant noted that after searching for information about a particular health condition, her feed began surfacing related content within hours — content she had not sought out and did not want to see. Another described how his engagement with certain political commentary had gradually shifted the character of his Explore page, making it feel, over time, like a narrower and more ideologically consistent space.
These accounts illustrate, in concrete terms, the mechanism of algorithmic behavioural shaping that the theoretical framework identifies. Instagram’s algorithm does not simply reflect user preferences; it amplifies and reinforces them. Each interaction — a like, a pause in scrolling, a tap on a profile — is registered and fed back into the system, which adjusts future content accordingly. The result, as several participants described, is a gradual narrowing of the information environment, experienced subjectively as personalisation but functioning structurally as a form of behavioural conditioning.
Participants also described a related phenomenon: self-censorship. Several said they had become more careful about what they posted or searched for, aware that these actions would shape what they were subsequently shown. This is a striking finding. It suggests that algorithmic influence does not only operate through content curation; it also shapes what users choose to express, creating a feedback loop between platform design and user behaviour that goes well beyond passive consumption.
Data Privacy in the Age of Artificial Intelligence
The third theme concerned participants’ broader attitudes toward AI and data privacy. Most expressed a general unease about the amount of data being collected about them, but felt practically unable to do anything about it. Several used phrases like “it’s just how things are” or “there’s nothing you can do,” which reflects precisely the sense of structural helplessness that Pasquale’s analysis of algorithmic opacity predicts. When the systems that govern one’s digital life are invisible and unaccountable, resistance feels futile.
Participants were also uncertain about what AI actually does with their data. Most understood that AI systems ‘learn’ from user behaviour, but very few had a clear sense of how this learning operates or what it produces. This uncertainty is itself significant. It speaks to the gap between the sophistication of the systems shaping users’ digital lives and the understanding that users bring to their interactions with those systems. Instrumental rationality, as Adorno and Horkheimer noted, tends to conceal its own workings: the smoother and more effortless a system appears, the harder it is to see what it is actually doing.
Conclusion
This paper has argued that algorithmic systems and digital surveillance operate together to modify user behaviour in ways that are systematic, largely invisible, and imperfectly understood by those they affect. Drawing on Zuboff’s concept of surveillance capitalism, Lyon’s account of the culture of surveillance, Pasquale’s analysis of algorithmic opacity, and Adorno and Horkheimer’s critique of instrumental rationality, it has developed a theoretical framework that situates contemporary platform capitalism within a longer historical process: the transformation of reason from a tool of emancipation into a tool of control.
The empirical findings from student interviews support and enrich this theoretical account. Participants demonstrated awareness of surveillance and algorithmic influence, but felt structurally unable to contest or escape it. They described experiences of content curation, behavioural conditioning, and self-censorship that correspond closely to the mechanisms identified in the literature. The gap between surface awareness and substantive understanding-knowing that data is collected without understanding what is done with it- reflects the opacity that Pasquale identifies as a deliberate structural feature of platform capitalism.
The implications of these findings extend beyond individual user experience. When algorithmic systems shape what information people encounter, they also shape the conditions of democratic deliberation. A public sphere structured by engagement-maximising algorithms is not a neutral forum for the exchange of ideas; it is an environment designed to provoke reaction, reinforce existing beliefs, and keep users on-platform. The consequences for political life are serious and have become increasingly visible in recent years, from the amplification of misinformation on Facebook to the role of recommendation algorithms in radicalisation on YouTube.
The question of what to do about this is genuinely difficult, and this paper does not pretend to resolve it. Pasquale’s reformist argument for transparency requirements and algorithmic auditing is compelling as far as it goes, but it does not address the deeper structural incentives that drive surveillance capitalism. Zuboff’s more radical critique — that the problem is the economic system itself, not merely its excesses — points toward more fundamental change, but leaves open the question of what that change would look like in practice. What this paper does suggest is that any adequate response will need to grapple with the political dimensions of algorithmic opacity: the fact that systems making decisions with profound social consequences are answerable, at present, to no one outside the corporations that own them.
Social media and digital platforms are not inherently inimical to human flourishing. The impulse to connect, share, and find community online is genuine and valuable. What this paper has tried to show is that the infrastructure through which those impulses are currently channelled is designed primarily to serve commercial objectives, and that this design has consequences for autonomy, knowledge, and democratic life that deserve far more sustained attention than they presently receive.
Acknowledgement- Tejaswani
References
Lyon, D. (2018). The Culture of Surveillance: Watching as a Way of Life. Polity Press.
