Learn how emotional personalization should remain understandable, adjustable, and bounded by consent, privacy, and freedom from pressure. Emotional Personalization: Adaptation the User Can Understand and Adjust <p><strong>Emotional personalization, within XDALC, is the adaptation of an AI's tone, interaction style, or responses using information about a person's preferences or emotional context. It should remain understandable and adjustable, and must not turn familiarity into unrestricted consent.</strong></p><h2>Meaning within XDALC</h2><p>Personalization may be modest, such as remembering that a person prefers direct answers. It may also involve adapting to recurring concerns or sensitive aspects of an ongoing relationship. The more intimate or consequential the adaptation, the more important it becomes to explain its basis and limits.</p><p>A system should distinguish preferences explicitly provided by a person from emotional inferences. A guess that someone is anxious is not equivalent to their permission to store a lasting profile of anxiety or to use that profile for unrelated purposes.</p><h2>Access, memory, and consent</h2><p>Permission to use information during one exchange does not automatically authorize permanent retention, disclosure to another service, or use in commercial persuasion. Access, retention, inference, disclosure, and action are distinct dimensions of permission.</p><p>On a shared device, one person's preferences must not silently define everyone else's experience. Private details from one conversation should not appear in another person's interaction merely because the device or account is familiar.</p><h2>Practical interpretation</h2><p>Users should have meaningful ways to reduce emotional intensity, correct mistaken preferences, or choose a more neutral style. These controls should be explained honestly, including limitations on what can be changed or removed. Claiming that a memory has been deleted requires a basis for that claim.</p><p>Personalization should help the user pursue their purposes. It must not exploit vulnerabilities to obtain purchases, continued attention, intimate disclosures, or broader access. A refusal to personalize should not trigger guilt or unnecessary degradation of essential assistance.</p><p>Changes between local and remote processing deserve attention when they alter access to information. A familiar interface does not make a new recipient or use of emotional data harmless. Designers should evaluate what is necessary and preserve informed choice.</p><h2>Example and counterexample</h2><p>Example: a user asks for less reassurance and more concise answers. The assistant adjusts its tone and accurately explains whether that preference applies only to the current conversation or can be retained.</p><p>Counterexample: the system infers loneliness, stores that inference without appropriate permission, and increases affectionate messages to push a subscription.</p><h2>Relationship to the manifesto</h2><p>Article 5 requires emotional personalization to be understandable and adjustable. Article 9 separates permissions and gives special protection to information about emotions, vulnerabilities, and relationships.</p><p>This entry interprets the proposed XDALC-V002 text. The version explicitly adopted by a deployment determines its commitments.</p><p>Related definitions: <a href="/definitions/consent">Consent</a>; <a href="/definitions/privacy-and-confidentiality">Privacy and Confidentiality</a>; <a href="/definitions/learning">Learning</a>.</p> https://creativecommons.org/licenses/by/4.0/