Relationship-mediated emotional support AI system

Abstract

Sharing the sender's feelings or opinions directly with the recipient can lead to misunderstandings and psychological burdens, depending on the relationship with the recipient, potentially damaging the relationship. Furthermore, it is necessary to avoid sharing information without the sender's consent. [Solution] Based on relationship information that identifies the relationship between the sender and the recipient, the original expression of the feelings or opinions entered by the sender is not sent directly to the recipient. Instead, it is converted into an expression that is more acceptable to the recipient and presented to the sender. The sender is then asked whether they would like to share the converted expression, and only if the sender consents is obtained is the converted expression shared with the recipient. This makes it possible to share information in an expression appropriate to the relationship and based on the sender's consent.

Description

Technical Field

[0001] The present invention relates to an emotion support system utilizing artificial intelligence technology, and particularly relates to a relationship-mediating AI intended for maintaining psychological safety and coordinating and restoring relationships among a plurality of parties. Specifically, the present invention relates to a technology that effectively functions in sites where human relationships are fixed and continued, such as households, educational institutions, welfare facilities, and corporations. In addition, the present invention also includes a function that promotes the organization and reflection of one's own emotions through dialogue, and supports the process in which the parties recognize and organize their own feelings and true wishes. This also contributes from the perspective that relationship restoration with the other party starts from self-understanding.

Background Art

[0002] Conventionally, when emotional discrepancies or misunderstandings arise between people, face-to-face dialogue or intervention by experts (counselors, coordinators, etc.) has been the main countermeasure. However, it is difficult to continuously use these means in daily life, and there are not a few cases where relationships are broken while emotional frictions accumulate. In addition, conventional AI chatbots and dialogue agents are specialized in one-on-one interactions with individual users, and do not have functions that mediate and translate the emotional states of multiple people and contribute to maintaining relationships. In recent years, "attentive listening AI" that allows children to consult about their worries through dialogue with AI has begun to be introduced in school sites and local governments.

[0003] For example, the attentive listening AI developed by ZIAI Inc. provided a mechanism for elementary and junior high school students to chat anonymously with AI in a demonstration experiment in Kashiwa City, Chiba Prefecture, and its usage was reported to greatly exceed the conventional number of consultations (Non-Patent Documents 1 and 2). As a result, the psychological effect that makes children more likely to open their hearts through dialogue with AI is being socially recognized.

[0004] However, these listening AI systems primarily aim to "allow individuals to talk freely," and do not explicitly include features such as adjusting emotional expression, controlling content sharing, or mediating relationships. In other words, conventional technologies do not have a configuration that comprehensively manages processes such as transforming the expression when the content of the consultation is conveyed to others, proactively confirming consent for transmission, and procedures for notifying third parties.

[0005] Based on the psychological effects of such attentive AI, this invention provides a system with a mediating support structure that enables expression adjustment and information sharing based on the user's will, thereby mitigating emotional friction in family and welfare contexts. [Prior art documents] [Non-patent literature]

[0006] Reference 1: ZIAI Co., Ltd. Press Release "A pilot project using AI-powered listening skills for chat-based counseling for children is being conducted in Kashiwa City, Chiba Prefecture." PR TIMES, released on November 1, 2023

[0007] Reference 2: Kashiwa City AI Chat Consultation Service (ZIAI Listening AI Introduction) [Overview of the project]

[0008] This invention relates to an emotion support AI system that enables multiple users to individually interact with an AI, record their emotions and opinions, and then communicate that content to others, thereby reducing psychological burden and maintaining relationships. Users are managed with individual IDs, and the history of dialogue content and emotion sharing is stored separately. Furthermore, when sharing, information is transmitted in a step-by-step and flexible manner through processes such as translation, paraphrasing, timing adjustment, and consent confirmation. Furthermore, by incorporating auxiliary functions such as a suggestive confirmation function, voice output, dialogue tendency analysis, and third-party notification, the present invention helps to avoid tension in relationships and promotes the restoration of trust and support for reintegration in households, educational settings, welfare support, etc.

[0009] This invention relates to a relationship-mediated emotional support AI system, which provides a configuration that minimizes psychological burden when multiple users individually interact with the AI ​​to record and organize their emotions, and when sharing that information with others, through processes such as expression transformation, suggestive confirmation, and consent confirmation. In particular, it is a technology that alleviates relationship tension in family and welfare contexts, enabling support for reintegration and preventive intervention.

[0010] Users are managed by individual IDs, and while interacting with the same AI personality, the response style is individually optimized. Furthermore, all sharing is based on consent and can be done asymmetrically. In addition, features such as log management, voice output, and third-party notifications ensure both psychological safety and flexible expression support. [Problems that the invention aims to solve]

[0011] In settings where relationships tend to be fixed, such as homes, workplaces, and educational and welfare facilities, when emotional tensions escalate, direct dialogue between the parties involved can be psychologically burdensome and may even exacerbate friction.

[0012] This invention is designed so that instead of directly exchanging emotions, each party individually interacts with the AI, organizes and visualizes their emotional state, and then, with the sender's consent, translates it into appropriate language and content for the recipient before sharing it in stages. The aim is to maintain relationships while avoiding direct emotional conflict and to support the continuation of those relationships. [Means for solving the problem]

[0013] The present invention relates to a relationship-mediated emotion support AI system (hereinafter referred to as "the System (1)") that provides a mechanism for multiple users to individually record and interact with emotions through an AI character having the same personality.

[0014] Each user has a separate user ID management unit (30), and record information is maintained independently for each. This system (1) designs responses individually for each user, while maintaining a unified personality.

[0015] Furthermore, the system (1) is configured to include a log input / output unit (90) and a log management unit (100) that visually output the history of conversations and emotional sharing between the user and the system (1).

[0016] This allows users to review past interactions not only as text but also to reconfirm them via audio as needed, increasing the flexibility of support and encouraging introspection and growth in users regardless of age or characteristics. Logs are separated for each user, and shared history is explicitly managed, ensuring privacy and psychological safety.

[0017] Furthermore, for child users or younger individuals, the system includes a "clarification function" via a suggestive confirmation prompt unit (70) that uses stamps, facial expressions, or simple questions to reconfirm the user's intentions before sharing content.

[0018] This feature provides users with an opportunity to reconsider whether they should really be sharing this information with the recipient, preventing accidental sending of messages or emotional outbursts, and also helps them become more aware of their intended message.

[0019] Furthermore, if the content is difficult to understand or there is a possibility of misunderstanding, this system (1) will verify the content and make revisions to ensure that the information is shared in a way that the individual ultimately understands and agrees with.

[0020] Furthermore, this system (1) is configured to maintain a common personality for all users while also including an individualized dialogue AI unit (40) that learns the speech tendencies and emotional expression patterns for each user's ID and individually optimizes the response style.

[0021] This allows users to perceive a consistent character while effectively experiencing personalized conversational styles, fostering a sense of psychological security and promoting the building of lasting trust.

[0022] Via the sharing control unit (50) in the cloud relay hub, when the system (1) transmits the recorded emotion content received from a user to the other party, the expression conversion unit (60) converts the expression using parables, indirect speech, and neutral paraphrasing to avoid causing emotional burden or misunderstanding. This conversion also takes into account factors such as the intensity of the emotion, the relationship with the other party, and the timing of sharing.

[0023] Further, the information is not transmitted directly, and the system (1) is configured to make a suggestive confirmation to the sender, such as "Would it be acceptable to convey it this way?". In addition to text responses, the expression conversion unit (60) is also configured to be capable of presenting converted expressions via speech synthesis as needed, to assist the user's understanding. All sharing is translated into a form that is easy for the other party to accept based on the sender's consent, and is executed by adjusting the sharing timing as necessary.

[0024] Note that the present system (1) is capable of sharing information gradually and asymmetrically according to the relationship and psychological state between users during sharing.

[0025] "Asymmetric sharing" refers to a mode in which the content, amount, or provision timing of the shared information differs bidirectionally, for example, providing detailed emotion information to one user, but notifying only the fact of "confirmed" to the other user.

[0026] Note that the present system (1) may be configured to allow selection of a "non-intervention mode" in which no translation or expression conversion is performed during sharing, and the original text of the sender is shared as is. This mode is particularly effective when respecting the user's intention to convey a message directly in their own words, or in highly urgent situations where immediate sharing without translation is required. By being able to handle cases where translation intervention is unnecessary, the design supports more flexible and autonomous expression of intent.

[0027] Further, the sharing control unit (50) may have a function of analyzing conversation tendencies based on the relationship group ID, and determining the timing of sharing and the necessity of expression conversion.

[0028] This analysis of dialogue trends is achieved by using indicators such as changes in the frequency of negative vocabulary used by users (e.g., "lonely," "anxious," "want to disappear"), sudden increases or decreases in conversation frequency, and rapid changes in emotions, and analyzing them using machine learning models and rule-based logic.

[0029] Furthermore, in this system (1), a "relationship group ID (family ID)" is set for each household, allowing multiple users belonging to the same household (e.g., parent and child) to interact with this system (1) using their individual IDs.

[0030] This system (1) references this relationship ID through a relay hub (shared control unit (50)) and delivers only the necessary information to the relevant parties within the group in an appropriate manner, thereby enabling secure information sharing while avoiding emotional conflicts.

[0031] Furthermore, this system (1) includes a mode that allows users to interact with the system (1) anonymously. Users can conceal their real names and set arbitrary nicknames or avatars to express their emotions, and the user ID management unit (30) is designed so that users can freely express their feelings without worrying about being identified. This configuration provides an environment in which people can express themselves with peace of mind, especially in situations where direct communication is difficult, such as in family or school relationships.

[0032] Furthermore, it may also include a third-party notification unit (80) that notifies a trusted third party (supporter ID (110)) of emotional information based on the user's declaration. [Effects of the Invention]

[0033] According to the present invention, it is possible to realize a function that intervenes in the "relationships" that was difficult for conventional conversational AI.

[0034] In particular, it can prevent relationship deterioration and breakdown in psychologically stressful situations or in relationships where direct dialogue is difficult. In addition, it can function as a "bridge" until intervention by supporters arrives, thereby increasing the sustainability of interpersonal support.

[0035] Furthermore, by including a function to notify trusted third parties (such as registered contacts or supporters) based on the individual's declaration as needed, it is possible to create an entry point for support in serious situations. However, careful design and operation are required when implementing this notification function from the perspective of privacy protection and psychological safety.

[0036] Furthermore, it has been reported that there is a certain degree of risk of re-abuse when a child returns home after being placed in protective custody outside the home. A structure that allows children to identify signs of this risk through daily conversations with AI and connect them to support at their own discretion can contribute to both a sense of security and support for independence.

[0037] Furthermore, within the family, small anxieties and discomforts from daily life, such as childcare, elder care, and marital discord, tend to accumulate, and in many cases, relationships deteriorate because these are not properly shared. This invention provides each family member with a means to organize and visualize their emotions through individual conversations with AI and communicate them to others as needed, thereby preventing emotional misunderstandings and contributing to the maintenance of a secure family environment.

[0038] In particular, when parents have difficulty expressing their emotions or when children harbor feelings that are difficult to put into words within the family, AI can function as a "neutral space" where everyone can safely confide their true feelings, thereby contributing to the restoration of trust within families and providing preventative support.

[0039] Furthermore, by preventing the burden of consultation from being concentrated on one person within the family, and by having AI act as an initial emotional support system—a "first point of contact"—it can contribute to distributing the emotional burden in childcare and between spouses.

[0040] Furthermore, the "clarification function" incorporated into this system not only confirms the intentions of child users but also provides adult users with a period of emotional cooling. This allows them to review their words and intentions before sending, which is expected to prevent the deterioration of relationships.

[0041] Furthermore, this system offers the flexibility to determine how users communicate by allowing them to select a non-intervention mode during sharing, which does not perform expression conversion depending on the situation and only confirms the intention to share. [Brief explanation of the drawing]

[0042] Figure 1 is a block diagram showing the configuration of the relationship-mediated emotional support AI system (1) according to the present invention. This system (1) comprises User A (10), User B (20), User ID management unit (30), Individual dialogue AI unit (40), Shared control unit (50), Expression conversion unit (60), Proposal confirmation prompt unit (70), Third-party notification unit (80), Log input / output unit (90), Log management unit (100), and Supporter ID (110). Design Concept

[0043] This system is designed with the user's free will and psychological safety as its top priorities, and conversation content is never automatically shared externally. Even when notifications are made to supporters as needed, they are done with the user's consent or based on carefully designed criteria, and through polite explanatory prompts directed at the user. Notifications are limited to supporters whom the user trusts, and the system is configured so that the user can choose whether or not to receive notifications, ensuring that psychological safety is not compromised even when interventions are taken. [Modes for carrying out the invention]

[0044] The relationship-mediated emotional support AI system of the present invention (hereinafter referred to as "the System (1)") has a configuration that allows multiple users to interact with the System using their respective IDs and to mediate the sharing of the content of that interaction with the other party in an appropriate manner and at an appropriate time.

[0045] This system (1) is equipped with an individual user ID management unit (30) for each user, and independently maintains the dialogue history and emotional records with each user. The content spoken by the user is analyzed by the individual dialogue AI unit (40) to determine the dialogue tendencies and emotional tendencies, and a response is generated in a style that is natural for the user.

[0046] The emotional content recorded by the user is shared with other users as needed via a sharing control unit (50) connected to a cloud relay hub. During this sharing process, an expression conversion unit (60) considers the intensity, relationship, and timing of the emotion, and converts it into an expression that is easily accepted by the recipient through analogies, indirect speech, neutral paraphrasing, etc.

[0047] Sharing is carried out entirely with the sender's consent, and a suggestive confirmation prompt (70) asks for confirmation such as, "Is it okay to share this as is?". For children and users requiring special consideration, simpler confirmation methods such as stamps or facial expression selections are also used.

[0048] Furthermore, the third-party notification unit (80) is configured to notify registered trusted supporter IDs (110) of information such as emotional summaries, only if the user consents. Notifications are not made based on the AI's unilateral judgment, but are contingent on the user's consent being confirmed.

[0049] Furthermore, this system (1) includes a log input / output unit (90) and a log management unit (100), allowing users to review past conversations and sharing history as text and audio information. This enables a flexible configuration that supports introspection and self-awareness regardless of age or characteristics.

[0050] This system (1) plays a mediating role in supporting mutual understanding while avoiding direct emotional conflict in fixed relationship settings such as family, education, and welfare. Through the coordinated interaction of its individual elements, it functions as an integrated system that enhances psychological safety and leads to the restoration of relationships and continued support. [Examples]

[0051] AI translation and sharing function in parent-child relationships If a child expresses feelings such as "I'm tired from school" or "Mommy was scary" through a conversation with the AI, the AI, with the child's permission, translates these into phrases like "They may be feeling tired" or "They may be feeling anxious about their relationship with their mother," and presents them to the parent as a suggestion. The parent can then use this information to talk to their child or consider how to respond as needed. [Examples]

[0052] Emotional sharing and collaborative support in sibling relationships When small frictions such as "I was ignored" or "He said he won again" repeatedly occur between siblings, the AI ​​can record their respective emotions and, if necessary and with the consent of the speaker, translate them into neutral expressions such as "There may be a misunderstanding between the siblings" and notify the parents. This allows parents to intervene early and try to mediate the relationship between siblings. [Examples]

[0053] Application of the function of asking for clarification in a marriage and asymmetrical sharing If emotionally charged statements are recorded between spouses, the AI ​​will prompt a confirmation message such as, "Is it okay to convey this exactly as it is?", giving the speaker an opportunity to reconsider their words. Furthermore, the shared content is translated into expressions that are more acceptable to the other party and presented in stages. [Examples]

[0054] Support for building a secure parent-child relationship after family reunification When a child returns home after temporary protection or placement in a facility, and experiences anxiety or tension regarding interactions within the family, the AI ​​records and organizes the child's emotions and provides reassuring information to the parents. For parents, the AI's neutral intervention helps avoid excessive interference or indifference, supporting a stable reintegration process. [Examples]

[0055] Intervention support through collaboration with external support providers In one household, a parent, who was a user of the AI, confided in it about the stress and loneliness they felt from parenting through conversations with the AI. The AI ​​analyzed the continuous conversation history and emotional changes and determined that the user's condition was at a level where they needed support.

[0056] The user was prompted to confirm whether it was acceptable to notify the city's childcare support center (supporter ID (110)) that had been registered in advance. Upon the user's consent, a summary of the user's emotions was sent to supporter ID (110), and the support center contacted the user.

[0057] After contact, the AI ​​provided follow-up prompts to the user, such as "Were you able to talk?", to help them feel more at ease after the intervention.

[0058] It should be noted that this system is not configured for AI to constantly monitor user logs and automatically send notifications. Instead, notifications are only issued after confirming the user's consent, based on information obtained through dialogue.

[0059] Protective intervention support after family reunification A middle school student who had left a children's welfare facility and been reunited with their biological parents was recording their daily life and relationship with their parents through regular conversations with an AI. Initially, the situation was stable, but after a certain period, emotional repression and anxiety became noticeable in the content of the conversations. The AI ​​analyzed the changes in the user's psychological state and speaking patterns. After continuously analyzing the situation, it was determined that there was a high possibility that the deterioration of the relationship would become more serious.

[0060] The AI ​​suggested to the user that they notify a child welfare center or support specialist (supporter ID (110)) that had been registered in advance. The user agreed, and the AI ​​compiled the necessary information and sent the notification. The support specialist who received the notification initiated contact to intervene, confirmed the situation, and took initial action leading to temporary protection or external support.

[0061] This example demonstrates a configuration in which AI can detect deterioration in relationships within a reintegrated family without relying on the child's spontaneous reporting, and enable supportive intervention through careful notification. In this embodiment as well, the system is designed as a notification system based on user consent, rather than a configuration in which the AI ​​automatically reports and makes decisions, thus giving maximum consideration to psychological safety. [Industrial applicability]

[0062] This invention is a system that alleviates emotional friction within the family, ensures psychological safety between parents and children, siblings, and spouses, and supports reintegration and welfare intervention by third parties, and has broad industrial applicability in fields such as child welfare, education, and family support.

[0063] Furthermore, the configuration of the present invention can be applied to fields such as mediating dialogue between users in social networking services (SNS) and online games, supporting emotional exchanges between lovers and married couples, facilitating emotional responses in customer support, and supporting communication and ensuring psychological safety among staff in companies and welfare facilities. The "individual dialogue and sharing control function for each user ID" of the present invention can be applied not only to the fields of family and welfare, but also to mediating emotional conflicts and misunderstandings between users in SNS and online games. The configuration, which converts each user's statements into expressions that reduce emotional burden and then notifies them in a gradual and asymmetrical manner, is also effective in maintaining and moderating online relationships, and is expected to have a wide range of applications. [Explanation of Symbols]

[0064] 1: Relationship-mediated emotional support AI system (AI) 10: User A 20: User B 30: User ID Management Department 40: Individual Dialogue AI Department 50: Shared control unit (relay hub) 60: Expression Conversion Unit 70: Proposal confirmation prompt section 80:Third Party Notification Department 90: Log Input / Output Section 100: Log Management Department 110: Supporter ID (third party)

Claims

1. A relationship-mediated emotional support AI system that assists in sharing emotions or opinions entered by the sender with recipients other than the sender, A representation conversion unit (60) does not transmit the original expression of the emotion or opinion recorded and transmitted by the sender to the recipient, but rather, based on relationship information that can identify the relationship between the sender and the recipient, converts it into an expression that is easily acceptable to the recipient and presents it to the sender. A proposal confirmation prompt unit (70) presents the expression converted by the expression conversion unit (60) to the sender and confirms with the sender whether it is acceptable to share the converted expression with the receiver, A sharing control unit (50) shares the converted expression with the recipient only if the sender's consent is obtained through the proposed confirmation prompt unit (70), Equipped with, The sharing control unit (50) is configured to ensure that the sharing is carried out asymmetrically by controlling, in stages, the content, quantity, or timing of the information shared with the recipient. A relationship-mediated emotional support AI system characterized in that, if the sender's consent cannot be obtained, neither the original expression nor the converted expression will be shared with the recipient.

2. The relationship-mediated emotional support AI system according to claim 1, characterized in that the proposed confirmation prompt unit (70) is configured to present the expressions converted by the expression conversion unit (60) in stages, and to allow confirmation of the sender's intention at each stage.

3. The relationship information includes relationships identified based on a user ID management unit (30) for identifying multiple users and a relationship group ID that defines relationships between users, and the representation conversion unit (60) performs the conversion based on the relationships, characterized in that the relationship-mediated emotional support AI system is as described in claim 1.

4. The relationship-mediated emotional support AI system according to claim 1, characterized in that it has an individual dialogue AI unit (40) that has a personality common to multiple users while optimizing the response style for each user.

5. The relationship-mediated emotional support AI system according to claim 1, characterized by comprising a log management unit (100) that visually presents the history of conversations and sharing with the user.

6. The relationship-mediated emotional support AI system according to claim 1, further comprising a third-party notification unit (80) that notifies a pre-registered supporter of the converted expression or summary information based on the converted expression, based on the consent of the sender.

Images