A Comparative Study of "Knowledge" Produced through Traditional Social Research Methods and AI: Reanalyses of Existing Comic Market Surveys

2022 Fiscal Year Final Research Report

• Project/Area Number: 20K20778
• Research Category: Grant-in-Aid for Challenging Research (Exploratory)
• Allocation Type: Multi-year Fund
• Review Section: Medium-sized Section 8:Sociology and related fields
• Research Institution: Kyushu University
• Principal Investigator: Sugiyama Akashi 九州大学, 比較社会文化研究院, 准教授 (60222056)
• Project Period (FY): 2020-07-30 – 2023-03-31
• Keywords: 社会調査法 / AI

Outline of Final Research Achievements

The purpose of this study was to compare the reanalysis of existing social survey data using AI with conventional statistical analysis and to explore the characteristics of each method. In the AI analysis, certain variables were made to predict after data learning. It succeeded in predicting with considerable accuracy, but due to the magnitude of computer processing power, it soon became overlearned, making it impossible to evaluate it as a result of the AI analysis. The data for reanalysis were two surveys with about 40,000 valid responses each, an exceptionally large number for a survey in the field of cultural sociology, but still not enough at all. On the other hand, the results of the estimation based on socio-statistical methods were not much different from those obtained by AI. It was confirmed that the conventional method is a method to obtain effective "knowledge" from small information quantity.

Free Research Field

文化社会学

Academic Significance and Societal Importance of the Research Achievements

現在、マーケティングや行政、政治戦略の構築において、これまで主流であった社会調査的手法に代わって、AIによる分析が広範に利用されつつある。両者は同じことを行う別の手法であると受け取られやすいが、比較的少数のデータから統計指標へと抽象化された“事実”を抽出する社会調査と、抽象レベルはソフト任せで、ビッグデータから個別の“現実”を予測するAIの差異は大きい。両手法の特徴を探った本研究は、人間にとって、社会を認識するということが何なのか、その含意を示唆するものとなった。