{"title":"通过体验式学习增强人工智能:分析用户生成内容的含义","authors":"Ashutosh Singh , Reeti Agarwal , Rsha Alghafes , Armando Papa","doi":"10.1016/j.techfore.2025.124261","DOIUrl":null,"url":null,"abstract":"<div><div>Artificial intelligence (AI) has evolved into generative artificial intelligence, offering users even greater benefits. The AI platforms provide generative AI-related services to support users' professional development and gather feedback to enhance the service through experiential learning. However, comprehending large volumes of unstructured datasets in the form of customer reviews presents an increasingly serious challenge as the number of users on AI platforms grows over time. We employ advanced machine learning techniques-topic modelling and word2vec- to extract more accurate insights from unstructured data. We collect customer reviews from AI content-creation platforms from 2022 to 2024. By combining topic modelling and word2vec, we uncover valuable insights. Our analysis identifies eight key topics: Playground, Support Hub, Content Lab, Productivity, User Experience, Access, Business Assistant, and Remix. The topic of regression analysis reveals that Content Lab, User Experience, Business Assistant, and Remix are more favourable in terms of customer satisfaction scores. The word2vec analysis with negative sampling indicates that Access and Playground demonstrate better cohesion scores compared to other themes. Conversely, themes such as Content Lab, Productivity, and Business Assistant have lower cohesion scores, indicating weak clustering among words within these themes. Our research offers several valuable insights for AI platform managers, which can further enhance services through experiential learning.</div></div>","PeriodicalId":48454,"journal":{"name":"Technological Forecasting and Social Change","volume":"219 ","pages":"Article 124261"},"PeriodicalIF":13.3000,"publicationDate":"2025-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Empowering AI with experiential learning: Implications from analysing user-generated content\",\"authors\":\"Ashutosh Singh , Reeti Agarwal , Rsha Alghafes , Armando Papa\",\"doi\":\"10.1016/j.techfore.2025.124261\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><div>Artificial intelligence (AI) has evolved into generative artificial intelligence, offering users even greater benefits. The AI platforms provide generative AI-related services to support users' professional development and gather feedback to enhance the service through experiential learning. However, comprehending large volumes of unstructured datasets in the form of customer reviews presents an increasingly serious challenge as the number of users on AI platforms grows over time. We employ advanced machine learning techniques-topic modelling and word2vec- to extract more accurate insights from unstructured data. We collect customer reviews from AI content-creation platforms from 2022 to 2024. By combining topic modelling and word2vec, we uncover valuable insights. Our analysis identifies eight key topics: Playground, Support Hub, Content Lab, Productivity, User Experience, Access, Business Assistant, and Remix. The topic of regression analysis reveals that Content Lab, User Experience, Business Assistant, and Remix are more favourable in terms of customer satisfaction scores. The word2vec analysis with negative sampling indicates that Access and Playground demonstrate better cohesion scores compared to other themes. Conversely, themes such as Content Lab, Productivity, and Business Assistant have lower cohesion scores, indicating weak clustering among words within these themes. Our research offers several valuable insights for AI platform managers, which can further enhance services through experiential learning.</div></div>\",\"PeriodicalId\":48454,\"journal\":{\"name\":\"Technological Forecasting and Social Change\",\"volume\":\"219 \",\"pages\":\"Article 124261\"},\"PeriodicalIF\":13.3000,\"publicationDate\":\"2025-07-04\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Technological Forecasting and Social Change\",\"FirstCategoryId\":\"91\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S0040162525002926\",\"RegionNum\":1,\"RegionCategory\":\"管理学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"BUSINESS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Technological Forecasting and Social Change","FirstCategoryId":"91","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0040162525002926","RegionNum":1,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"BUSINESS","Score":null,"Total":0}
Empowering AI with experiential learning: Implications from analysing user-generated content
Artificial intelligence (AI) has evolved into generative artificial intelligence, offering users even greater benefits. The AI platforms provide generative AI-related services to support users' professional development and gather feedback to enhance the service through experiential learning. However, comprehending large volumes of unstructured datasets in the form of customer reviews presents an increasingly serious challenge as the number of users on AI platforms grows over time. We employ advanced machine learning techniques-topic modelling and word2vec- to extract more accurate insights from unstructured data. We collect customer reviews from AI content-creation platforms from 2022 to 2024. By combining topic modelling and word2vec, we uncover valuable insights. Our analysis identifies eight key topics: Playground, Support Hub, Content Lab, Productivity, User Experience, Access, Business Assistant, and Remix. The topic of regression analysis reveals that Content Lab, User Experience, Business Assistant, and Remix are more favourable in terms of customer satisfaction scores. The word2vec analysis with negative sampling indicates that Access and Playground demonstrate better cohesion scores compared to other themes. Conversely, themes such as Content Lab, Productivity, and Business Assistant have lower cohesion scores, indicating weak clustering among words within these themes. Our research offers several valuable insights for AI platform managers, which can further enhance services through experiential learning.
期刊介绍:
Technological Forecasting and Social Change is a prominent platform for individuals engaged in the methodology and application of technological forecasting and future studies as planning tools, exploring the interconnectedness of social, environmental, and technological factors.
In addition to serving as a key forum for these discussions, we offer numerous benefits for authors, including complimentary PDFs, a generous copyright policy, exclusive discounts on Elsevier publications, and more.