Listening to a teenager in the presence of a robot requires abandoning a script that expects predictable curiosity, obedience, or rejection. Silence, enthusiasm, and discomfort are not responses to classify quickly. They may reveal how the machine is understood, which risk is perceived, and what support is needed. Dignity recognises the young person's voice, safety enables stopping, and repair changes the system when listening fails. The robot is useful only while remaining subordinate to that human relationship.
When the conversation already has a correct answer#
A project may invite teenagers after purchasing technology and ask whether they like it. The question appears participatory but allows only validation or rejection of a closed decision. Someone expressing doubt may be considered fearful; someone showing enthusiasm, unreflective. Listening means opening issues that can still change: purpose, tasks, data, appearance, schedules, alternatives, and withdrawal conditions.
A project may invite teenagers after purchasing technology and ask whether they like it.
Listening becomes an examination when form is valued above content. An immediate, verbal, orderly explanation is expected, while silence, written messages, augmentative communication, or later responses are discarded. A teenager may need more time, sign language, easy-read material, or a lower-stimulation space. Communication accessibility does not provide an advantage; it allows experience to arrive without imitating an adult style.
Questions can also close the result. Asking how to make the robot more attractive avoids discussion of whether it should be used. Asking why a person lacks trust locates the problem in them before sensors, privacy, or earlier experience are examined. Facilitation uses open and concrete prompts: what happened, what helped, what caused discomfort, what would change, and which alternative would be acceptable. Understanding is then checked.
Understanding power, dependency, and freedom#
A teenager may depend on the educational, health, or residential institution proposing the robot. In that relationship, agreement does not guarantee freedom. Refusing interaction should not mean losing support, receiving a sanction, or being labelled uncooperative. Genuine options, consequences, and limits are explained. A person can accept some functions, refuse others, and change their mind without justifying every preference.
Freedom does not mean deciding without support. It includes accessible information, time, a chosen trusted person, and opportunities to ask. Relatives and professionals may contribute perspectives without always speaking for the teenager. Where protective duties or legally constrained decisions exist, they should be explained precisely: what is not negotiable, why, who reviews it, and which choices remain.
A robot's social appearance adds power. A warm voice, animated eyes, or affectionate wording may imply that it understands, keeps secrets, or cares. Explanation distinguishes simulation from human experience. The device identifies itself as a tool, communicates limitations, and refers to a person. It does not use guilt, disappointment, or insistence to obtain interaction. Technological connection never replaces access to human relationships.
Privacy and safety in youth settings#
Sensors may capture voice, image, location, movement, health, routines, and relationships. In a classroom or home, they also record peers and relatives. Before use, people are told which data is collected, for what purpose, where it is processed, who can access it, and when it is deleted. Information is provided in suitable formats without treating a lengthy policy as proof of understanding. Default settings minimise data.
Consent to receive a service does not automatically include research, advertising, or model training. Each secondary purpose needs a legitimate basis and separate choice where required. Emotion or diagnosis should not be inferred from gestures without sufficient evidence. Teenagers can inspect and correct records. Privacy is not presented as the inevitable price of innovation.
Safety includes the ability to stop. Control must be accessible through redundant options, with a human team able to respond. A stop brings no punishment or automatic restart. Test scenarios include noise, irregular connectivity, unexpected movement, and varied bodies. Where risk exceeds controls, the function is limited or withdrawn despite its appeal.
Preparing listening that can modify the project#
Before conversation, purpose, participants, duration, possible decisions, and confidentiality limits are shared. The agenda arrives early and allows additions. Communication preferences are requested without requiring clinical data. Oral, written, and asynchronous participation is possible. The space is accessible and safe, while independent facilitation may be needed where the listening institution also controls the service.
During the meeting, one question at a time and genuine response time are provided. Facts, interpretations, and emotions are separated without automatic hierarchy. A professional does not promise absolute secrecy where safeguarding duties exist; those limits are explained before sensitive information is received. Notes distinguish the teenager's words from adult interpretation. Disagreements remain recorded.
Afterwards, an accessible summary sets out agreements, responsible people, and dates. The teenager can correct it. Proposals receive a reasoned response: accepted, under trial, infeasible, or pending. If nothing can change, consultation was decorative. The institution also explains complaints and how to leave a pilot without losing care. Closing the loop builds trust.
Case 1: a social robot in a classroom#
A school introduces a robot to support social skills. The team expects enthusiasm and records a demonstration. An autistic student avoids interaction, and this behaviour is interpreted as a therapeutic target. Nobody asks whether volume, animated gaze, or cameras cause discomfort. Classmates receive instructions to encourage participation. Listening becomes collective pressure.
The school stops individual use, withdraws the recording, and communicates through the student's chosen channel. Sound, distance, and appearance are adjusted, and an activity without the robot is available. The student chooses to try one function without cameras and with a stop word. Success is measured through educational purpose and wellbeing rather than interaction time. Repair recognises exposure and changes consent for later projects.
Case 2: a conversational assistant for wellbeing#
A youth service provides a conversational robot to ask about mood. A teenager shares a concern believing the exchange is private. The system creates an alert, but nobody explained who would receive it or within what time. Human response arrives late and mentions the conversation in front of others. Although intended to protect, the mechanism removes her control and she stops using the service.
Review separates general support, assessment, and emergency. Before conversation, the system explains limits with examples, offers human contact, and permits omission. Alerts have an owner and deadline; if capacity is unavailable, the function is not activated. Records are minimised and disclosed information corrected. The teenager selects a follow-up channel. Repair does not require returning to the robot. Learning redefines safety as reliable response and understandable confidentiality.
Case 3: robotics in rehabilitation#
A teenager uses a robotic device to support movement after injury. The protocol fixes repetition and pace, while professional questions expect confirmation. He smiles and continues, then later sends a message explaining that he experienced pain and did not know how to stop without disappointing the team. The report records a completed session. Apparent cooperation conceals authority and an unheard signal.
The service pauses progression, assesses pain, and reviews possible harm. A later session, only if chosen, includes unloaded demonstration, accessible control, breaks, and selected pace. The professional uses several scales and formats. Continuing consent is recorded rather than only an initial signature. Metrics include confidence and willingness to return. The device retains a limited therapeutic role under human judgement and adolescent participation.
Dignity in language and documentation#
Records should not automatically describe refusal as resistance, manipulation, or immaturity. Observations, the explanation offered, and the person's perspective are documented. Labels can affect future decisions, so a correction route is necessary. Information is shared only with those who need it. An anecdote is not used in presentations without specific consent.
The robot's language matters too. It should avoid infantilisation, claims of affection, and blame. A reminder can be direct and respectful, offer choice, and accept silence. Personalisation does not justify storing unnecessary intimate details. Jokes, names, and voice are configurable. The teenager can reset preferences and erase a profile.
Dignity includes payment where consultation becomes design work. The same people should not always be asked to educate the team. Participation includes breaks, support, and feedback. Lived knowledge complements engineering and professional practice; it is not a decorative story for legitimising a project.
Safety that learns from disagreement#
Objections and withdrawal are safety data. They are recorded without public identification and investigated for patterns. A low complaint rate does not prove success where the channel is unknown or unsafe. Near misses, human interventions, and questions provide signals. The organisation protects reporters and responds within a stated time.
Stopping criteria are agreed before deployment: harm, persistent distress, data loss, repeated failures, insufficient supervision, inequality, or unauthorised change. Professionals and teenagers know how to pause. The supplier does not decide alone whether operation restarts. Evidence, explanation, and renewed consent where experience changes are required.
Investigation examines the system, not merely the last action. Design, data, environment, training, metrics, staffing, and power are reviewed. A supposed user error may reveal unclear instruction. Repair combines individual support with structural change. Every action has an owner, date, and verification in real use.
Governance with adolescent participation#
A plural body includes diverse teenagers, families, professionals, educators or care workers, engineering, privacy, accessibility, and safeguarding. It defines permitted and prohibited uses. Young participants receive support and genuine authority to change scope. Commercial relationships and conflicts of interest are declared. Decisions are communicated in plain language.
Indicators cover purpose, autonomy, understanding, refusal, alternatives, human contact, workload, incidents, privacy, and repair. Averages are carefully disaggregated to identify inequality without revealing identities. A function may increase completed tasks while worsening wellbeing. Evaluation must be able to limit or withdraw, not only expand.
The life cycle covers purchase, testing, update, maintenance, and exit. A software change can alter voice, sensors, or data use. Re-evaluation follows. When funding or support ends, service continuity and secure deletion or migration are guaranteed. Nobody should depend on a temporary project without an alternative.
Institutional sources for practice#
UNICEF guidance on adolescent participation helps distinguish meaningful involvement from symbolic consultation. The United Nations Convention on the Rights of the Child places expression and consideration of views within a rights framework. These references guide processes adapted to age, context, and safeguards.
The World Health Organization provides guidance on ethics and governance of artificial intelligence for health. The UNESCO Recommendation on the Ethics of Artificial Intelligence offers principles concerning rights, diversity, and oversight. They need applicable local rules, technical evaluation, and the particular teenager's experience.
Leaving the script to provide better care#
Listening does not require accepting every proposal or abandoning adult responsibilities. It requires explaining limits, providing alternatives, and allowing received information to change decisions. Dignity recognises knowledge; safety protects against risk and retaliation; repair acts when a process fails. Disagreement can improve a project.
A useful robot knows its place when it performs a bounded task, acknowledges uncertainty, accepts being stopped, and refers to a person. It does not interpret silence as permission or interaction as emotional connection. When an institution gives up its script, the teenager ceases to be a test subject and participates in defining care. Technology becomes more modest, safer, and genuinely useful.
