Impact of AI on Romanian Businesses (survey dataset description)

Impact of AI on Romanian Businesses (survey dataset description) BACKGROUND AND RATIONALE Rapid advances in artificial intelligence (AI) have prompted growing interest in how these technologies affect businesses worldwide. However, evidence from recent statistics indicates that AI adoption in Romania remains nascent. In 2023, only about 1.5% of Romanian enterprises (with ≥10 employees) had implemented AI solutions – the lowest share in the EU, compared to roughly 8% EU-wide (Eurostat, 2024)[1][2]. This gap underscores the importance of understanding local drivers and barriers to AI uptake. Romanian policymakers have also emphasized the need for strategic action: a National AI Strategy was launched to maximize the benefits of AI integration and align with European trends (Autoritatea pentru Digitalizarea României & MCID, 2024). Within the private sector, there is a mix of enthusiasm and caution. For instance, a recent entrepreneurship survey found 71% of Romanian firms had increased their use of AI in the past year (EY Romania, 2025)[3], yet 37% reported challenges in training employees on AI tools[4]. In this context, the present survey was conceived to systematically assess the impact of AI on businesses in Romania, filling an empirical gap by capturing business stakeholders’ experiences, readiness, and concerns regarding AI. The survey addresses a wide array of themes – from workforce implications and cybersecurity challenges to changes in financial services, ethical considerations, and government policy related to AI – reflecting the multifaceted nature of AI’s business impact[5]. The primary rationale for the project was to gather robust data that can inform both academic analysis and practical strategies for AI adoption in the Romanian business environment. SURVEY STRUCTURE AND INSTRUMENT The “Impactul Inteligenței Artificiale asupra Afacerilor din România” survey is a structured questionnaire developed in Romanian language using the LimeSurvey online platform. It was designed as a self-administered web survey targeting individuals knowledgeable about their organisation’s use of AI (e.g. executives, managers, IT specialists, entrepreneurs). The instrument comprises 20 questions organised into nine thematic sections, ensuring coverage of both contextual profile information and specific AI-related topics. Below is an overview of the survey structure: · Section I – Organisational Profile: 3 questions on company size, industry sector, and the respondent’s role in the company. · Section II – AI Familiarity & Adoption Motives: 2 questions assessing the respondent’s familiarity with AI applications in business and the primary motives for adopting AI solutions. · Section III – AI Adoption Barriers: 1 multi-select question identifying the main perceived obstacles to implementing AI in the organisation. · Section IV – AI and Employment: 2 questions on how AI is expected to affect jobs (e.g. job creation vs. replacement) and what measures the company is taking (or would take) to train or prepare employees for AI integration. · Section V – AI in Business Future: 2 questions examining anticipated high-impact AI application areas for business value and opinions on AI’s potential to drive economic growth. · Section VI – Regulation & Ethics: 2 questions regarding the need for AI-specific regulation to mitigate risks and the primary ethical concerns related to AI use in business. · Section VII – Global Competitiveness: 2 questions gauging perceptions of Romania’s competitiveness in AI adoption relative to other European countries and how Romania should position itself in the global AI market. · Section VIII – Emerging Topics & Strategy: 5 questions covering strategic outlooks, including AI’s environmental impact, the role of government incentives in AI adoption, expected changes to the company’s business model due to AI, anticipated new AI innovations in the industry (open-ended), and approaches to measuring return on investment (ROI) for AI projects. · Section IX – Final Considerations: 1 open-ended question inviting any final comments or suggestions from respondents regarding AI’s impact on their business or the Romanian market. This survey structure was purposefully designed to capture both quantitative inputs (through closed-ended questions in single-choice or multiple-choice format) and qualitative insights (through two open-ended questions in Sections VIII and IX). The closed questions use categorical response scales tailored to each topic (including ordinal scales for level-of-familiarity and agreement where appropriate), while multiple-selection questions allow respondents to choose all applicable options (e.g. selecting all relevant barriers or training measures). The instrument underwent content validation to ensure clarity in Romanian business context, and it was estimated to require ~10 minutes to complete, balancing depth with respondent effort[6]. The overall purpose of the survey instrument is to obtain a comprehensive yet focused snapshot of how AI is perceived, adopted, and managed in Romanian businesses, providing a basis for descriptive analysis without delving into individual companies’ proprietary data. DATASET DESCRIPTION The responses collected from this survey have been compiled into an anonymised dataset (in CSV format) for research and archival use. The dataset captures each respondent’s answers across all questions, with one row per completed response. Table 1 summarises the key characteristics of the dataset and survey content: DATASET ATTRIBUTE DESCRIPTION Total responses 212 completed responses (unique organisations/respondents) Survey period 2024–2025 (data collected in this interval) Total questions 20 questions, structured into 9 thematic sections (as detailed above) Question formats 13 single-choice categorical questions; 6 multiple-selection questions; 1 open-ended question Total variables (columns) 139 columns in the CSV dataset, including all survey question fields and sub-items for multi-select options, plus technical metadata Thematic coverage Organisational demographics; AI familiarity and usage; drivers and barriers to AI adoption; workforce impact; strategic outlook on AI (business value, competitiveness, innovation); regulatory and ethical perspectives; and open-ended feedback Each row in the dataset corresponds to an individual survey submission, identified by a unique response ID. The 139 columns encompass: system fields (e.g. response ID and timestamps); one column for each single-choice question; multiple binary or categorical columns for each multi-select question (indicating which options were chosen, with accompanying free-text columns for any specified “Other” inputs); and text fields for the open-ended questions. All variable names are coded for brevity in English or alphanumeric identifiers, while the survey questions were administered in Romanian. A codebook is provided along with the data, mapping each column to its survey question and response options. In terms of data types: most variables are nominal or ordinal categorical data (recorded as numeric codes or text labels for selected categories), and a smaller number are free text strings (for the open-ended responses). There are no continuous numerical measurements in this survey; rather, the emphasis is on categorical perception and decision data. The structure enables researchers to perform frequency analysis, cross-tabulations (e.g. comparing responses by company size or industry), and qualitative content analysis on open-ended answers, without needing any personal identifiers. Notably, the dataset reflects a diverse range of company profiles and AI engagement levels due to the broad outreach of the survey. While this documentation does not interpret the results, basic counts indicate that the 212 respondents cover various industries and organizational sizes, making the dataset a valuable cross-section of Romanian business perspectives on AI. The multi-topic coverage, from cybersecurity concerns to ROI evaluation, allows for multifaceted descriptive analysis. For instance, researchers can identify how many companies cite high implementation cost as a barrier versus those that cite lack of expertise (from the multiple-choice barriers question), or enumerate the kinds of AI applications that respondents consider most valuable to their business (from the multi-select question on high-potential AI applications). DATA ANONYMISATION AND ACCESS This dataset has been thoroughly anonymised to protect respondent privacy and confidentiality. No personally identifiable information (PII) was collected in the survey instrument – questions focused on organisational characteristics and attitudes rather than individual identities, and the survey did not ask for names, contact details, or company names. Furthermore, any indirect identifiers (such as IP addresses or precise timestamps) have been removed or generalized in the public dataset. All responses are stored under randomised ID codes. As a result, individual respondents or organisations cannot be identified from the data, in compliance with research ethics and data protection standards. The full survey questionnaire remains available online for consultation at the original LimeSurvey portal (in Romanian) at https://www.info-world.ro/so/ index.php/444639?lang=ro . This ensures transparency of the instrument and allows interested parties to view the exact phrasing of questions and options as presented to participants. The online availability of the survey also enables ongoing data collection; however, the dataset described here corresponds to a fixed export of completed responses as of the survey’s initial analysis period. Researchers and practitioners are welcome to use this dataset under the terms of its open data license, and to cite this documentation report in any derived publications. By providing a detailed description of the survey instrument and dataset, we aim to facilitate the reusability of the data and encourage further scholarly examination of AI’s business impact in emerging markets such as Romania. REFERENCES Autoritatea pentru Digitalizarea României (ADR) & Ministerul Cercetării, Inovării și Digitalizării (MCID). (2024). Strategia națională în domeniul inteligenței artificiale 2024–2027 (National AI Strategy 2024–2027). Government of Romania. Eurostat. (2024, May 29). 8% of EU enterprises used AI technologies in 2023 [News release]. Eurostat Statistics Explained. [1][2] EY Romania. (2025, May 29). EY Entrepreneurship Barometer Study: 71% of companies adopt Artificial Intelligence, while 75% fear economic risks in Romania [Press release]. EY.com. [3][4] [1] [2] 8% of EU enterprises used AI technologies in 2023 - News articles - Eurostat https://ec.europa.eu/eurostat/web/products-eurostat-news/w/ddn-20240529-2 [3] [4] EY Entrepreneurship Barometer Study: 71% of companies adopt Artificial Intelligence, while 75% fear economic risks in Romania | EY - Global https://www.ey.com/en_ro/newsroom/2025/050/ey-entrepreneurship-barometer-study- [5] [6] INFLUENȚA INTELIGENȚEI ARTIFICIALE ÎN MEDIUL DE AFACERI DIN ROMÂNIA https://surveys.tmedia.ro/index.php/854239?lang=en人工智能对罗马尼亚企业的影响（调研数据集说明） ## 背景与研究动因 快速演进的人工智能（Artificial Intelligence, AI）技术在全球范围内引发了各界对其企业影响的广泛关注。但最新统计数据显示，罗马尼亚的人工智能采纳仍处于起步阶段。2023年，仅约1.5%的罗马尼亚企业（员工≥10人）部署了AI解决方案——这是欧盟最低比例，而欧盟整体约为8%（欧盟统计局[Eurostat]，2024）[1][2]。这一差距凸显了理解本地AI采纳驱动因素与阻碍因素的重要性。罗马尼亚政策制定者也强调需采取战略行动：该国已出台国家人工智能战略，以最大化AI整合的效益并契合欧洲发展趋势（罗马尼亚数字化管理局[Autoritatea pentru Digitalizarea României, ADR]与研究、创新与数字化部[Ministerul Cercetării, Inovării și Digitalizării, MCID]，2024）。私营部门则呈现出热情与谨慎并存的态势：例如，近期一项创业调研显示，71%的罗马尼亚企业在过去一年提升了AI使用量（安永[EY]罗马尼亚，2025）[3]，但37%的企业表示在员工AI工具培训方面存在挑战[4]。在此背景下，本调研旨在系统性评估人工智能对罗马尼亚企业的影响，通过捕捉企业利益相关者的AI使用体验、准备度与担忧，填补相关实证研究空白。本调研覆盖多类主题——从劳动力影响、网络安全挑战到金融服务变革、伦理考量以及AI相关政府政策——全面反映了人工智能对企业影响的多面性[5]。本项目的核心动因在于收集可靠数据，以为罗马尼亚商业环境下的AI采纳提供学术分析与实践策略参考。 ## 调研结构与工具 本次名为"Impactul Inteligenței Artificiale asupra Afacerilor din România"（即"罗马尼亚企业人工智能影响"）的调研是采用罗马尼亚语开发的结构化问卷，依托LimeSurvey在线平台搭建，为自填式网络调研，目标受众为熟悉所在组织AI使用情况的人员（如高管、经理、IT专员、创业者）。调研工具包含20个问题，划分为9个主题模块，兼顾组织背景信息与特定AI相关议题的覆盖。以下为调研结构概览： · 第一部分——组织概况：3个问题，涉及企业规模、行业领域以及受访者在企业中的角色。 · 第二部分——AI熟悉度与采纳动机：2个问题，评估受访者对商业AI应用的熟悉程度，以及采纳AI解决方案的主要动机。 · 第三部分——AI采纳阻碍因素：1个多选问题，用于识别组织内实施AI的主要感知障碍。 · 第四部分——AI与就业：2个问题，探讨AI对就业的预期影响（如岗位创造与岗位替代），以及企业当前或计划采取的员工AI整合培训与准备措施。 · 第五部分——未来商业AI应用：2个问题，调研预期可创造高商业价值的AI应用领域，以及对AI驱动经济增长潜力的看法。 · 第六部分——监管与伦理：2个问题，涉及针对AI的专项监管需求（以缓解相关风险）以及企业使用AI时的主要伦理关切。 · 第七部分——全球竞争力：2个问题，衡量受访者对罗马尼亚在AI采纳方面相较于其他欧洲国家的竞争力认知，以及罗马尼亚应如何在全球AI市场定位自身。 · 第八部分——新兴议题与战略：5个问题，涵盖战略展望，包括AI的环境影响、政府激励对AI采纳的作用、AI驱动的企业商业模式预期变革、行业内预期的新AI创新（开放式问题），以及AI项目投资回报率（Return on Investment, ROI）的衡量方法。 · 第九部分——最终意见：1个开放式问题，邀请受访者就AI对其业务或罗马尼亚市场的影响发表最终评论或建议。 本调研结构旨在同时收集定量输入（通过封闭式单选或多选问题）与定性洞察（通过第八、第九部分的2个开放式问题）。封闭式问题采用适配各主题的分类响应量表（包括针对熟悉度与同意程度的有序量表），多选问题允许受访者选择所有适用选项（如选择所有相关障碍或培训措施）。该工具已通过内容效度验证，以确保其在罗马尼亚商业语境中的清晰度，且预计完成时长约为10分钟，在调研深度与受访者负担之间取得平衡[6]。本调研工具的整体目标是全面且聚焦地展现罗马尼亚企业对AI的认知、采纳与管理现状，为描述性分析提供基础，且无需涉及单个企业的专有数据。 ## 数据集说明 本次调研收集的响应已整理为匿名化数据集（格式为CSV），供研究与存档使用。数据集收录了每位受访者所有问题的答案，每份完整响应对应一行数据。下表总结了数据集与调研内容的关键特征： | 数据集属性 | 说明 | | --- | --- | | 总响应数 | 212份完整响应（对应唯一的组织/受访者） | | 调研周期 | 2024–2025（此区间内收集的数据） | | 总问题数 | 20个问题，划分为9个主题模块（详见上文） | | 问题格式 | 13个单选分类问题；6个多选问题；1个开放式问题 | | 总变量（列数） | CSV数据集共139列，包含所有调研问题字段、多选问题的子项字段，以及技术元数据 | | 主题覆盖范围 | 组织人口统计学特征；AI熟悉度与使用情况；AI采纳的驱动因素与阻碍因素；劳动力影响；AI战略展望（商业价值、竞争力、创新）；监管与伦理视角；以及开放式反馈 | 数据集中的每一行对应一份独立的调研提交，以唯一的响应ID标识。139列涵盖：系统字段（如响应ID与时间戳）；每个单选问题对应一列；每个多选问题对应多个二进制或分类列，用于标记所选选项，并附带“其他”选项的自由文本列；以及开放式问题的文本字段。所有变量名称均采用英文或字母数字标识符以实现简洁编码，而调研问题均以罗马尼亚语呈现。随数据集附带一份编码手册，用于将每一列映射至对应的调研问题与响应选项。就数据类型而言：多数变量为名义或有序分类数据（以数值代码或所选类别的文本标签记录），少量变量为自由文本字符串（对应开放式响应）。本调研未包含连续数值型测量指标，核心重点为分类感知与决策数据。该结构支持研究人员开展频次分析、交叉制表（如按企业规模或行业对比响应结果）以及开放式答案的定性内容分析，且无需任何个人识别信息。 值得注意的是，得益于本次调研的广泛覆盖，数据集反映了多样化的企业概况与AI参与程度。虽然本说明未对调研结果进行解读，但基础统计数据显示，212名受访者来自不同行业与组织规模，使得该数据集成为罗马尼亚企业AI认知视角的宝贵横截面样本。本次调研覆盖从网络安全关切到ROI评估的多类主题，支持多维度描述性分析。例如，研究人员可统计将高实施成本列为阻碍因素的企业数量，对比将缺乏专业技能列为阻碍因素的企业数量（来自多选阻碍因素问题），或枚举受访者认为对其业务最具价值的AI应用类型（来自高潜力AI应用的多选问题）。 ## 数据匿名化与获取 本数据集已完成全面匿名化处理，以保护受访者的隐私与机密性。调研工具未收集任何个人可识别信息（Personally Identifiable Information, PII）——问题聚焦于组织特征与态度，而非个人身份，且未要求受访者提供姓名、联系方式或企业名称。此外，公开数据集中的任何间接标识符（如IP地址或精确时间戳）均已被移除或泛化处理。所有响应均以随机生成的ID代码存储。因此，无法从数据中识别出单个受访者或组织，符合研究伦理与数据保护标准。 完整的调研问卷仍可通过原LimeSurvey门户（罗马尼亚语版本）在线查阅，网址为https://www.info-world.ro/so/index.php/444639?lang=ro。此举确保了调研工具的透明度，允许相关方查看向参与者呈现的问题与选项的确切措辞。调研的在线可用性也支持持续的数据收集；但本说明所描述的数据集对应至初始分析周期时导出的完整响应固定版本。研究人员与从业者可根据开放数据许可条款使用本数据集，并在衍生出版物中引用本说明文档。通过详细阐述调研工具与数据集，本说明旨在促进数据的可复用性，并鼓励对罗马尼亚等新兴市场中AI的企业影响开展进一步学术研究。 ## 参考文献 [1] 罗马尼亚数字化管理局（Autoritatea pentru Digitalizarea României, ADR）与研究、创新与数字化部（Ministerul Cercetării, Inovării și Digitalizării, MCID）. (2024). 2024–2027年国家人工智能战略（National AI Strategy 2024–2027）. 罗马尼亚政府. [2] 欧盟统计局（Eurostat）. (2024年5月29日). 2023年欧盟8%企业使用人工智能技术[新闻稿]. 欧盟统计局统计解释. [3] 安永（EY）罗马尼亚. (2025年5月29日). 安永创业晴雨表研究：71%的罗马尼亚企业采用人工智能，75%担忧经济风险[新闻稿]. EY官网. [4][5][6] 罗马尼亚企业人工智能影响调研 [INFLUENȚA INTELIGENȚEI ARTIFICIALE ÎN MEDIUL DE AFACERI DIN ROMÂNIA] https://surveys.tmedia.ro/index.php/854239?lang=en