Generative AI Thinking Evaluation System

The generative AI thinking evaluation system analyzes semantic transformations of input sentences to objectively assess cognitive progression, addressing the limitations of conventional evaluation methods by providing a quantitative thinking depth score and visual feedback.

JP7883813B1Active Publication Date: 2026-07-02小澤 暢吾
View PDF 6 Cites 0 Cited by

Patent Information

Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
小澤 暢吾
Filing Date
2026-01-26
Publication Date
2026-07-02

AI Technical Summary

Technical Problem

Conventional educational evaluation methods struggle to objectively assess the thinking process of learners interacting with generative AI, particularly in evaluating the semantic transformation of input sentences and the depth of thinking, as they often rely on final products rather than the thought process itself.

Method used

A generative AI thinking evaluation system that analyzes the semantic transformation of input sentences using semantic vectors, classifies these transformations into predefined types, and calculates a thinking depth score to objectively evaluate the learner's cognitive progression.

Benefits of technology

Enables objective and quantitative evaluation of the learner's thinking process, promoting self-reflection and enhancing educational evaluation by providing a visual representation of cognitive development and supporting instructional decisions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007883813000001_ABST
    Figure 0007883813000001_ABST
Patent Text Reader

Abstract

This system provides an objective evaluation of the thought process of learners through the transformation of the questions they input into a generative AI during learning using dialogue with the AI, without relying on the output. [Solution] The Generative AI Thinking Evaluation System 10 acquires multiple input sentences entered by the learner 2 into the Generative AI in chronological order, converts each input sentence into a semantic vector to calculate the semantic difference, and classifies the changes in the input sentences into predetermined change types based on that difference. Furthermore, it evaluates the degree of deepening of thinking based on the change type and presents the result as reflection support information.
Need to check novelty before this filing date? Find Prior Art