US20260288819
2026-09-24
Physics
G06F16/33295
The system employs a processor to coordinate multiple generative artificial intelligence models to execute tasks simultaneously. It generates prompts for these models, which then produce answers that are evaluated by other AI models. This process ensures the selection of a final answer based on evaluation results, enhancing the reliability and consistency of responses. Furthermore, the system analyzes the user's emotional state, providing this analysis to the AI models to refine and personalize the dialogue.
This application claims priority from a Japanese patent filed in 2025. It addresses limitations in conventional dialog systems that use a single AI model, which often leads to unreliable or emotionally insensitive responses. Previous systems lacked mechanisms for cross-checking answers or considering the user's emotional state, reducing user satisfaction and trust.
The system relates to AI-driven dialogue systems and builds upon existing methods that utilize persona-based chatbot controls. Unlike traditional systems, it leverages multiple AI models to generate and evaluate responses, ensuring better accuracy and emotional alignment with the user. This approach solves issues of factual inaccuracy and emotional insensitivity found in prior art.
The processor generates prompts for parallel execution by multiple AI models, which independently produce answers. These answers undergo evaluation by other models, considering criteria like validity, consistency, and emotional appropriateness. The system selects the highest-scoring answer, aligning it with the user's emotional state to optimize dialogue quality.
Key components include the processor, memory, storage, and communication interfaces. The processor may consist of various computational units such as CPUs or GPUs. Memory serves as temporary storage for processing tasks, while non-volatile storage retains programs and parameters. Communication interfaces facilitate connectivity using standards like 5G or Wi-Fi.