{"title":"The threat of AI chatbot responses to crowdsourced open-ended survey questions","authors":"Frederic Traylor","doi":"10.1016/j.erss.2024.103857","DOIUrl":null,"url":null,"abstract":"<div><div>Many researchers use crowdsourced and online surveys and open-ended survey questions to gather new ideas and concerns from the public regarding emerging energy and climate technologies. The availability of large language model chatbots, most notably ChatGPT, presents a threat to the utility of these approaches. While closed-ended questions and paradata analysis have been previously used to screen inattentive and ingenuine respondents, and open-ended question responses can be manually examined for unrelated content, these chatbots can mimic satisfactory answers to evade typical detection. Using a question asking people about their thoughts about hydrogen energy, in a side-by-side comparison of a managed panel sample (<em>N</em> = 834) to one drawn from Amazon Mechanical Turk (<em>n</em> = 1166), I find that responses from the latter appeared higher-quality but were more likely to originate from AI chatbots, even after screening based on close-ended questions and survey paradata. Survey designs should thus incorporate structural changes to prevent fraudulent responses, and analysis going forward must improve methods to detect them.</div></div>","PeriodicalId":48384,"journal":{"name":"Energy Research & Social Science","volume":"119 ","pages":"Article 103857"},"PeriodicalIF":6.9000,"publicationDate":"2024-11-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Energy Research & Social Science","FirstCategoryId":"96","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2214629624004481","RegionNum":2,"RegionCategory":"经济学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"ENVIRONMENTAL STUDIES","Score":null,"Total":0}
引用次数: 0
Abstract
Many researchers use crowdsourced and online surveys and open-ended survey questions to gather new ideas and concerns from the public regarding emerging energy and climate technologies. The availability of large language model chatbots, most notably ChatGPT, presents a threat to the utility of these approaches. While closed-ended questions and paradata analysis have been previously used to screen inattentive and ingenuine respondents, and open-ended question responses can be manually examined for unrelated content, these chatbots can mimic satisfactory answers to evade typical detection. Using a question asking people about their thoughts about hydrogen energy, in a side-by-side comparison of a managed panel sample (N = 834) to one drawn from Amazon Mechanical Turk (n = 1166), I find that responses from the latter appeared higher-quality but were more likely to originate from AI chatbots, even after screening based on close-ended questions and survey paradata. Survey designs should thus incorporate structural changes to prevent fraudulent responses, and analysis going forward must improve methods to detect them.
期刊介绍:
Energy Research & Social Science (ERSS) is a peer-reviewed international journal that publishes original research and review articles examining the relationship between energy systems and society. ERSS covers a range of topics revolving around the intersection of energy technologies, fuels, and resources on one side and social processes and influences - including communities of energy users, people affected by energy production, social institutions, customs, traditions, behaviors, and policies - on the other. Put another way, ERSS investigates the social system surrounding energy technology and hardware. ERSS is relevant for energy practitioners, researchers interested in the social aspects of energy production or use, and policymakers.
Energy Research & Social Science (ERSS) provides an interdisciplinary forum to discuss how social and technical issues related to energy production and consumption interact. Energy production, distribution, and consumption all have both technical and human components, and the latter involves the human causes and consequences of energy-related activities and processes as well as social structures that shape how people interact with energy systems. Energy analysis, therefore, needs to look beyond the dimensions of technology and economics to include these social and human elements.