Virtual Person Generation via Personality-to-Prompt Conversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The configuration of generative AI systems like ChatGPT requires domain expertise and programming skills, making it challenging for domain experts to effectively set up and adapt these systems without costly trial and error.
Innovation Solution
An automated method for generating a virtual person involves receiving user-input personality parameters, converting them into machine-readable prompts, and using a large language model to generate a virtual person, allowing domain experts to configure the system without programming knowledge.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If domain experts configure generative AI systems through trial and error, then the system can be set up, but it requires costly access and repeated attempts
Solution Approach 1:
The patent introduces an automated configuration system as an intermediary between domain experts and the generative AI system. This intermediary translates human-understandable personality parameters into machine-readable prompts automatically, eliminating the need for experts to directly interact with complex system configuration interfaces through trial and error.
Solution Approach 2:
The patent replaces the mechanical process of manual trial-and-error configuration with an automated computational process. The system automatically converts personality parameters into prompts and generates virtual persons, substituting the iterative manual adjustment mechanism with a direct automated transformation process.
2Ease of operation
If software developers program the configuration instructions, then the system can be implemented, but domain experts lack programming skills
Solution Approach 1:
The automated configuration system serves as an intermediary layer that translates between the simple human-understandable personality parameters that domain experts can provide and the complex machine-readable prompts required by the generative AI system. This eliminates the need for domain experts to learn programming skills while maintaining system functionality.
Solution Approach 2:
The system performs self-service by automatically converting personality parameters into configuration prompts without requiring user intervention in the technical conversion process. Domain experts simply provide personality parameters in human-understandable formats, and the system autonomously handles the technical transformation and virtual person generation.
3Ease of operation
If domain experts provide personality parameters in human-understandable format, then they can configure without programming skills, but the system needs automated conversion to machine-readable format
Solution Approach 1:
The patent replaces manual conversion processes with automated computational conversion. The system automatically transforms human-understandable personality parameters into machine-readable prompt formats, eliminating the need for domain experts to manually format their inputs while maintaining full compatibility with the generative AI system.
Solution Approach 2:
The conversion process is handled by the system itself rather than requiring user involvement. The automated configuration system performs the translation from human-understandable to machine-readable formats autonomously, allowing domain experts to simply provide parameters in their natural format without technical overhead.
Data Source
Figure 1
AI summary
The disclosure relates to an automated method for generating a virtual person (VP) for interaction with a real person (RP), the method comprising the steps: - generating (S1) a user interface, the user interface requesting a personality parameter for the virtual person, - receiving (S2) the personality parameter via a receiving option, wherein the personality parameter is received in a human-understandable format, - converting (S3) the personality parameter into a prompt for a large language model, the prompt being in a machine-readable format according to a programming language, - generating (S4) a digital model based on the prompt by the large language model, and - generating (S5) the virtual person (VP) for interaction with the real person (RP) based on the digital model. A corresponding system, a computer program product, and a computer-readable storage medium are disclosed as well.