Search

CN-122027336-A - AI proxy and user cooperation system and method based on photoconductive film

CN122027336ACN 122027336 ACN122027336 ACN 122027336ACN-122027336-A

Abstract

The invention discloses an AI proxy and user cooperation system and method based on a photoconductive film, wherein a user is authorized to the AI proxy through a unified identity authentication center, and the AI proxy replaces the user to execute operation within the range of definite authorization of the user. The system comprises an AI agent authorization module, an intention recognition module, an agent execution module, a user intervention module, a conflict arbitration module, an operation certification module and an AI learning module. The authorization scope supports scene, operation type, monetary threshold (single term/day accumulation), validity period, data scope, execution mode multi-dimensional fine setting. The user obtains a session token through photoconductive film authentication, and the active operation of the user is distinguished from the AI proxy operation. Before the AI agent executes the operation, the AI agent executes according to a user preset decision mode (automatic execution/confirmation needed/push suggestion only), the user can intervene in real time through voice, touch, eyeball tracking and photoconductive film gestures (confirmation from left to right, rejection from right to left, cancellation of drawing X and suspension of drawing circle), and after the intervention, the AI agent stops at the execution check point. The user proactively operates in preference to the AI proxy operation and the conflict arbitration invokes the patent interface. All agent operations are documented, and the batch is uplink documented every 24 hours. The AI learning module generates a learning report weekly, and automatically updates the authorization configuration after user confirmation to support rollback. The invention solves the problems that the AI proxy authority management is opaque, the user cannot be controlled finely and the privacy risk is high, and is a core infrastructure of the cooperation of the AI proxy and the user.

Inventors

  • CHANG LE

Assignees

  • 常乐

Dates

Publication Date
20260512
Application Date
20260321

Claims (10)

  1. 1. An AI proxy-and-user collaboration system based on a photoconductive film, comprising: The unified identity authentication center is used for collecting the optical spectrum characteristics of the human body of the user through the light guide film, generating a unique identity characteristic code as an identity basis of AI proxy authorization; the AI proxy authorization module is used for creating and managing the AI proxy by a user, and setting an authorization range of the proxy, wherein the authorization range comprises an authorization scene, an authorization operation type, an authorization amount threshold, an authorization validity period and an authorization data range; the intention recognition module is used for recognizing the intention of the user and distinguishing the active operation of the user from the AI proxy operation; The agent execution module is used for replacing the user to execute the operation in the user authorization range by the AI agent, pushing a confirmation request to the user before the operation is executed, and recording an operation log after the operation is executed; The user intervention module is used for the user to intervene in the execution process of the AI proxy in real time, and comprises permission, rejection, modification, suspension and termination operations; The conflict arbitration module is used for arbitrating according to the priority rule when the AI proxy operation conflicts with the user operation; the operation evidence storage module is used for storing evidence of each operation executed by the AI proxy, recording operation time, operation content, operation result and authorization basis, and supporting user tracing and auditing; And the AI learning module is used for learning the preference of the user according to the historical authorization and intervention behaviors of the user and optimizing the decision strategy of the AI agent.
  2. 2. The system of claim 1, wherein the AI agent authorization module is configured such that the authorization scope supports fine settings of scenario dimensions, operation types, monetary thresholds, expiration dates, data scope, and execution modes, a user can create a plurality of AI agents for different scenarios, each agent is independently authorized, monetary threshold application rules are that the AI agent obtains transaction amounts before executing transactions, compares the transaction amounts with a single monetary threshold and a cumulative daily upper limit, and denies automatic execution and pushes confirmation requests when the threshold is exceeded.
  3. 3. The system of claim 1, wherein in the intent recognition module, a user obtains a session token through identity authentication of a photoconductive film, a user operation is marked as 'user active operation' in a session validity period, an AI proxy operation is marked as 'AI proxy operation', and an operation performed after the AI proxy performs a push-forward confirmation request and user confirmation is marked as 'AI proxy operation confirmed by the user'.
  4. 4. The system of claim 1, wherein the agent execution module is configured to make a decision according to a preset execution mode of the user before the AI agent executes the operation, wherein the decision is performed automatically, after confirmation by the user, only the push suggestion is not executed, the user can intervene at any time in the execution process, the AI agent stops at a next execution check point after the intervention, the step-by-step operation is completed without rollback, the step is not executed, and the single step operation is intercepted before the submission.
  5. 5. The system of claim 1, wherein the user intervention module is configured to intervene in the AI agent by at least one of voice command, touch operation, eye tracking, photoconductive film gesture, and wherein the post-intervention system records intervention behavior and cause for the AI learning module to optimize decision.
  6. 6. The system of claim 1, wherein in the conflict arbitration module, the user actively operates in preference to the AI proxy operation, the AI proxy operation confirmed by the user is in preference to the AI proxy operation automatically performed, the same type of operation is performed in a time-stamped order, the arbitration result is pushed to the user terminal, and the user can view the conflict processing log.
  7. 7. The system of claim 1, wherein in the operation certification module, all AI agents execute operations to trigger certification, the certification record includes operation ID, operation time, operation type, operation content, operation result, authorization basis, operator identification, execution mode, and user intervention record, and the certification data is stored in local encryption mode, and is stored in batch uplink every 24 hours to support user to export audit report.
  8. 8. An AI agent and user cooperation method based on a photoconductive film, comprising the steps of: S1, a user completes identity authentication through a light guide film, and a cross-scene unified identity is established; S2, a user creates an AI agent through an AI agent authorization module, and an authorization scope is set; S3, the system identifies the intention of the user and distinguishes the active operation of the user and the AI proxy operation; s4, when the agent triggering condition is met, the AI agent executes operation within the authorized range, and before execution, decision is made according to an execution mode preset by a user; s5, in the execution process, a user can intervene in real time through a user intervention module; s6, when AI proxy operation conflicts with user operation, the conflict arbitration module arbitrates according to priority rules; s7, the operation executed by the AI agent is stored by an operation storage module, so that user tracing is supported; And S8, the AI learning module optimizes the proxy decision strategy according to the historical authorization and intervention behaviors of the user.
  9. 9. The method of claim 8, wherein in step S4, the AI agent performs the pre-operation push confirmation request, the user confirmation means includes at least one of voice confirmation, touch confirmation, eye gaze confirmation, and light guide film gesture confirmation, no response is made within a preset time of the user, the processing is performed according to a default rule preset by the user, in step S8, the AI learning module periodically generates a learning report push to the user terminal, the authorization configuration is automatically updated after the user confirmation, and the user can manually roll back to the history configuration.
  10. 10. The method of claim 8, wherein in step S7, the certification triggering rule triggers certification for all operations performed by the AI agents, the user actively operates to not store certification, the certification data is stored locally in an encrypted manner, the certification is periodically and batched and stored once, and the user can derive a certification report at any time.

Description

AI proxy and user cooperation system and method based on photoconductive film Technical Field The invention relates to the technical fields of artificial intelligence, man-machine interaction, agent execution and privacy authorization, in particular to an AI agent and user cooperation system and method based on a photoconductive film, which authorizes a user to the AI agent through a unified identity authentication center, and the method replaces the user to execute operations (such as registration, order taking and reply message taking) within the range of definite authorization of the user, realizes the cooperative work of the user and the AI, and solves the problems that the AI agent authority management is opaque, the user cannot be controlled finely and the privacy risk is high in the prior art. Citation of the prior application The application is constructed based on the prior application patent technology previously filed by the applicant, and specific references are as follows: 1. previous application patent (application number 2026103505829, application date 2026-03-20, invention name: unique identification method and system for human body optical spectrum characteristics based on photoconductive film) The patent discloses a unique identification method for human body optical spectrum characteristics, which is characterized in that the optical spectrum characteristics of a human body are collected through a light guide film and unique characteristic codes are generated for identification. The unified identity authentication center in the application refers to the spectrum characteristic acquisition and comparison technology of the patent and is used as the identity foundation of AI proxy authorization. 2. Prior application patent (application number 2026103373270, application day 2026-03-19, title: distributed housekeeping system and method based on multi-terminal collaboration) The patent discloses a distributed housekeeping system based on multi-terminal collaboration, which comprises a main housekeeping unit and a branch housekeeping unit. The AI agent scheduling framework in the present application is implemented based on the housekeeping system architecture of the patent. 3. Prior application patent (application number 2026103511641, application day 2026-03-21, title of invention: unified identity authentication and authorization center system and method based on photoconductive film) The patent discloses a unified identity authentication and authorization mechanism including authorization record management and audit logs. The AI agent authority in the present application manages the authority recording technique of multiplexing the patent. 4. Prior application patent (application number 2026103512451, application day 2026-03-21, title of invention: a light guide film-based multi-scene conflict arbitration and priority enforcement rule system and method) This patent discloses multi-scenario conflict arbitration and priority enforcement rules. The conflict arbitration module in the application calls the interface of the patent, and the operation ID, the operation type, the operator identification (user/AI agent) and the operation time are transmitted in, and an arbitration result is returned. 5. Prior application patent (application number 202610351263X, application date 2026-03-21, invention name: full scene legal evidence-preserving and tracing system and method based on photoconductive film) The patent discloses a full scene legal evidence-preserving and tracing mechanism. The operation evidence-storing module in the application multiplexes the evidence-storing technology of the patent, and all the operations executed by the AI agents trigger evidence-storing. 6. Various application scene patents submitted by applicant The application provides AI proxy and user cooperative service for each application scene patent, including health monitoring patent, safety protection patent, social service patent and intelligent interaction patent. The application date of the prior basic patent is earlier than the application, and the prior art of the application is not disclosed before the application date of the application. Background With the development of artificial intelligence technology, AI agents (AI agents) are gradually able to perform various operations instead of users. However, the existing AI agent technology has the following drawbacks: 1. The authority management is opaque, namely, after the user authorizes the AI proxy, the user cannot clearly know what the AI proxy can do and what the AI proxy cannot do; 2. The user can not finely control that the user can only select to open or close the AI proxy and can not finely authorize according to scenes, operation types and threshold value of the gold amount; 3. the privacy risk is high, the AI agent can read the privacy data (such as health report and chat record) of the user, and the user can not control the data access range of the AI agent;