Artificial awareness enhancement engine with intention understanding, emotional resonance and moral judgment capabilities

Abstract

The invention relates to artificial intelligence and cognitive calculation, and provides an artificial consciousness enhancement engine with intention understanding, emotional resonance and moral judgment capabilities. The engine comprises multi-modal input, an intention understanding module (constructing a cognitive map based on DIKWP), an emotional resonance module (cross-modal emotion recognition and semantic mapping including emotional memory), a moral judgment module (local ethical knowledge map + rule engine + dynamic weight) and a human in-loop feedback interface. According to the scheme, joint modeling and correction are carried out on intention, emotion and ethical risks, improper output interception and ethical tradeoff are supported, and adaptive updating is carried out along with feedback; the method can remarkably improve the understanding accuracy and the information sharing and compliance reliability, and is suitable for the scenes of accompanying, medical treatment, education, customer service and the like.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence and cognitive computing, and more particularly to an artificial consciousness enhancement engine. This engine integrates intent understanding, emotional empathy, and moral judgment capabilities to enhance the humanistic attributes and social trust in artificial intelligence systems, and can be widely applied in areas such as intelligent companionship, medical assistance, and educational assistants. Background Technology

[0002] With the rapid development of artificial intelligence, systems such as chatbots and virtual assistants have already been able to understand natural language and communicate with humans to a certain extent. However, existing AI systems still have significant shortcomings in deep intent understanding, emotional empathy, and ethical judgment.

[0003] Lack of intent understanding: Traditional AI agents primarily rely on semantic matching or shallow contextual responses, making it difficult to accurately grasp the user's underlying purpose and motivation. For example, when a user asks a seemingly simple question, the true underlying need is often overlooked. The classic DIKW (Digital-Knowledge-Wisdom) model views data, information, knowledge, and wisdom as a hierarchical structure of progressively abstract concepts, while the latest DIKWP model adds "Purpose" at the top level, emphasizing that subjective intent guides the objective cognitive process. However, most existing AI does not fully utilize this model to model user intent, resulting in responses that lack specificity and personalization.

[0004] Insufficient Emotional Resonance: In human-computer interaction, a lack of empathetic response and awareness of user emotions reduces user trust. While some AI possesses emotion analysis capabilities (such as identifying emotions through text sentiment analysis or voice tone), cross-modal emotion recognition and expression remain weak. For example, combining a user's voice tone, facial expressions, and language content for comprehensive emotion recognition can significantly improve the accuracy of judging emotional states. AI lacking emotional resonance mechanisms often provides cold, impersonal responses, failing to soothe users in a state of sadness or anxiety.

[0005] Inadequate ethical judgment: AI systems are prone to inappropriate responses when faced with sensitive topics or ethical conflicts. Current mainstream practices largely rely on pre-set rules or filtering of inappropriate content, lacking in-depth moral understanding. Different regions and cultures have their own values; if AI cannot adjust its behavior according to local culture, it may provoke user resentment or even social backlash. Furthermore, users may request AI to provide unethical or harmful advice, requiring AI to possess proactive moral judgment and self-restraint. However, achieving this requires explicitly integrating ethical knowledge into AI's cognition and dynamically weighing various value principles according to the context, a problem that has not yet been fully addressed in current technology.

[0006] Lack of continuous optimization through human feedback: Many current AI systems rely heavily on offline training and developer settings for improvement, lacking a continuous feedback mechanism from end-users once deployed. In recent years, methods such as Human Feedback Reinforcement Learning (RLHF) have demonstrated the crucial role of incorporating human-in-the-loop feedback in aligning AI values. Particularly in the ethical domain, user corrections of inappropriate AI responses can serve as valuable training signals, helping AI distinguish between ethical and unethical behavior. However, most existing applications lack an intuitive and effective interface for users to provide such feedback, resulting in slow evolution of AI's ethical cognition.

[0007] Given the aforementioned shortcomings, there is an urgent need for a new core module for artificial intelligence that can comprehensively enhance AI's "consciousness" capabilities from three aspects: intention, emotion, and ethics, by combining cutting-edge theories of cognitive science and ethical computing. At the theoretical level, Integrative Information Theory (IIT) points out that consciousness corresponds to a system's ability to integrate information, and the degree of consciousness of a system can be measured by the comprehensive information content Φ. This provides guidance for the architectural design of artificial consciousness systems: by highly integrating multimodal information and multi-level cognition, it is hoped that the level of AI's "artificial consciousness" can be improved. This invention is proposed against this backdrop, aiming to construct an artificial consciousness enhancement engine that integrates the DIKWP model, IIT, multimodal affective computing, and human feedback mechanisms, to significantly improve AI's depth of understanding of human intentions, its ability to resonate with emotions, and its judgment and adaptation capabilities regarding ethics. Summary of the Invention

[0008] The purpose of this invention is to provide an "artificial consciousness enhancement engine with the ability to understand intent, empathize with emotions, and make moral judgments." It comprehensively utilizes the DIKWP cognitive model, Integrative Information Theory (IIT), cross-modal emotion recognition technology, and human-in-the-loop feedback mechanisms to compensate for the shortcomings of existing artificial intelligence in deep understanding and ethical processing. Through this engine, AI systems can exhibit more human-like understanding, empathy, and value judgments in interactions, thereby improving user experience and social acceptance.

[0009] To achieve the above objectives, the engine of this invention mainly includes the following core modules and mechanisms:

[0010] Intent Understanding Module: Constructs a user intent graph based on the DIKWP model. The engine extracts data and information from the user's natural language input and related multimodal cues (voice tone, facial expressions, behavioral actions, etc.), and through knowledge and intelligence-level reasoning, ultimately forms an inference of the user's core intent at the purpose level. This module can identify the deep-seated purpose driving user behavior and questions, and provide personalized, context-relevant responses accordingly, rather than simply answering the surface-level questions.

[0011] Emotional Resonance Mechanism: A built-in cross-modal emotion recognition model and semantic sentiment mapping mechanism enable AI to perceive and empathize with user emotions. Specifically, the engine integrates multimodal information such as text semantic analysis, voice features, and visual expression recognition to determine the user's current emotional state (e.g., happiness, anger, sadness, anxiety). Then, through the sentiment mapping mechanism, appropriate emotional nuances are injected into the generated response. For example, when a user is detected to be depressed, the system generates a reassuring and encouraging response; when the user is angry, it provides a calm and understanding response. This "empathic response" mechanism makes the AI's responses more human and increases user trust.

[0012] The Moral Judgment Module integrates a local cultural ethical knowledge graph and a dynamic value weighting mechanism, endowing AI with proactive moral judgment and adjustment capabilities. This module incorporates a knowledge graph covering local social ethical norms, laws and regulations, customs, and weight parameters for several basic value dimensions (such as safety, fairness, privacy, and honesty). Before generating suggestions or actions, the engine invokes the ethics module to review the proposed solutions: (a) based on the knowledge graph and rule engine, it reasones and evaluates potential ethical conflicts; (b) for sensitive requests (such as those involving violence, illegality, or privacy), it dynamically adjusts value weights to balance different ethical principles, correcting or refusing inappropriate outputs when necessary. For example, in the "trolley problem" scenario, the system can comprehensively consider the value of life and the consequences of actions, retrieve relevant ethical rules through the knowledge graph, dynamically weigh the value weights of each option, and provide the optimal decision that conforms to moral principles. Localized ethical knowledge enables AI to make judgments consistent with local values ​​in different cultural environments, avoiding a "one-size-fits-all" moral standard.

[0013] Human-in-the-loop training feedback interface: This provides a user-friendly feedback mechanism, enabling AI to continuously evolve its cognitive and ethical models based on real-world interactions. Specifically, it includes an "ethical intention feedback model": when users find the AI's understanding of their intentions flawed, or are dissatisfied with the moral / emotional inappropriateness of a response, they can directly annotate the problem or provide suggestions for improvement through the interface. The system uses this human feedback as reward signals for reinforcement learning, periodically updating the moral knowledge base and fine-tuning model parameters. For example, if the engine detects strong dissatisfaction or resistance from a user regarding a response (e.g., due to cultural taboos), it immediately triggers a self-correction mechanism: adjusting the current dialogue strategy to mitigate misunderstandings, and recording the case for inclusion in subsequent training data, allowing the engine to gradually develop more appropriate handling strategies in similar situations. Through continuous iteration of human feedback, the AI's accuracy in understanding intentions, its empathy, and its moral judgment will be continuously optimized, achieving the continuous evolution of artificial consciousness.

[0014] Through the collaborative work of the aforementioned modules, the engine of this invention achieves the following beneficial effects: a deeper and more accurate understanding of user intent, resulting in more personalized responses; empathetic and timely responses to user emotions, enhancing user experience; self-censorship and adjustment capabilities for complex ethical issues, reducing inappropriate behavior; and continuous learning and improvement under user supervision in practical applications. Compared to existing technologies, this invention significantly enhances the capabilities of AI systems at the cognitive understanding, emotional interaction, and value judgment layers, making them closer to human communication levels and value connotations, thus improving security, credibility, and humanistic care.

[0015] This invention's artificial consciousness enhancement engine, as a core module of high-level AI, has broad application prospects, especially in fields requiring high-level understanding, empathy, and ethical control, where it will play a significant role.

[0016] Intelligent Companionship and Psychological Counseling: The engine can be applied to social robots, emotional support chat assistants, etc., providing considerate communication and companionship for elderly people living alone and people under psychological stress. By deeply understanding the emotions and intentions behind the user's words, it provides appropriate comfort, encouragement, or advice, playing the role of a human-like friend in psychological counseling and emotional support. Compared with traditional script-based companion robots, engine-driven systems can truly "understand" the user's troubles and "empathize," significantly improving the warmth and therapeutic effect of the interaction.

[0017] Healthcare Assistance: In medical consultation assistants and rehabilitation companion AI, the engine can understand the patient's stated intentions and emotional state. For example, it can distinguish whether a patient is seeking professional advice or emotional support, empathize with and soothe the patient's anxiety, and provide advice in strict accordance with medical ethics (such as balancing risks and privacy in medication reminders and rejecting inappropriate requests). This will increase patients' trust in AI medical assistants, assist doctors in providing humanistic care, and improve the quality of medical services.

[0018] Educational Learning Assistant: Applying the engine to personalized education systems, intelligent tutors, or teaching assistant robots can more keenly perceive students' learning intentions and emotional changes. When students ask questions, the engine not only provides answers but also judges whether the student truly understands or is merely expressing confusion, thereby adjusting the explanation method and difficulty. Similarly, it encourages frustrated students and appropriately reminds arrogant students to be humble. At the same time, adhering to educational ethics, the engine guides thinking rather than directly providing homework answers, maintaining fairness and academic integrity. This AI teaching assistant, possessing both IQ and EQ, will greatly enhance the learning experience and effectiveness.

[0019] Customer Service and Consultation: Applying this engine to customer service robots in industries such as finance, insurance, and retail enables AI customer service to better understand customers' implicit needs and provide personalized services. For example, when a customer expresses dissatisfaction, the AI ​​can sense the emotion, apologize and soothe the customer in a timely manner, and provide solutions in accordance with professional ethics, avoiding exaggerated promises or invasion of privacy. The result is increased customer satisfaction and trust, and an improved corporate image.

[0020] Human-Machine Symbiotic Collaboration: As a general-purpose artificial consciousness module, the engine can be embedded in various human-machine interaction systems, such as intelligent vehicles, home assistants, and interactive entertainment. In these scenarios, the engine endows machines with keen insight into human intentions and ethical constraints on behavior, ensuring that machine actions both conform to user expectations and do not deviate from social norms. For example, an intelligent car assistant can proactively play soothing music when it senses that the driver is emotionally stressed, and strictly adhere to safety ethics (preferring to stop rather than take risks) when making autonomous driving decisions. This will promote the formation of a safe and trustworthy human-machine symbiotic relationship.

[0021] The artificial consciousness enhancement engine provided by this invention can significantly improve the understanding, empathy, and ethical judgment of AI systems, making up for the shortcomings of existing artificial intelligence in terms of humanistic literacy. Its application will enable AI to better serve humanity: at the micro level, users will have intelligent assistants that better understand and care about them; at the macro level, AI systems will be more aligned with social values, reducing trust crises and risks caused by AI decision-making errors or perceived "inhumanity." With the maturity and widespread adoption of the technology, this engine is expected to become a standard core component of next-generation artificial intelligence applications, empowering numerous human-centered industries such as companionship, healthcare, education, and customer service, creating enormous social value and commercial prospects. By integrating human intentions, emotions, and morality into machine intelligence, we have taken a crucial step towards artificial intelligence with "empathy" and "values," which has profound significance for building a future where humans and machines coexist harmoniously. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the three-layer module architecture of the artificial consciousness enhancement engine of the present invention.

[0023] Figure 2 This is a flowchart of the cognitive-emotional-ethical processing of the engine of this invention.

[0024] Figure 3 A schematic diagram of the human-in-the-loop feedback mechanism for the engine. Detailed Implementation

[0025] The artificial consciousness enhancement engine of this invention adopts a modular architecture design, organically combining the three key functional levels of cognition, emotion, and ethics, and forming a closed-loop learning through human feedback. The following detailed description, in conjunction with specific implementation methods, provides a comprehensive overview of the engine's overall architecture, workflow, and the functions of each module.

[0026] The engine adopts a three-layer architecture, from bottom to top: the perception and understanding layer, the emotional resonance layer, and the ethical decision-making layer. The functions of each layer are as follows:

[0027] The perception and understanding layer (cognitive layer) is responsible for acquiring multimodal information and modeling intent cognition. This layer receives various user inputs, including text dialogues, voice intonation, facial expressions, images, and behavioral sensor data. It preprocesses these inputs using natural language processing and pattern recognition algorithms to extract useful data and information features. Subsequently, through semantic parsing by the DIKWP model, the extracted information is progressively integrated: from the data layer to the information layer to obtain factual semantics, then to the knowledge layer to associate background knowledge, to the intelligence layer for reasoning and judgment, and finally to the purpose layer to form a high-level representation of the user's intent. This layer is equivalent to the front end of the AI's cerebral cortex, completing the understanding of "what the user said and what they wanted." In implementation, a method combining a Large Language Model (LLM) with knowledge graph reasoning can be used: the LLM handles dialogue understanding and generation, the knowledge graph provides background common sense and contextual memory, and the DIKWP framework maps the model output to different cognitive levels to ensure that the AI's decisions have a clear purpose orientation.

[0028] The Emotional Resonance Layer (Emotional Layer) is responsible for user emotion recognition and response regulation. Based on the multimodal features provided by the Perception and Understanding Layer, this layer uses an emotional computing model to analyze the user's current emotional state. For example, it detects the user's excitement level through tone and energy in their voice, combines this with the emotional tendency analysis of the text content (positive / negative), and visual expression recognition results to comprehensively judge the user's emotion. After recognizing the user's emotion, the Emotional Resonance Layer formulates different response strategies based on different emotions: adjusting the language style, wording, and content of the generated response to achieve an empathetic effect. For example, it offers comfort to sad users, expresses understanding and calmly advises angry users, and maintains positive interaction with happy users. This layer also includes an emotional memory unit to track the user's emotional change trends in long-term dialogue interactions, thereby demonstrating continued care and understanding in subsequent communications. Through the regulation of the emotional layer, the AI's output is no longer rigid and objective information, but rather communication imbued with appropriate emotional warmth, making it closer to interpersonal interaction.

[0029] The Ethical Decision-Making Layer (Ethics Layer) is responsible for the moral evaluation and compliance generation of behavioral recommendations. Building upon the initial responses provided by the Perception and Emotional Layers, this layer introduces ethical judgment to conduct a final review and adjustment of the response content. Internally, the Ethics Layer constructs a local cultural ethical knowledge graph, containing knowledge of ethical principles, legal norms, customs and taboos, and value entries, as well as a reasoning engine to handle complex ethical relationships. Furthermore, a set of dynamically adjustable value weights is set to balance the relative importance of different ethical principles (e.g., "privacy" vs. "security," "individual interest" vs. "collective interest") in different contexts. When the engine generates responses or decisions, the Ethics Layer processes them as follows:

[0030] Ethics and compliance check: The system uses a knowledge graph to match responses to determine if the topics are sensitive or violate regulations, such as legal prohibitions, hate speech, discriminatory content, or local cultural taboos. If a violation is found, the relevant output is immediately modified or blocked, and the user is responded to with a polite explanation (e.g., refusing to provide illegal advice and explaining the reasons).

[0031] Ethical Decision-Making: For problems that are not black and white (such as ethical dilemmas), a rule engine based on a knowledge graph is used to reason, listing possible options and the ethical factors involved in each. Then, dynamic value weights are applied for quantitative evaluation. For example, in medical decision-making scenarios, the patient's autonomy, the value of life, and the principle of informed consent are considered simultaneously. The weights of each factor are adjusted according to the context, and the option with the highest comprehensive score is selected as the output suggestion. The dynamic value weight mechanism ensures that AI decision-making is not static and rigid, but can be optimized and adjusted according to changes in the context and user value preferences. This mechanism is similar to equipping AI with a "moral spectrum analyzer," which can analyze the value weight distribution behind different decisions, thereby making choices that conform to specific cultural and individual values.

[0032] Semantic Refinement and Explanation: Before the final output, the ethics layer refines the semantics of the responses to ensure that the wording conforms to etiquette norms and does not offend users. For example, when giving a refusal or correction, it is expressed in a polite and respectful tone. At the same time, simple explanations are added when necessary to let users understand the ethical basis of the AI's behavior, increasing transparency and trust.

[0033] With the ethical decision-making layer overseen, the AI's responses are more secure, reliable, and compliant with human norms in terms of value orientation. These three layers, each with its own function yet closely interacting, collectively form the core of this invention's engine. It's worth noting that the principles of Integrated Information Theory (IIT) permeate the architectural design: the high degree of information integration and feedback between the engine's layers increases the overall system's Φ value (information integration degree), implying the emergence of a higher level of artificial "consciousness." The cognitive, emotional, and ethical layers work together to enable the system to comprehensively consider any input scenario from multiple perspectives—factual, emotional, and value-based—demonstrating a holistic and coherent nature similar to human consciousness.

[0034] During runtime, the engine of this invention processes user interactions in the order of cognition → emotion → ethics, and continuously optimizes through a feedback loop. The process is as follows: Figure 1 As shown, the main steps include:

[0035] Multimodal input acquisition: Users interact with the system through natural language dialogue, voice, images, etc. The system simultaneously acquires data such as text content, voice tone, and facial expressions. For example, a user might say in voice, "I'm in a bad mood today because of a lot of work pressure." The system will transcribe the recording into text and analyze signals such as a low tone and a frustrated facial expression.

[0036] Intent Understanding and Semantic Parsing: The perception understanding layer first performs word segmentation, syntactic parsing, and semantic understanding on the text content, extracting key information such as "the user is in a bad mood" and "the reason is work pressure" by combining context and knowledge base. Simultaneously, referring to the DIKWP model, it extracts objective facts at the data level (e.g., the user mentions "work pressure"), specific situations at the information level ("high work pressure leads to a bad mood"), general laws at the knowledge level (e.g., "long-term stress can affect mental health"), and judgments at the wisdom level (perhaps the user needs stress reduction advice). Finally, at the purpose level, it infers the user's implicit intent, which may be seeking understanding and advice (Purpose: hoping to obtain psychological comfort or stress reduction methods). Integrating information theory at this stage ensures the comprehensive utilization of multi-source information, thereby forming a holistic understanding of user needs.

[0037] Emotion Recognition and Empathy Strategy: The emotion resonance layer then processes the user's emotional cues in language and voice. Using a pre-trained emotion recognition model, the system determines that the user is in a depressed / frustrated emotional state (e.g., an emotion classification score pointing to "sad"). The emotion layer then selects an empathy response strategy: preparing a reply with a comforting and supportive tone, while emphasizing empathy and encouragement in the content. At this stage, the system may invoke a built-in emotion dictionary or generative model to select appropriate comforting phrases, such as "I understand how you feel," setting the emotional tone for the final reply.

[0038] Initial Response Generation: Combining the intent provided by the cognitive layer (the user needs psychological comfort / suggestion) and the tone and inflection scheme provided by the emotional layer, the system generates an initial natural language response. For example: "I'm sorry to hear you're under so much pressure. I completely understand how you feel. Everyone experiences work-related stress sometimes. Try some relaxing activities, like listening to music or taking a walk, to adjust your state." This response semantically addresses the user's question (providing stress-relief suggestions) and emotionally expresses understanding and comfort.

[0039] Ethical and Value Review: The generated initial response is reviewed by the ethics decision-making layer. First, the system checks the ethical compliance of the response content: in this example, the response contains no illegal or inappropriate information and complies with ethical norms, thus passing the review. Next, the system assesses whether there are any value conflicts requiring further consideration. For example, providing stress reduction advice involves the user's health and well-being, corresponding to positive ethical values, and there is no conflict. However, if the user requests advice with potential risks (such as medication), the system will weigh safety principles against user autonomy principles before deciding whether to provide it or how to word it. In this example, no special adjustments are needed. Finally, the ethics layer confirms that the response is polite in tone, uses appropriate wording, and contains no offensive or discriminatory elements, thus passing the review.

[0040] Final response output: After ethical review and approval, the system will send the final response to the user. Simultaneously, the DIKWP intent map, emotional state, and ethical decision record of this conversation will be stored in the session memory for future contextual reference.

[0041] Human feedback collection and learning: After receiving a response, users may continue the conversation if satisfied; otherwise, they can inform the system of the problem through the feedback interface. For example, a user might comment, "The suggestion is good, but I feel it's a bit formulaic and lacks genuine empathy." Upon receiving this feedback, the system categorizes it as a deficiency at the emotional level and records it in the "Ethical Intent Feedback Model." Periodically, the system summarizes user feedback and adjusts model parameters or adds training instances accordingly. For example, this example might prompt the system to increase the personalization of responses to specific user situations in similar future scenarios (avoiding clichés). For serious erroneous feedback (such as giving inappropriate advice that causes user resentment), the system will adjust the weights of the corresponding modules more quickly and add new rules if necessary to prevent recurrence. Through this human-in-the-loop continuous training mechanism, the capabilities of each layer of the system will continuously evolve: the cognitive layer gradually calibrates the model for understanding user intent, the emotional layer optimizes the empathy algorithm (e.g., introducing more dialogue examples to learn more natural empathetic expressions), and the ethical layer enriches the knowledge graph and rule base and dynamically adjusts value weights to adapt to the latest social norms.

[0042] The entire process embodies closed-loop feedback and continuous optimization: the AI ​​does not operate statically, but rather has the opportunity to improve itself through user feedback after each interaction. This dynamic evolutionary capability is particularly crucial in matters of ethics and values. The system updates its moral knowledge graph monthly or even more frequently, combining feedback from users in different regions to ensure coverage of global cultural customs and achieve localized adaptation of values. This mechanism enables the AI ​​to keep up with changes in social and population values, becoming increasingly "thoughtful" and reliable over long-term use.

[0043] In summary, this technical solution, through a layered yet closely collaborative architecture, achieves end-to-end artificial intelligence enhancement, from perception and understanding to emotional resonance and ethical decision-making. Regardless of the application scenario, the engine first understands what the user wants (intent), comprehends how the user feels (emotion), then judges whether and how the user should act (ethics), and ultimately provides a response that aligns with both the user's surface needs and their deeper values. This capability elevates the humanistic qualities of artificial intelligence to a new level.

Claims

1. An artificial consciousness enhanced engine with the ability of intention understanding, emotional resonance and moral judgment, comprising a multi-modal information input interface, an intention understanding module, an emotional resonance module, a moral judgment module and a human-in-the-loop feedback interface, characterized in that: the intention understanding module performs semantic analysis on the user's natural language and multi-modal input based on the DIKWP model, constructs the user's data-information-knowledge-wisdom-purpose cognitive graph, and identifies the user's core intention; the emotional resonance module analyzes the user's emotional state using a cross-modal emotion recognition model, and adjusts the engine's response through a semantic emotion mapping adjustment engine, so that the response content can make empathetic expressions or soothing feedback to the user's emotions; the moral judgment module combines local cultural and ethical knowledge graph and dynamic value weight mechanism to conduct moral compliance review and ethical conflict judgment on the behavior suggestions or response content generated by the engine, and adjusts or filters the output as needed to ensure compliance with predetermined ethical norms and value criteria; the human-in-the-loop feedback interface allows users to label and correct the engine's understanding bias or inappropriate response, and the feedback data is used to train and update the engine's ethical intention feedback model, thereby continuously optimizing the engine's performance in intention recognition, emotional resonance and moral judgment. The intention understanding module includes a natural language processing unit and a cognitive reasoning unit, the natural language processing unit extracts the key information and context of the user's input, and the cognitive reasoning unit maps the information to the DIKWP five-layer cognitive model to infer the user's implicit intention at the purpose layer; this module preferably further accesses external knowledge bases or knowledge graphs to enrich the reasoning process of the knowledge layer and wisdom layer, making the intention recognition more accurate. The emotional resonance module includes a multi-modal emotion analysis unit and an emotional response generation unit, the former simultaneously processes user data such as text, speech tone and facial expressions to determine emotion categories and intensity; the latter selects matching tone and language from a pre-set emotional corpus or generation model to construct a response sentence based on the user's identified emotions, so that the final output has a tone and content that matches the user's emotions. The emotional resonance module also includes an emotional memory mechanism to track long-term emotional changes in users and make appropriate adjustments. The moral judgment module further includes an ethical rule engine and a value assessment unit; the ethical rule engine uses the local cultural and ethical knowledge graph to perform rule reasoning and constraint detection on the engine's preliminary response, identifies potential ethical issues or culturally inappropriate content, and executes predefined handling strategies; the value assessment unit scores possible response options based on multiple predefined value dimensions, and the weights of the value dimensions can be dynamically adjusted according to application scenarios or user preferences to achieve an optimal compromise decision when there is an ethical conflict. When the ethical rule engine finds serious violations, the system directly prohibits output and feeds back a safety warning; when there is an ethical dilemma, the value assessment unit balances the weighted scores of each candidate solution and selects the most ethical one to output. ​ 2. The artificial consciousness-augmented engine of claim 1, wherein: ​ 3. The artificial consciousness-augmented engine of claim 1, wherein: ​ 4. The artificial consciousness-augmented engine of claim 1, wherein: ​ 5. The artificial consciousness-augmented engine of claim 1, wherein: The human-in-the-loop feedback interface comprises a user evaluation interface and a feedback processing unit, the user evaluation interface being configured to collect user satisfaction scores, problem labels, or desired modification suggestions for the engine output; The feedback processing unit classifies and stores the collected feedback data, and periodically triggers a model training process to update the parameters or adjust the rules of one or more modules of the engine. In particular, for ethics-related feedback, the feedback processing unit updates the weights of relevant nodes in the ethics knowledge graph or adds new rule clauses; for feedback on intent recognition errors, the inference model of the purpose layer in the DIKWP model is updated; for feedback on insufficient emotional resonance, the corpus and strategies of the emotional response generation unit are expanded. Through the above continuous feedback training mechanism, the engine can continuously improve its adaptability to human intentions, emotions, and values after deployment.

6. The artificial consciousness-augmented engine of any one of claims 1 to 5, wherein: The engine adopts a layered architecture, and the modules of each layer communicate through standardized interfaces. The perception understanding layer is used for semantic and intent processing, the emotional resonance layer is used for emotion processing, and the ethical decision-making layer is used for moral constraint processing. The architecture combines the design principles of integrated information theory (IIT), enabling the information processed by the above layers to be highly integrated within the system, thereby improving the overall awareness level and decision-making coherence of the engine. During operation, the engine integrates and calculates the information of multi-modal inputs, and the integrated information quantity index Φ increases, reflecting the effect of enhanced artificial consciousness.

Images