What Do We Mean by Gentle Technology? A Relational Lens within Technology Ethics

Abstract

Technologies do more than perform functions. They structure choices, distribute power, remember human histories and shape the conditions under which people can act, refuse, recover from mistakes and affect others.

This paper introduces gentle technology as a relational lens within technology ethics. It proposes that the ethical quality of a technological relationship becomes especially visible when an interaction does not proceed as expected: during error, uncertainty, vulnerability, refusal and conflict.

Five preliminary dimensions are developed: legibility, proportionality, recoverability, respect for refusal and relational attentiveness. The paper also considers the significance of digital identity and argues that responsibility within technological relationships is mutual but not symmetrical. Users remain moral participants, while designers and institutions bear greater responsibility because they possess greater power to structure choices, incentives and consequences.

The framework is presented as a starting point for contextual evaluation and future research rather than a complete ethical theory or certification standard.

Keywords: gentle technology, technology ethics, artificial intelligence, human–computer interaction, digital identity, relational responsibility, personalization, autonomy, care ethics

1. Introduction

Technologies do not merely complete tasks. They organize possibilities, shape expectations and establish conditions under which people act. A system may make one action effortless and another nearly impossible. It may preserve a person’s work or erase it, explain a decision or conceal it, allow refusal or repeatedly challenge it.

These conditions matter ethically because technology increasingly mediates activities that carry personal and social significance: communication, healthcare, education, work, entertainment, finance and participation in communities.

Existing approaches in technology ethics examine values embodied in design, the political character of artifacts, the mediation of morality, human autonomy, care, accessibility, privacy, safety and justice. This paper develops a complementary relational question:

What kind of relationship does a technological system create with the people who use it and with others affected through it?

The term gentle technology is proposed as a lens for examining this relationship.

Here, gentle does not mean merely pleasant, frictionless, forgiving or aesthetically soft. Nor does it mean that technologies should avoid boundaries, difficult information or necessary restrictions. The concept concerns how technological and institutional power is exercised: whether it remains understandable, proportionate, responsive to context and open to correction.

The quality of this relationship becomes especially visible when an interaction does not proceed as expected. What happens when a person makes an error, does not understand, becomes vulnerable, refuses a request or challenges the system? These moments reveal assumptions that may remain hidden during ordinary use.

This paper has four aims. First, it situates gentle technology within existing work in technology ethics and related fields. Second, it proposes five critical moments through which technological relationships can be examined: error, uncertainty, vulnerability, refusal and conflict. Third, it develops five preliminary dimensions of gentleness: legibility, proportionality, recoverability, respect for refusal and relational attentiveness. Finally, it examines digital identity and argues that responsibility within technological relationships is mutual but not symmetrical.

The framework is conceptual and preliminary. It is intended to support research and ethical reflection, not to provide a universal score or certify a technology as ethical.

2. Conceptual foundations

Gentle technology develops within a broader history of research showing that technologies are neither neutral instruments nor isolated objects.

Winner (1980) argues that artifacts may embody political arrangements and distribute power. Nissenbaum (2001) similarly demonstrates that computer systems embody values through their design, classifications and assumptions. Value Sensitive Design provides methods for identifying and incorporating human values throughout technological development (Friedman, Hendry, & Borning, 2017).

Technologies also mediate moral experience and action. Verbeek (2011) describes how technological artifacts help shape perception, decision-making and behaviour. This perspective shifts ethical attention beyond the intentions of designers and users toward the ways technologies participate in practical situations.

Care ethics contributes a relational understanding of morality. Tronto (1993) emphasizes attentiveness, responsibility, competence and responsiveness within practices of care. These ideas are relevant to technological systems, particularly when systems respond to vulnerability or claim to support human well-being. Yet technological care requires caution: a system may present surveillance, restriction or manipulation as concern for the user.

Relational approaches to autonomy further challenge the idea that autonomy consists only of isolated individual choice. Human agency develops within social relationships, material conditions and structures of power (Mackenzie & Stoljar, 2000). A technically available option may therefore fail to constitute meaningful choice when refusal carries disproportionate costs or when alternatives are deliberately obscured.

Borgmann (1984) examines how technologies organize patterns of engagement with the world, while Weiser and Brown’s (1996) account of calm technology proposes that technologies should move appropriately between the centre and periphery of attention. Calmness may reduce cognitive burden, but gentle technology cannot be defined by calmness alone. A calm interface may still conceal exploitation, and an ethically necessary warning may appropriately demand attention.

These traditions establish essential foundations: technologies embody values, distribute power, mediate action, influence autonomy and participate in relationships of care. Gentle technology does not replace these approaches. It adds a narrower analytical focus:

How is technological power exercised within an ongoing relationship, particularly when the interaction encounters resistance, uncertainty or failure?

3. A relational definition of gentle technology

A technology should not be described as gentle solely because of its appearance, intended purpose or isolated features. Gentleness is a quality of the relationship created among a technological system, the people who interact with it, the institutions that govern it and others affected through its use.

This paper proposes the following preliminary definition:

Gentle technology is a sociotechnical arrangement that exercises its power with discipline by remaining legible, responding proportionately, supporting recovery, respecting refusal and attending to the contexts of those materially affected.

The definition is relational because the same feature may have different ethical meanings in different situations. A reminder may support one person while pressuring another. Personalization may reduce cognitive effort or become a mechanism of surveillance. Friction may create an unnecessary obstacle or protect someone from an irreversible decision.

Gentleness therefore cannot be located entirely within an interface or artifact. It depends on institutional policies, business models, data practices, human support, accessibility, enforcement procedures and the distribution of consequences.

The phrase “exercises its power” does not imply that a machine necessarily possesses intentions or independent moral agency. Technologies participate in shaping action because designers and institutions configure possibilities, defaults, incentives and restrictions through them. Responsibility remains attributable to the relevant human and institutional actors according to their knowledge, power and capacity to intervene.

The phrase “with discipline” is equally important. Gentleness is not the absence of power. Technologies inevitably enable some actions, constrain others and affect how people understand their options. Ethical discipline concerns whether this power is exercised for defensible purposes and whether its use remains intelligible, proportionate and open to challenge.

Gentle technology is therefore not a complete theory of technological ethics. A relationship may appear respectful at the point of interaction while depending on exploitative labour, environmental damage, discrimination or invasive data practices. Gentleness must be evaluated alongside justice, safety, privacy, accessibility, accountability and sustainability.

Its contribution is to direct attention toward the lived and institutional character of technological interaction:

• How does the system respond when a person cannot follow the expected path?
• Can its interpretation of the person be corrected?
• Does it preserve meaningful room for refusal?
• Are restrictions proportionate to the harm they address?
• Who benefits from the interaction, and who bears its hidden costs?
• What forms of recovery remain possible after error, conflict or exclusion?

These questions become especially important during moments in which the preferred interaction breaks down.

4. Five critical moments

The ethical quality of a technological relationship is often most visible when ordinary interaction becomes difficult. Five moments are especially revealing: error, uncertainty, vulnerability, refusal and conflict.

These moments are not mutually exclusive. A person may be uncertain because the system has produced an error, vulnerable because access to an essential service is threatened, or in conflict because their refusal was not respected. The categories function as analytical prompts rather than rigid classifications.

4.1 Error

Errors may be made by users, systems, institutions or several participants at once.

A gentle response to error should not assume that every mistake reflects negligence, incompetence or malicious intent. It should help the affected person understand what happened, preserve recoverable work and provide a path toward correction where possible.

Relevant questions include:

• Does the system explain the error in understandable language?
• Does it distinguish accidental mistakes from deliberate misuse?
• Can the person correct inaccurate information?
• Is unfinished work preserved?
• Are consequences proportionate to the seriousness and recurrence of the error?
• Does the system learn from recurring failures, or repeatedly impose their costs on users?

Designing for error means recognizing that fallibility is an ordinary condition of both human and technological action.

4.2 Uncertainty

Technologies frequently operate under uncertainty, particularly when they predict, recommend, classify or generate information.

A gentle system should not disguise uncertainty as certainty merely to appear competent. It should communicate limitations in a form appropriate to the stakes and to the person’s capacity to act on the information.

Too little explanation may create misplaced trust. Too much technical detail may transfer interpretive burden to the user without improving understanding.

Relevant questions include:

• What is known, estimated or unknown?
• How reliable is the output in this context?
• What consequences could follow from relying on it?
• Is additional verification advisable?
• Can the person understand why the system reached its conclusion?
• Does the system allow uncertainty to be questioned?

The appropriate expression of uncertainty depends on context. A low-stakes entertainment recommendation and a medical decision-support system do not require the same level of explanation or caution.

4.3 Vulnerability

Vulnerability may arise from disability, illness, age, financial precarity, emotional distress, language barriers, limited digital literacy, dependence on a service or an emergency situation. It may be temporary, structural or created by the technology itself.

Gentleness requires attention to vulnerability without reducing a person to it.

A system should ask whether its expected interaction path assumes capacities, resources or stability that not everyone possesses. It should also consider whether personalization, monitoring or intervention claimed to protect vulnerable users may instead increase institutional control.

Relevant questions include:

• Who is likely to experience greater difficulty or harm?
• Does the system provide accessible alternatives?
• Can support be obtained without excessive disclosure?
• Does assistance expand the person’s agency or replace it?
• Are people permanently classified according to a past period of vulnerability?
• Are indirect stakeholders exposed to transferred risk or labour?

A gentle response provides support that is proportionate and revisable rather than silently imposing paternalistic decisions.

4.4 Refusal

Refusal reveals whether consent is meaningful.

A person may decline data collection, reject a recommendation, refuse personalization, pause an interaction or leave a service. A system that technically offers refusal but makes it confusing, punitive or practically impossible does not genuinely respect the decision.

Relevant questions include:

• Is refusal presented clearly?
• Are the consequences explained without intimidation?
• Does the system continue pressuring the person after a decision?
• Are unrelated services withheld to obtain consent?
• Can consent later be withdrawn?
• Can the person leave while preserving data or work that could reasonably be retained?

Respecting refusal does not mean that every requested action must remain available. Institutions may impose legitimate conditions or prevent harmful conduct. The ethical question is whether restrictions are relevant, proportionate and honestly communicated.

4.5 Conflict

Conflict occurs when the interests, interpretations or values of participants diverge.

A user may challenge a moderation decision, dispute an automated classification or use a system in ways that harm others. Institutions may need to protect participants, enforce rules or restrict access.

Gentleness does not require avoiding judgement. It requires that judgement remain contextual, explainable and proportionate.

Relevant questions include:

• Can the affected person understand the decision?
• Is there a meaningful opportunity to provide context?
• Are repeated and severe harms distinguished from isolated mistakes?
• Are the interests of other affected people considered?
• Is appeal available where appropriate?
• Can access be restored under reasonable conditions?
• Does enforcement protect safety without becoming humiliating or arbitrary?

Conflict demonstrates why responsibility cannot be assigned only to the immediate user or only to the system. The wider institutional arrangement determines whose account is heard, which harms are recognized and what possibilities for repair remain.

5. Five dimensions of gentleness

The five critical moments show when the ethical quality of a technological relationship becomes especially visible. The following five dimensions describe qualities that may support gentleness across those moments: legibility, proportionality, recoverability, respect for refusal and relational attentiveness.

These dimensions overlap and may sometimes conflict. Increasing legibility may create cognitive burden. Supporting recovery may conflict with privacy or security. Attending to context may require information that should not be collected. They should therefore guide contextual judgement rather than function as automatic requirements.

5.1 Legibility

Legibility concerns whether people can form an adequate understanding of the system, the relationship and the consequences of their actions.

A legible technology does not require every person to understand its full technical architecture. It provides the information necessary for meaningful participation: what the system is doing, what information it uses, why an important decision occurred, what uncertainty remains and what options are available.

Legibility includes more than transparency through disclosure. Information must be accessible, timely and relevant. A lengthy policy that technically contains an explanation but cannot realistically be understood does not create meaningful legibility.

5.2 Proportionality

Proportionality concerns whether technological and institutional responses correspond to the seriousness, likelihood and context of the relevant harm.

Warnings, restrictions, verification requirements and sanctions may be justified. Yet they should not exceed what is reasonably necessary for their legitimate purpose.

A proportionate system distinguishes among accidental error, misunderstanding, repeated misconduct and deliberate harm. It considers the stakes of the situation, the vulnerability of affected participants and the reversibility of consequences.

5.3 Recoverability

Recoverability concerns whether people can repair mistakes, correct interpretations and regain meaningful participation after disruption.

Technologies frequently treat efficiency during ordinary use as more important than recovery after failure. Yet a system’s ethical character may depend more on what happens after something goes wrong.

Recoverability may include undo functions, preserved progress, correction of inaccurate data, accessible support, appeal procedures, reinstatement and data portability. Not every consequence can or should be reversed, but irreversible outcomes require stronger justification and clearer safeguards.

5.4 Respect for refusal

Respect for refusal concerns whether a person can meaningfully decline, pause, withdraw or leave.

Refusal should not be treated automatically as an obstacle to overcome. Repeated prompts, hidden decline options, punitive defaults and artificial urgency may convert nominal choice into pressure.

Respecting refusal includes recognizing that people may change their minds. Consent should therefore be revisable where possible, and the system should distinguish between a temporary pause, rejection of a particular feature and departure from the entire relationship.

5.5 Relational attentiveness

Relational attentiveness concerns whether a system recognizes the relevant contexts, dependencies and differences among those affected.

It asks not only what an individual clicks or selects, but what conditions shape that action. Language, disability, financial constraint, prior experience, social pressure and dependence on a service may all affect whether an interaction is genuinely voluntary or accessible.

Attentiveness should not become unlimited data collection. A system does not need to know everything about a person in order to treat them responsibly. The ethical task is to identify which contextual differences materially matter, use the least intrusive means available and allow people to correct or reject the system’s interpretation.

Relational attentiveness also extends beyond the immediate user. A convenient interaction for one person may transfer hidden labour, risk or environmental cost to others. The relationship must therefore be evaluated across the broader sociotechnical arrangement.

6. Digital identity and relational responsibility

Digital technologies increasingly retain traces of human activity: conversations, creative work, preferences, relationships, reputation, purchases, locations and patterns of behaviour. These traces may become practically and emotionally significant.

A digital profile should not be treated as a complete representation of a person. Data are selective, context-dependent and vulnerable to error. Inferences may become outdated, and behaviour recorded in one situation may be misinterpreted in another.

At the same time, it would be inadequate to describe digital identity as merely disposable information. A platform may contain years of work, social relationships or participation in a community. Losing access may therefore affect more than convenience.

Gentle technology should recognize both claims:

• digital identity may possess genuine personal and social significance;
• no technological representation should become a permanent or total definition of the person.

This creates several responsibilities for institutions. People should be able to inspect and correct important information where feasible. Significant automated interpretations should be open to challenge. Historical data should not silently determine every future interaction. Removal, suspension or loss of access should be proportionate to the relevant harm and attentive to the value stored in the account.

These responsibilities do not imply unlimited rights to remain on any platform or preserve every form of access. Services may need to remove harmful content, restrict conduct or protect other participants. Yet such decisions should consider what is being lost, whether relevant data can be exported and whether a path toward repair exists.

Responsibility within technological relationships is mutual but not symmetrical.

Users remain moral participants. They are responsible for how they use technological capabilities, how their actions affect others and whether they exploit, deceive or harm through a system. Describing technology as morally influential should not erase human agency.

However, institutions and designers usually possess greater power to shape the relationship. They determine defaults, incentives, classifications, available choices, enforcement mechanisms and the distribution of visibility and risk. Greater structural power creates greater responsibility for the conditions under which others act.

Responsibility should therefore be distributed according to knowledge, power, capacity to intervene and contribution to the relevant harm. It should not be assigned automatically to the person closest to the visible outcome.

7. From principle to practice

Gentle technology can inform design, institutional governance and research. Its purpose is not to prescribe a single interface style but to create questions that can be examined throughout the life of a technological system.

7.1 Design and development

During design and development, teams may ask:

• What forms of error are foreseeable?
• Which actions are difficult or impossible to reverse?
• Where might users misunderstand the system’s capabilities or certainty?
• How can refusal be made clear and meaningful?
• Which groups are disadvantaged by the expected interaction path?
• What contextual information is genuinely necessary?
• Who performs hidden labour to make the interaction convenient?
• What happens to a person’s work, data or relationships if access is lost?

Prototypes should include failure, refusal and recovery paths rather than presenting only the ideal interaction. Accessibility testing and participatory methods may reveal forms of vulnerability or transferred burden that designers do not anticipate.

7.2 Institutional governance

Many conditions of gentleness cannot be implemented through interface design alone.

Appeal procedures, moderation policies, customer support, data retention, subscription cancellation, account suspension and access to human review are institutional decisions. A clear explanation cannot compensate for a policy that remains substantively disproportionate.

Organizations should therefore examine whether their incentives support or undermine gentleness. A business model based on maximizing attention, discouraging departure or collecting as much personal data as possible may conflict with respect for refusal and relational attentiveness.

7.3 Research and evaluation

The framework can support qualitative and quantitative investigation.

Possible methods include interviews, usability studies, participatory workshops, policy analysis, incident review, accessibility research and controlled experiments. Relevant outcomes may include understanding, perceived pressure, recovery success, appeal accessibility, trust calibration, error correction and the distribution of burden among stakeholders.

Evaluation should include ordinary use but also deliberately examine the five critical moments. A system may perform well when everything proceeds according to plan while failing precisely when ethical support is most necessary.

The framework should not be reduced too quickly to a single gentleness score. Aggregation may conceal serious failures: strong performance in visual clarity cannot compensate for the absence of meaningful refusal or the transfer of harm to workers.

8. A preliminary evaluation matrix

The framework may be used as a matrix connecting the five critical moments with the five dimensions of gentleness. The purpose of the matrix is not to produce a universal score, but to help identify questions that might otherwise remain unnoticed.

The matrix makes several tensions visible. A highly legible intervention may still be disproportionate. A recoverable system may nevertheless undermine refusal through persistent pressure. Attentiveness to context may improve fairness while also creating privacy risks if it depends on excessive data collection.

Critical moment	Legibility	Proportionality	Recoverability	Respect for refusal	Relational attentiveness
Error	Can the person understand what happened?	Do the consequences correspond to the error?	Can work, access or accurate information be restored?	Can the person reject an incorrect interpretation?	Does the response consider context and differing capacities?
Uncertainty	Are limitations and confidence communicated meaningfully?	Is the level of caution appropriate to the stakes?	Can decisions be reconsidered when new information appears?	Can the person decline or seek another form of guidance?	Is uncertainty communicated appropriately for those affected?
Vulnerability	Are risks, options and support understandable and accessible?	Is intervention limited to what is reasonably necessary?	Can a temporary condition or classification later be revised?	Can assistance be declined without disproportionate cost?	Does support respond to relevant needs without reducing the person to them?
Refusal	Are the available choices and consequences clear?	Are any restrictions resulting from refusal justified?	Can consent be withdrawn or later renewed?	Is the decision accepted without manipulation or repeated pressure?	Do dependency, accessibility and power differences affect whether refusal is meaningful?
Conflict	Can the decision and its basis be understood?	Are enforcement measures appropriate to the harm?	Are correction, appeal or restoration possible?	Can the person challenge the system’s interpretation?	Are the interests and contexts of all materially affected participants considered?

For this reason, evaluation requires judgement rather than mechanical compliance. The relevant questions, evidence and acceptable trade-offs will differ across domains. Healthcare, entertainment, education, finance and social communication involve different stakes and relationships of dependence.

The matrix should therefore be treated as a starting structure for inquiry. Its usefulness depends on participation by those affected, attention to institutional incentives and willingness to examine harms beyond the immediate interface.

9. Limitations and future research

The concept of gentle technology has several limitations in its present form.

First, the framework remains broad and requires adaptation to particular domains. Legibility, proportionality and recoverability will have different meanings in healthcare, education, entertainment, finance and public infrastructure. Future research should develop domain-specific interpretations without losing the relational character of the framework.

Second, gentleness is partly dependent on perspective. An intervention experienced as supportive by one person may feel intrusive or paternalistic to another. Participatory research involving users, non-users, workers and indirectly affected communities is therefore necessary.

Third, tensions among the dimensions cannot always be eliminated. Security may limit recoverability, privacy may constrain contextual attentiveness and protection from harm may justify certain restrictions on refusal. The framework identifies these tensions but does not resolve them automatically.

Fourth, relational qualities at the interface may conceal wider injustice. A system may communicate respectfully while relying on exploitative labour, environmentally damaging infrastructure or discriminatory institutional practices. Gentle technology must therefore remain connected to broader analyses of justice, sustainability, safety and political power.

Fifth, the language of gentleness may itself be misused. Organizations may adopt calm aesthetics or caring language while preserving manipulative business models and disproportionate control. Gentleness should be evaluated through practices, incentives and consequences rather than branding or declared intention.

Future research could examine:

• how people from different social and cultural contexts interpret technological gentleness;
• whether the five dimensions can be operationalized without reducing them to superficial metrics;
• how gentle responses affect trust, autonomy, understanding and willingness to seek correction;
• how institutional incentives enable or undermine recoverability and respect for refusal;
• how the framework applies to artificial intelligence, public services, healthcare technologies, games and digital platforms;
• how the interests of workers, communities and other indirect stakeholders can be incorporated into evaluation;
• when technological intervention becomes appropriately protective and when it becomes paternalistic.

Empirical research may also reveal that the proposed moments and dimensions require revision. The framework should remain open to correction by the people and communities whose technological relationships it seeks to understand.

10. Conclusion

Technology ethics often asks what systems do, which values they embody and what consequences they produce. This paper has proposed an additional relational question: how does a technology exercise power within its relationship with people, particularly when interaction departs from the expected path?

Gentle technology has been defined as a sociotechnical arrangement that exercises its power with discipline by remaining legible, responding proportionately, supporting recovery, respecting refusal and attending to the contexts of those materially affected.

The framework directs attention to five critical moments: error, uncertainty, vulnerability, refusal and conflict. These moments reveal whether a system treats people as meaningful participants capable of understanding, questioning, correcting and leaving—or primarily as objects to be classified and directed.

Gentleness does not mean unlimited permissiveness, aesthetic softness or the absence of boundaries. Technologies and institutions may need to warn, restrict or intervene. The ethical demand is that such power be exercised for defensible purposes, in proportion to relevant harms and with meaningful possibilities for explanation, challenge and repair.

Responsibility within technological relationships is mutual but not symmetrical. Users remain responsible for how they act through technological systems, while designers and institutions bear greater responsibility for the structures, incentives and consequences they create.

The concept is not offered as a certification label or substitute for justice, safety, privacy, accessibility and sustainability. It is a starting point for examining a dimension of technological ethics that can disappear behind functionality and efficiency: the quality of the relationship itself.

A technology becomes most ethically visible when something goes wrong. How it responds in that moment shows not only what the system can do, but how it understands the human being with whom it interacts.

References

1. Borgmann, A. (1984). Technology and the character of contemporary life: A philosophical inquiry. University of Chicago Press.

2. Friedman, B., Hendry, D. G., & Borning, A. (2017). A survey of value sensitive design methods. Foundations and Trends® in Human–Computer Interaction, 11(2), 63–125. https://doi.org/10.1561/1100000015

3. Mackenzie, C., & Stoljar, N. (Eds.). (2000). Relational autonomy: Feminist perspectives on autonomy, agency, and the social self. Oxford University Press.

4. Nissenbaum, H. (2001). How computer systems embody values. Computer, 34(3), 118–120. https://doi.org/10.1109/2.910905

5. Tronto, J. C. (1993). Moral boundaries: A political argument for an ethic of care. Routledge.

6. Verbeek, P.-P. (2011). Moralizing technology: Understanding and designing the morality of things. University of Chicago Press.

7. Weiser, M., & Brown, J. S. (1996). The coming age of calm technology. Xerox PARC.

8. Winner, L. (1980). Do artifacts have politics? Daedalus, 109(1), 121–136.

Disclosure of AI assistance

ChatGPT, developed by OpenAI, was used during the preparation of this article to support literature discussion, conceptual development, structural organization, language refinement and editing. The author reviewed the cited sources, critically evaluated the arguments and retains responsibility for the article’s ideas, interpretations and final text.