欧洲人工智能（AI）技能、使用与态度概要结果

Artificial Intelligence (AI) in Europe Lasse Marz and Anja-Kristin AbendrothBielefeld University Topline Results Skills, Use and Attitudes towardsArtificial Intelligence (AI) in Europe Lasse Marz and Anja-Kristin Abendroth, Bielefeld University Introduction Artificial intelligence (AI) – thedigital technology capableof producing its own content(be it videos, images or text),making data-based decisions,and simulating humancommunication – stressors at work. With theEU AI Act, which was adoptedin 2024, policymakers havealready responded to thesedevelopments and created alegal framework aiming to definewhat AI is allowed to be used for. The EUpromotes a humancentred approachby, for example,prohibiting chatbotsfrom pretending tobe human or AI frommaking autonomousdecisions withouthuman oversight. has the potential totransform people’slives. We first investigateEuropeans’ self-assessed ICT andAI skills in terms oftext creation andinformation literacy.Secondly, we observehow frequently they useAI for text creation andadaptation. As AI-generatedcontent becomesharder to identify, itincreases the riskof disinformationbecause manipulatedor false informationcan appear morecredible and spreadmore easily. This is particularlytrue on the Internet and in socialmedia. In the workplace, AI canbe used to support and evenreplace several managerialand occupational tasks. Thisraises hopes about efficiencygains, but also concerns aboutjob loss, restrictions in jobautonomy, deskilling or new The ‘Make it Digital’module of theCRONOS-3 panel provides afirst insight into Europeans’ AIskill, AI use and views on AIand technological developmentin general. The CRONOS-3panel is a cross-national studyconsisting of data collectedbetween October 2024 and May2025 in 11 European countries(Austria, Belgium, Czechia, Finland, France, Hungary, Iceland,Poland, Portugal Slovenia, and theUnited Kingdom). About 12,000panellists were recruited from thetwo latest rounds of the EuropeanSocial Survey (rounds 10 and11). The data was collected overfive waves. Each Thirdly, we report the managerialtasks for which employeeswould allow their employers toimplement AI in workplaces, andinvestigate whether they believejobs in their field may be at riskbecause of AI. Finally, we analysewhat kind of changes for the societyindividuals expectthrough technologicaladvancements, suchas ChatGPT or socialmedia platforms. questionnaireconsisted ofabout 90 itemsand included fivedifferent topicslabelled as Green,Healthy, Equal,Strong and Digital.This report drawsupon the module“Make it Digital” from Waves 1, 4 ...we analysewhat kind of changesfor the societyindividuals expectthrough technologicaladvancements, suchas ChatGPT or socialmedia platforms. All of our analysesinclude groupcomparisons, eitherby age, education, or country. We choose this approach becauseresearch on digital inequalitieshas shown that they shape digitalskills, as well as use, acceptance,and perceptions of informationand communication technologies.Thus, they are likely to matterfor the prevalence of AI skills,and use, as well the levels ofconcerns. and 5 of the CRONOS-3 Panel. We first investigate Europeans’self-assessed ICT (Informationand Communication Technology)and AI skills in terms of textcreation and information literacy.Secondly, we observe howfrequently they use AI for textcreation and adaptation. ICT and Artificial Intelligence Skills asked about their ability to checkthe truthfulness of informationand content they found on theInternet. Overall, Europeansare quite confident to verifyinformation and/or content(Figure 1) but we observeconsiderable gaps between agegroups and educational levels.1 Since digital devices havebecome widespread throughoutsociety, people’s ability touse them – often referred toas digital or ICT skills – hasbecome of central importancefor job prospects and generalsocial participation. Advancingtechnologies and the rise ofmisinformation have required theindividual to adapt constantly,with Artificial Intelligence (AI)tools the most recent example todemonstrate this challenge. Nearly every respondent agedunder 35 years old is at leastsomewhat confident that theycan check the truthfulness ofinformation and content they findon the internet (96.4%). OlderEuropeans are a little less likelyto report this information literacyskill (83.7%). In CRONOS-3, respondents wereasked to assess to what extentthey know how to carry outcertain activities, some of whichexplicitly included the presence ofAI tools. The activities representoverarching competencydimensions, such as informationliteracy and content creation. Figure 1 also shows thatindividuals with higher levelsof education are more likelyto feel confident that they cancombat misinformation on theInternet. Only half of those withthe lowest levels of educationfeel completely confident in thisability, compared to almost four infive people who are educated tothe highest level. Information literacy refers toi