{"title":"人们对标注为人工智能生成的标题持怀疑态度,即使是真实的或人为的,因为他们假定人工智能已经完全自动化。","authors":"Sacha Altay, Fabrizio Gilardi","doi":"10.1093/pnasnexus/pgae403","DOIUrl":null,"url":null,"abstract":"<p><p>The rise of generative AI tools has sparked debates about the labeling of AI-generated content. Yet, the impact of such labels remains uncertain. In two preregistered online experiments among US and UK participants (<i>N</i> = 4,976), we show that while participants did not equate \"AI-generated\" with \"False,\" labeling headlines as AI-generated lowered their perceived accuracy and participants' willingness to share them, regardless of whether the headlines were true or false, and created by humans or AI. The impact of labeling headlines as AI-generated was three times smaller than labeling them as false. This AI aversion is due to expectations that headlines labeled as AI-generated have been entirely written by AI with no human supervision. These findings suggest that the labeling of AI-generated content should be approached cautiously to avoid unintended negative effects on harmless or even beneficial AI-generated content and that effective deployment of labels requires transparency regarding their meaning.</p>","PeriodicalId":74468,"journal":{"name":"PNAS nexus","volume":null,"pages":null},"PeriodicalIF":2.2000,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11443540/pdf/","citationCount":"0","resultStr":"{\"title\":\"People are skeptical of headlines labeled as AI-generated, even if true or human-made, because they assume full AI automation.\",\"authors\":\"Sacha Altay, Fabrizio Gilardi\",\"doi\":\"10.1093/pnasnexus/pgae403\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><p>The rise of generative AI tools has sparked debates about the labeling of AI-generated content. Yet, the impact of such labels remains uncertain. In two preregistered online experiments among US and UK participants (<i>N</i> = 4,976), we show that while participants did not equate \\\"AI-generated\\\" with \\\"False,\\\" labeling headlines as AI-generated lowered their perceived accuracy and participants' willingness to share them, regardless of whether the headlines were true or false, and created by humans or AI. The impact of labeling headlines as AI-generated was three times smaller than labeling them as false. This AI aversion is due to expectations that headlines labeled as AI-generated have been entirely written by AI with no human supervision. These findings suggest that the labeling of AI-generated content should be approached cautiously to avoid unintended negative effects on harmless or even beneficial AI-generated content and that effective deployment of labels requires transparency regarding their meaning.</p>\",\"PeriodicalId\":74468,\"journal\":{\"name\":\"PNAS nexus\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":2.2000,\"publicationDate\":\"2024-10-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11443540/pdf/\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"PNAS nexus\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1093/pnasnexus/pgae403\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"MULTIDISCIPLINARY SCIENCES\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"PNAS nexus","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1093/pnasnexus/pgae403","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"MULTIDISCIPLINARY SCIENCES","Score":null,"Total":0}
People are skeptical of headlines labeled as AI-generated, even if true or human-made, because they assume full AI automation.
The rise of generative AI tools has sparked debates about the labeling of AI-generated content. Yet, the impact of such labels remains uncertain. In two preregistered online experiments among US and UK participants (N = 4,976), we show that while participants did not equate "AI-generated" with "False," labeling headlines as AI-generated lowered their perceived accuracy and participants' willingness to share them, regardless of whether the headlines were true or false, and created by humans or AI. The impact of labeling headlines as AI-generated was three times smaller than labeling them as false. This AI aversion is due to expectations that headlines labeled as AI-generated have been entirely written by AI with no human supervision. These findings suggest that the labeling of AI-generated content should be approached cautiously to avoid unintended negative effects on harmless or even beneficial AI-generated content and that effective deployment of labels requires transparency regarding their meaning.