Share E-Book

AI Interfaces and the Redesign of Agency (Donghee Shin)(Z-Library)

Author

Rating 3.5 AI initial rating

Log in to rate

AI
Language English

In contemporary life, people increasingly encounter AI not as an autonomous intelligence or distant technical infrastructure, but as an interface that quietly organizes perception, judgment, and action. Recommendation feeds shape what comes into view and what is ignored, generative systems influence how questions are asked and how answers appear, and automated labels, scores, and rankings signal what counts as credible, risky, or acceptable. Navigation tools determine routes, hiring platforms order candidates, and decision support systems suggest outcomes, often without issuing explicit instructions. Individuals are rarely coerced or commanded; instead, they move through environments of preconfigured options that present themselves as neutral, efficient, and helpful. This book argues that such interfaces are reshaping the conditions of human agency by positioning people as interface users who choose among alternatives already filtered, ranked, and framed by AI systems. Freedom is not removed, but reorganized, as choice persists within environments where scope, visibility, and friction are designed in advance. By shifting attention away from models, data, and abstract ethics toward lived experience, the book conceptualizes interfaces as architectures of action that redistribute authority, normalize automated trust, and subtly displace responsibility. Its central claim is that the most consequential power of AI today operates neither through command nor persuasion, but through interface design that quietly reorganizes how people decide, act, and understand their own choices.

Format PDF
Size 3.1 MB
(First 20 pages)

Registered users can read the full content for free

Register as a Gaohf Library member to read the complete e-book online for free and enjoy a better reading experience.

Page 1
AI Interfaces and the Redesign of Agency The Architecture of Choice Donghee Shin
Page 2
AI Interfaces and the Redesign of Agency
Page 3
Donghee Shin AI Interfaces and the Redesign of Agency The Architecture of Choice
Page 4
ISBN 978-3-032-39277-0 ISBN 978-3-032-39278-7 (eBook) https://doi.org/10.1007/978-3-032-39278-7 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2026 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. Unless otherwise expressly agreed by the Publisher in writing, the Publisher does not authorize use or reproduction of all or any part of this work in any manner for the purpose of training, fine-tuning, evaluating, or developing artificial intelligence models, technologies, or systems. In accordance with Article 4(3) of the EU Directive 2019/790 on copyright and related rights in the Digital Single Market, the Publisher expressly reserves all rights with respect to this work under the text and data mining exception of the directive. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Palgrave Macmillan imprint is published by the registered company Springer Nature Switzerland AG. The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland If disposing of this product, please recycle the paper. Donghee Shin KIMEP University Almaty, NA, Kazakhstan
Page 5
v Competing Interests The author has no competing interests to declare that are relevant to the content of this manuscript.
Page 6
vii Contents 1 Designed Freedom: Human Agency in the Age of Algorithmic Interfaces 1 2 Algorithmic Interfaces: Mediating Human-AI Interaction 7 3 Reconfiguring Agency in Algorithmic Interface 21 4 Algorithmic Authority and Epistemic Trust 37 5 Agentic AI and Interface 57 6 Synthetic Intimacy: Relational AI and the Interface of the Self 79 7 Conclusion: Living in Designed Freedom 93
Page 7
1© The Author(s), under exclusive license to Springer Nature Switzerland AG 2026 D. Shin, AI Interfaces and the Redesign of Agency, https://doi.org/10.1007/978-3-032-39278-7_1 CHAPTER 1 Designed Freedom: Human Agency in the Age of Algorithmic Interfaces Abstract Contemporary AI is reframed as an interface rather than a dis- crete tool, with algorithmic systems shown to exercise power through design and curation, not command. The concept of “designed freedom” is developed—autonomy that feels self-directed but is pre-structured by invisible choice architectures—alongside an account of how frictionless interfaces erode deliberation via automation bias. Calling for a shift from post-hoc transparency toward “interface accountability” and Design Ethics, the discussion sets up the book’s central question: how human agency and cognitive sovereignty can be reclaimed within environments engineered to preemptively narrow choice. Keywords Designed freedom • Algorithmic interface • Choice architecture • Interface accountability The power of contemporary AI lies not in algorithmic computation itself, but in the design of interfaces that pre-structure the conditions under which human choice, judgment, and selfhood are formed.
Page 8
2 The Fallacy oF The DiscreTe Tool For decades, both scholarly discourse and public imagination have been governed by an instrumentalist conception of technology, the assumption that computers, software, and even advanced neural networks are neutral tools deployed by humans to achieve predefined ends (Miller, 2021). Within this paradigm, the human user is the autonomous pilot and tech- nology merely the vessel. We use AI to draft emails, query models for infor- mation, and deploy algorithms to sort data. Yet as AI has migrated from back-end computation to the foreground of everyday experience, this framing has become increasingly untenable. To continue treating contem- porary AI as a discrete tool is not merely misleading. It is a categorical error. AI systems today no longer function as instruments we simply pick up and put down. Instead, they operate as interfaces through which per- ception, action, and meaning-making are organized. They are not bounded objects but pervasive mediation layers. When an algorithm determines which news appears in a feed, which route a driver follows, or which appli- cant appears at the top of a hiring dashboard, it is not merely assisting a decision. It is actively constructing the conditions under which that deci- sion becomes possible (Shin, 2024). This shift reframes AI as an ambient structure. Unlike overt authorities that govern through instruction or command, ambient structures exert power by shaping environments so seamlessly that their influence feels natural, even inevitable. AI is no longer a discrete resource we selectively employ; it has become a pervasive mediation layer that edits our reality in real-time. In this pro- cess, the interface produces a state of “designed freedom.” By subtly dis- tributing attention and framing choices, it creates an environment where users feel they are exercising autonomous agency, while their paths are, in fact, pre-structured by invisible design logics. The algorithmic interface is the primary site where digital power is enacted. Addressing this shift requires a move toward “interface accountability,” demanding a rigorous examination of how these architectures distribute power and reconfigure human agency. Within the seamless convenience of the algorithmic wrap, we must continually question how to maintain the critical distance neces- sary for true autonomous thought. By redirecting analytical attention from the hidden “black box” of the model to the visible surface of the interface, we can begin to identify where algorithmic power is actually exercised and where it so often escapes scrutiny (O’Reilly et al., 2024). D. SHIN
Page 9
3 The locus oF Power: Design as insTrucTion If we look for power only at the moment of a final algorithmic output, we miss the most consequential phase of influence: the design of the interface itself. Power in the age of AI is not exercised through “command,” but through “curation.” As Winner (1980) famously argued, technological artifacts possess inherent political qualities based on their design; AI inter- faces are the modern manifestation of this politics of design (Kalluri, 2020). Consider the ordinary interface user. This individual interacts with a digital environment where rankings, defaults, and confidence cues serve as silent instructions. When a search engine ranks a result at the top, it isn’t forcing a click; it is making that click the path of least resistance (Sunstein, 2017). When an AI assistant provides a “confidence score,” it is not providing raw data, but a psychological nudge toward trust (Shin, 2021). In these envi- ronments, design logics make certain choices feel natural, efficient, or inev- itable. This is power through design. Foregrounding interfaces rather than underlying models also sharpens debates about AI governance. The harm embedded in biased AI is rarely a single wrong answer; it is the systemic narrowing of the user’s perceptual horizon through curated options, a pruning of possibility that precedes any individual choice (Shin, 2025). The erosion oF DeliberaTion: auTomaTion bias anD ambienT sTrucTure One of the primary psychological consequences of the AI-as-interface shift is the rise of automation bias, the tendency for humans to favor suggestions from automated systems even when they contradict human judgment (Shin, 2025). As AI becomes an ambient structure, it creates a “frictionless” expe- rience. Friction in human thought is often where deliberation happens; it is the moment of doubt that forces us to think critically (Kahneman, 2011). By removing friction, the AI interface effectively bypasses the deliberative process. Users comply with algorithmic outcomes without the need for coercion or explicit instruction (Ananny, 2024). The most profound impact of large language models is not what they do for us, but what they do to us (Shin, 2023). By organizing our perception and preemptively narrowing our options, these systems quietly reshape the character of judgment itself, gradually transitioning a society of deciders into a society of clickers navigat- ing a world whose architecture has already been decided. 1 DESIGNED FREEDOM: HUMAN AGENCY IN THE AGE OF ALGORITHMIC…
Page 10
4 The ParaDox oF FreeDom: choice wiThin consTrainTs A central question of this book is: Why does freedom persist within AI-mediated environments while simultaneously feeling structured or con- strained by design? This is the paradox of the modern interface user. We feel free because we are still the ones making the final selection. However, that freedom is exercised within a “choice architecture” that has already done the heavy lifting of exclusion (Thaler & Sunstein, 2008). If an inter- face presents three options out of a possible thousand, the user’s “free- dom” to choose among those three is a “managed autonomy.” This section explores how design mediates the tension between autonomy and automation. It argues that the “feeling” of freedom is often used to mask the reality of constraint. To understand AI governance, we must move beyond the “right to an explanation” and toward a framework that pro- tects human judgment from being eroded by the convenience of algorith- mic defaults (Shin, 2024). TowarD a new Framework The argument culminates in a fundamental reframing of responsibility: How can human agency be reclaimed in environments whose very archi- tecture has been engineered to preemptively narrow choice? If algorith- mic power has indeed migrated from the hidden black box to the visible surface of the interface (Ananny, 2024), then ethical responses limited to post-hoc transparency, model audits, or explainability requirements are insufficient. What is required instead is a turn toward Design Ethics. The ethical question is no longer simply, “Is the AI accurate?” but rather, “Is the AI-mediated world one in which human autonomy can meaning- fully flourish?” (Shin, 2025). Reclaiming cognitive sovereignty will not be achieved through better data alone or more efficient models. It demands the deliberate reinsertion of deliberative friction—design interventions that interrupt automation bias and compel users to reengage in active, reflective judgment (Kahneman, 2011). This book offers a theoretical and empirical roadmap for ensuring that technological innovation does not come at the expense of self-determination. By re-centering the interface as the primary site of governance, we can begin to move from passive interface users toward what we might call cognitive citizens, individuals capable of navigating, questioning, and contesting the architectures of an D. SHIN
Page 11
5 increasingly automated world. The chapters that follow demonstrate how this shift can be operationalized across domains ranging from public dis- course to technical communication. reFerences Ananny, M. (2024). Making generative artificial intelligence a public problem. Javnost  - The Public, 31(1), 89–105. https://doi.org/10.1080/1318322 2.2024.2291986 Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux. Kalluri, P. (2020). Don’t ask if artificial intelligence is good or fair, ask how it shifts power. Nature, 583(7815), 169. https://doi.org/10.1038/ d41586- 020- 02003- 2 Miller, B. (2021). Is technology value-neutral? Science, Technology, & Human Values, 46(1), 53–80. https://doi.org/10.1177/0162243919900965 O’Reilly, T., Strauss, I., & Mazzucato, M. (2024). Algorithmic attention rents: A theory of digital platform market power. Cambridge University Press. Shin, D. (2021). The effects of explainability and causability on perception, trust, and acceptance: Implications for explainable AI. International Journal of Human-Computer Studies, 146, 102551. https://doi.org/10.1016/j. ijhcs.2020.102551 Shin, D. (2023). Algorithms, humans, and interactions: How do algorithms interact with people? Designing meaningful AI experiences. Routledge. https://doi. org/10.1201/b23083 Shin, D. (2024). Artificial misinformation: Exploring human-algorithm interac- tion online. Palgrave Macmillan. Shin, D. (2025). Debiasing AI: Rethinking the intersection of innovation and sus- tainability. Routledge. Sunstein, C. R. (2017). #Republic: Divided democracy in the age of social media. Princeton University Press. Thaler, R. H., & Sunstein, C. R. (2008). Nudge: Improving decisions about health, wealth, and happiness. Yale University Press. Winner, L. (1980). Do artifacts have politics? Daedalus, 109(1), 121–136. 1 DESIGNED FREEDOM: HUMAN AGENCY IN THE AGE OF ALGORITHMIC…
Page 12
7© The Author(s), under exclusive license to Springer Nature Switzerland AG 2026 D. Shin, AI Interfaces and the Redesign of Agency, https://doi.org/10.1007/978-3-032-39278-7_2 CHAPTER 2 Algorithmic Interfaces: Mediating Human-AI Interaction Abstract Beyond serving as a neutral window into computational sys- tems, the algorithmic interface is a mediated space defined by choice archi- tecture and sociotechnical affordances. This analysis identifies a fundamental departure from traditional human-computer interaction; we are moving from static representations toward dynamic, adaptive filtra- tion. Instead of presenting a fixed set of options, contemporary interfaces deliver curated, personalized content driven by machine learning and behavioral histories. By theorizing the interface as a convergence of tech- nical design, social practices, and economic imperatives, this work distin- guishes it from adjacent layers like platforms or digital infrastructure. The interface acts as a primary mechanism of digital power, translating opaque processes into actionable experiences that recursively shape user cognition and behavior. Keywords Algorithmic interface • Choice architecture • Epistemic autonomy • Sociotechnical mediation
Page 13
8 The inTerFace as a concePTual FounDaTion To understand the unique power of contemporary digital systems, one must first unpack the conceptual history of the interface itself. Historically, the interface was treated as a passive boundary, a terminal endpoint, or a neutral window: a transparent pane of glass designed simply to reveal the underlying architecture of a computational system. In early human- computer interaction (HCI) paradigms, which borrowed heavily from the ergonomics of industrial design, the interface operated under a strict rep- resentational logic. It was imagined as a stable, spatial map of static func- tions, where available actions corresponded directly and predictably to visible options. Under this legacy framework, user interaction was guided by ideals of structural transparency, consistency, and absolute user com- mand. The interface was assumed to be an objective, flat space through which a human agent issued direct, unmediated instructions to a machine, which then executed them with passive fidelity. However, a critical turn in media theory and software studies reveals that the interface was never merely a passive skin or a neutral viewport. It has always been an active, mediated space and a primary site of choice architecture. An interface does not merely show; it fundamentally orga- nizes, categorizes, and translates. Long before the advent of machine learning, the design of a user interface required an intentional distribution of visibility. By choosing what to foreground on a desktop, what to rele- gate to an obscure sub-menu, what to color-code as urgent, and how to sequence interaction paths, traditional interfaces have always structurally delimited the conditions of human visibility and action. They did not sim- ply mirror system capabilities; they curated the user’s operational reality, quietly nudging attention and defining what was possible within the system. Consequently, the transition to an algorithmic interface does not invent mediation out of whole cloth; rather, it weaponizes and dynamizes it. Traditional interfaces, despite their selective framing, remained culturally legible as fixed workspaces—their choice architectures were rigid, uni- form, and visible to all users alike. The algorithmic mutation fundamen- tally shatters this stability. By injecting automated, probabilistic, and real-time predictive modeling into this interactional layer, the interface shifts from a static tool into a highly volatile, responsive environment. It transitions from a passive canvas into an active interlocutor. This algorithmic layer establishes an ongoing cybernetic feedback loop, where the system constantly reads the user’s micro-behaviors to instantly D. SHIN
Page 14
9 reconstitute the visual environment. The crucial epistemic shift is that the interface now actively pre-structures user perception before the user even formulates a conscious intent or executes an explicit command. By pre- dicting desire and preemptively flattening alternative pathways, the algo- rithmic interface ceases to be a tool used by a human; it becomes an immersive architecture that shapes the human who uses it. Crucially, the user “freedom” native to these hyper-adaptive environ- ments must not be conflated with mere convenience. While frictionless design, predictive text, and hyper-personalized recommendations are fre- quently marketed by platforms as liberating or empowering, they function as a highly optimized, restrictive choice architecture. Convenience is often the primary mechanism through which algorithmic power operates; by systematically removing the cognitive friction of searching, evaluating, and deciding, the interface renders user behavior deeply predictable and compliant. Genuine digital freedom, therefore, must be conceptualized far beyond superficial usability metrics or the ease of clicking a pre-selected link. It demands an examination of epistemic autonomy—the structural capacity of the human subject to recognize, resist, and look behind the invisible, predictive filtration structures that condition their reality in the first place. If freedom is reduced to the “convenience” of choosing Option A or Option B from a curated feed, the user yields their algorithmic sovereignty to a system that has already decided what is worth seeing. DeFining The algoriThmic inTerFace To define the algorithmic interface, one must first reject the long-standing assumption that interfaces function as neutral windows onto computa- tional systems. This assumption, deeply embedded in earlier paradigms of human–computer interaction, rests on the idea that interfaces transpar- ently represent underlying system capabilities. Under such representa- tional logic, the interface was imagined as a stable map of functions, where available actions corresponded directly to visible options and where user interaction was guided by transparency and predictability (Oeldorf-Hirsch et al., 2025). Contemporary algorithmic environments, however, fundamentally dis- rupt this logic. In algorithmically mediated systems, the interface no lon- ger reflects the system’s total capacity; instead, it presents a contingent, dynamically generated subset of possibilities shaped by ongoing predictive 2 ALGORITHMIC INTERFACES: MEDIATING HUMAN-AI INTERACTION
Page 15
10 modeling and contextual inference. What is presented to the user is not a comprehensive option set, but a curated slice of the system that continu- ously adapts in response to prior interactions and inferred user intent. This shift marks a transition from representation to algorithmic filtration, in which visibility itself becomes an outcome of machine learning and data- driven inference (Kitchin, 2021; Seaver, 2022). This transformation is both epistemic and experiential. Under algorith- mic mediation, users do not encounter information as a given external resource; they encounter a continuously constructed informational field shaped by adaptive logics. The interface does more than present content; it participates in producing what counts as relevant, salient, or actionable knowledge. Consequently, the relationship between user and system is reconfigured: users are situated within an interactional space where each interaction feeds back into the system’s adaptive predictions, shaping sub- sequent engagements (Bucher, 2024; Christin, 2020). Chun (2021) simi- larly argues that contemporary machine learning systems train users into forms of predictable behavior through recursive systems of recognition and correlation, transforming prediction into an infrastructure of social organization. The algorithmic interface thus operates as a mediating layer that trans- lates machine-level operations into legible to human user experiences. This translation is neither neutral nor exhaustive. It is structured through probabilistic inference, continuously adapting to inferred user preferences, behavioral histories, and contextual signals. Machine learning systems operate behind the scenes to rank, filter, and prioritize content, but it is the interface that renders these operations perceptible and actionable (Rader & Gray, 2015; Seaver et al., 2025). Importantly, this translation involves reduction of complexity. Computational systems process data across multiple dimensions and scales, yet the interface must compress these outputs into forms that users can process cognitively, affectively, and visually. In doing so, it foregrounds some elements while backgrounding or omitting others. This selective reduction is not incidental; it is constitu- tive of how the interface operates. Users therefore encounter not the full scope of algorithmic reasoning, but a selective representation calibrated toward usability engagement, and strategic goals over transparency (Eslami et al., 2019; Sundar et al., 2025). Simultaneously, the interface actively shapes how algorithmic processes are experienced and understood. Through design elements such as layout hierarchies, visual affordances, feedback cues, and interactive prompts, the D. SHIN
Page 16
11 interface directs user attention, signals importance, and structures sequences of possible actions. These elements work together with algo- rithmic outputs to produce a coherent experiential environment that shapes interpretation and response (Seaver, 2022; Walter & Friesike, 2025). This framing function has significant implications for user cognition and behavior. By organizing information in particular ways, interfaces influence how users interpret content, assess relevance, and make deci- sions. Prominently displayed items may be perceived as more credible or important regardless of underlying criteria, while the absence or obscurity of certain options can create the false impression that the visible choices are exhaustive. The interface thus participates in shaping not merely user action, but user perception and judgment (Bruineberg, 2025; Weinmann et al., 2016). Moreover, the adaptive nature of algorithmic interfaces introduces an important temporal dimension to this process. Unlike static interfaces, which present fixed structures to all users, algorithmic interfaces evolve over time in response to ongoing interaction. Each click, pause, or scroll contributes to a feedback loop that refines the system’s predictions and, consequently, the interface’s future configurations. This recursive dynamic blurs the boundary between system and user, as each interface iteration increasingly reflects a constructed model of the user shaped by prior engagement (Veale et al., 2018; Rieder et al., 2023). This temporal fluid- ity creates a state of perpetual variation: the interface is never finished. It exists as an active, unfolding dialogue where human habituation and machine optimization continuously rewrite the visual and interactive boundaries of the digital workspace. In this sense, the algorithmic interface functions as both reflective and generative. It reflects patterns of user behavior as interpreted by the sys- tem, but it also generates new patterns by structuring subsequent interac- tions. Users are not merely responding to the interface—they are being shaped by it, as the range of visible and actionable options is pre-curated to align with predicted interests or institutional priorities. This recursive shaping underscores the need to understand the interface not as a static object, but as a dynamic process of mediation (Bucher, 2024; van Dijck et al., 2021). Taken together, these considerations suggest that the algorithmic inter- face occupies a central and indispensable position in contemporary digital environments. It is the point where complex computational abstraction meets human experience, where data-intensive operations are translated 2 ALGORITHMIC INTERFACES: MEDIATING HUMAN-AI INTERACTION
Page 17
12 into perceptible and actionable forms. By filtering, framing, and adapting information, the interface does more than facilitate access: it actively con- figures the conditions under which knowledge is encountered and action is taken. Any analysis of algorithmic power that neglects the interface risks overlooking the very mechanism through which that power is enacted in everyday life. The inTerFace as a socioTechnical arrangemenT Conceptualizing the algorithmic interface requires moving beyond a nar- row technical definition that treats it as a simple conduit between system and user. Instead, the interface should be theorized as a sociotechnical arrangement, a layered and dynamic site where technical infrastructures, social practices, and economic imperatives converge to produce mediated experiences and shape behavior (Bucher, 2024; van Dijck et al., 2021; Xu & Chen, 2025). This framing foregrounds the interface not as a static artifact but as an active configuration of multiple logics, each influencing how information is presented, interpreted, and acted upon. At the technical level, the interface comprises more than visual compo- nents or interaction menus. It includes code structures, APIs, data pipe- lines, content recommendation algorithms, interactive affordances, and micro-interactions that together render algorithmic outputs into perceiv- able and manipulable forms (Shneiderman et al., 2022). The architecture of these elements determines how information is prioritized, sequenced, and displayed, often embedding implicit assumptions about the order of importance, the salience of content, and the ease of certain actions over others. These design decisions shape the very possibilities for engagement, subtly guiding users toward particular paths while constraining alternative actions. However, technical design alone cannot explain the full influence of the interface. The social dimension is equally constitutive. Users approach interfaces with a range of cognitive schemas, habitual behaviors, cultural norms, and affective orientations that interact with design ele- ments to produce meaning (Ruckenstein & Turunen, 2020). Interfaces do not simply respond to static users; they anticipate patterns of attention, expectation, and emotional engagement, exploiting these tendencies to produce desired behaviors. Users interpret, negotiate, and sometimes resist the cues and affordances embedded in the interface, yet this inter- pretive work always occurs within the constraints and possibilities the interface establishes. In this way, social practices and technical design D. SHIN
Page 18
13 co-produce each other: the interface shapes behavior, while user interac- tion continuously informs the evolution of the interface itself. Overlaying the technical and social layers is the economic logic of the platform, which fundamentally structures interface design. Interfaces are rarely neutral with respect to institutional goals; they are calibrated to optimize performance metrics such as engagement, retention, conversion, or advertisement revenue (Nieborg et al., 2025; Zuboff, 2019). The posi- tioning, visibility, and interactivity of content are often dictated by these economic imperatives rather than user interest or informational relevance alone. In this sense, the interface functions as a site where optimization strategies converge, aligning individual user actions with broader organi- zational objectives. Importantly, the sociotechnical interface is dynamic and adaptive. It continuously integrates user interactions, algorithmic outputs, and perfor- mance metrics to produce a system that is responsive to both behavioral patterns and platform goals. Each user engagement contributes to a feed- back loop, refining the interface’s predictions and shaping the subsequent visibility and accessibility of information (Veale et al., 2018; Rieder et al., 2023). It is both reflective, representing patterns of user activity, and gen- erative, producing conditions that shape future behavior. Finally, conceptualizing the interface as a sociotechnical arrangement highlights its normative and ethical significance. By embedding assump- tions about attention, cognition, and engagement, and by operationaliz- ing economic goals, interfaces exert subtle forms of governance over digital behavior (Bucher, 2024). Understanding interfaces as sites where technical, social, and economic logics intersect allows scholars to trace the mechanisms through which power, influence, and agency are distributed in algorithmic systems, and to identify points of intervention for ethical design and regulatory oversight. This sociotechnical framing positions the algorithmic interface as a cen- tral analytic object. It is not merely a point of interaction but a site of co- production, where code, human behavior, and institutional objectives converge to shape the visibility, interpretation, and enactment of informa- tion in digital environments. 2 ALGORITHMIC INTERFACES: MEDIATING HUMAN-AI INTERACTION
Page 19
14 DisTinguishing The inTerFace From aDjacenT layers Analytical clarity in studying digital ecosystems requires distinguishing the interface from other components that are often subsumed under the gen- eralized label of “the algorithm.” While algorithms, platforms, and infra- structures each exert distinct forms of influence, the interface occupies a unique and comparatively underexamined position within this assemblage, serving as the locus at which computational outputs, institutional logics, and human perception intersect. Algorithms function primarily at the level of data processing. They are responsible for sorting, ranking, and predicting according to defined cri- teria, and their power lies in determining what is computationally relevant or salient (Kitchin, 2021). Platforms, in contrast, establish the broader ecosystem within which these algorithms operate. They define the rules of participation, enforce normative and economic frameworks, and structure the circulation of content and value through governance mechanisms and business models (van Dijck et al., 2021). Infrastructure provides the foun- dational layer, encompassing the physical and digital systems—servers, networks, cloud storage, and transmission protocols—that enable compu- tational operations and access (Xu & Chen, 2025). The interface differs fundamentally from these layers because it consti- tutes the immediate site of encounter between user and system. Its power is primarily interpretive: it shapes how algorithmic outputs are perceived, understood, and acted upon. While an algorithm may identify certain con- tent as highly relevant according to quantitative metrics, it is the interface that determines how this relevance is rendered experientially, whether a recommendation appears as a large, visually salient banner, a small sidebar link, or an obscured option at the bottom of a feed (Bucher, 2018). In this sense, the interface does not merely transmit algorithmic judgments; it frames them, translating computational significance into perceptible affor- dances that guide attention, shape interpretation, and influence behavior. Without this demarcation, critiques of digital ecosystems become mono- lithic. Isolating the interface reveals that system power does not just lie in the code that sorts data (the algorithm), nor in the business models that fund it (the platform), but in the visual and interactive choreography (the interface) that makes those architectures intuitive, desirable, and absolute to the human user. This framing function underscores that influence cannot be understood solely in terms of algorithmic computation or platform governance. The D. SHIN
Page 20
15 interface actively mediates visibility, constructing a perceptual and interac- tional landscape in which certain options appear intuitive and others mar- ginal or inaccessible. By foregrounding some possibilities and backgrounding others, interfaces produce a form of interpretive power that operates alongside, and in conjunction with, the sorting power of algorithms and the institutional power of platforms. The result is a layered ecology of digital influence, where the interface functions as the critical mediator of experience, bridging the technical operations of algorithms with the social and economic logics of platforms. The inTerFace as The siTe oF meDiaTion anD acTion The algorithmic interface functions at the intersection of computational systems and human experience, serving as the surface through which oth- erwise opaque processes are rendered actionable and intelligible. It is at this site that abstract computational operations acquire experiential and perceptual form. In the absence of the interface, algorithmic outputs remain inaccessible to users; conversely, without algorithmic processing, the interface would lack substantive content to present. Their relationship is therefore mutually constitutive, yet asymmetrical with respect to the user: while algorithms operate largely invisibly, the interface is the locus of encounter, interpretation, and action (Bucher, 2024). This centrality gives the interface a distinctive role as an agent of visibil- ity (Bitencourt, 2025). It delineates the boundaries of what can be seen, selected, and acted upon within a given digital environment. By structur- ing the field of available options, the interface does more than facilitate interaction; it actively shapes it. Through the design of interaction mecha- nisms, including buttons, menus, prompts, and feedback cues, the inter- face establishes a form of choice architecture that channels user behavior along particular trajectories, subtly guiding attention, engagement, and decision-making (Lin et al., 2026; Weinmann et al., 2016). In this framework, the interface is also a co-constitutive element in the construction of user agency. It does not determine behavior in a determin- istic fashion, but it organizes the conditions under which decisions are made. By privileging certain pathways and obscuring others, the interface renders specific actions more intuitive, accessible, or frictionless, while simultaneously making alternative options less visible or more cumber- some to enact. This shaping of possible actions represents a subtle yet pervasive form of influence: the interface not only structures what users 2 ALGORITHMIC INTERFACES: MEDIATING HUMAN-AI INTERACTION
The above is a preview of the first 20 pages. Register to read the complete e-book.

Recommended for You

Loading recommended books...
Failed to load, please try again later

Tip the Site

Scan the WeChat Pay or Alipay code to tip. No login required.

WeChat Pay
Alipay
← Back to List