This article is part of weekly updates on new developments in the use of AI methods and tools of surveys (households, individuals, farms…) and administrative data for official statistics

Coverage Period: 16–22 February 2026

Key words: AI, survey research, official statistics, machine learning, data quality, automation, household surveys

AI-Powered Voice Surveys Revolutionizing Data Collection

Voice AI surveys are emerging as a transformative technology for data collection, combining the personal touch of telephone interviews with the efficiency and scalability of artificial intelligence [1]. These systems leverage advanced natural language processing to conduct automated telephone interviews, offering significant advantages over traditional methods. A recent analysis highlights that voice AI can conduct thousands of interviews simultaneously, reducing fieldwork time from weeks to hours and offering up to 70% cost reduction compared to traditional phone surveys. The technology stack includes Automatic Speech Recognition (ASR), Natural Language Processing (NLP), Natural Language Generation (NLG), and Text-to-Speech (TTS) to create a natural conversational experience for respondents.

Future developments in this area are expected to include emotional intelligence to detect respondent sentiment, predictive sampling for real-time sample optimization, and hyper-personalization of conversation style.

Rapid AI Adoption in Research Data Workflows

A new report on the state of open data reveals a significant increase in the use of AI for research data workflows between 2024 and 2025 [2]. The use of AI for data processing jumped from 22% to 32%, while its use for metadata creation rose from 16% to 25%. This rapid adoption indicates that AI is becoming a standard part of the research data landscape, streamlining data preparation and supporting the principles of FAIR (Findable, Accessible, Interoperable, and Reusable) data.

Machine Learning for Household Survey Data Editing

A groundbreaking study published in the Journal of Official Statistics presents a machine learning approach to prioritize editing in household survey data [3]. The authors developed a score function using a Gradient Boosting Trees classifier to identify cases affected by severe errors and omissions, a method that outperforms traditional selective editing strategies. This is the first application of machine learning to prioritize editing in household finance survey data, offering significant potential to reduce the time and resources allocated to manual data editing.

New UN Handbook on Household Surveys

The United Nations Statistics Division has launched a new Handbook of Surveys on Households and Individuals to address the fragmented and outdated guidance on household surveys [4]. The handbook provides technical standards, best practices, and case studies on emerging methodologies for National Statistical Offices and other organizations conducting surveys. While primarily designed for traditional face-to-face surveys, it also addresses emerging methodologies, which is crucial for integrating new technologies like AI into survey operations.

AI Transforming the ECB’s Corporate Telephone Survey

The European Central Bank (ECB) is using AI to transform its Corporate Telephone Survey, demonstrating a practical implementation of AI across the entire survey workflow [5]. The ECB is using AI for automated transcription and summarization of interviews, interview scoring, and workflow automation. This has resulted in significant productivity gains, with an estimated 31 person-hours saved in the latest survey round. The ECB’s experience highlights that while human expertise remains indispensable, a human-machine synergy can significantly enhance the efficiency and analytical capabilities of survey operations.

Advanced AI Tools for Survey Analysis and Reporting

New AI-powered tools are automating survey analysis, reporting, and the analysis of open-ended responses [6]. Platforms like Sopact, Zonka Feedback, and Thematic offer capabilities such as automatic theme extraction, sentiment analysis, and real-time reporting. These tools can compress weeks of manual work into minutes, enabling researchers and statistical offices to derive insights from survey data more efficiently.

AI-Readiness for Official Statistics

The UN Statistics Division hosted a seminar on AI-readiness for Official Data and Statistics, emphasizing the need for National Statistical Offices (NSOs) to prepare their data for AI consumption [7]. The objective is to provide AI systems with correct, timely, and contextually relevant official data and metadata. This requires NSOs to take on a stewardship role over the AI ecosystem, including understanding AI models, establishing guardrails for the use of official data, and developing standards for testing AI-mediated results.

The Threat of AI Bots to Survey Research Quality

Alongside the opportunities, the increasing use of AI also presents significant challenges. A major concern is the infiltration of surveys by AI bots, which can threaten the quality and reliability of survey data [8]. Recent analyses have found that AI chatbots can account for up to 45% of responses in some social science surveys. This has led to calls for new bot-detection strategies and a re-evaluation of survey incentive structures to mitigate the risk of data contamination.

Conclusion

The developments this week highlight a rapid and transformative integration of AI into all stages of the survey research process. From data collection and processing to analysis and dissemination, AI is offering new opportunities for efficiency, scalability, and deeper insights. However, the increasing prevalence of AI also brings new challenges, particularly in ensuring data quality and authenticity. For researchers and statistical offices, the key to harnessing the benefits of AI will be to embrace these new technologies while also developing the necessary governance, standards, and methodological innovations to address the associated risks.

References

[1] Voice AI Survey: The Future of Market Research is Here

Blog Sonalyx

[2] Open data sharing: What a decade of researcher insights tells us

For Researchers

Springer Nature

[3] A Score Function to Prioritize Editing in Household Survey Data: A Machine Learning Approach

Journal of Official Statistics

[4] Please join me at this UNSC side event on Thursday (online) where a small part of the amazing team that have produced this new survey handbook will present an overview of what it contains, how it…

Peter Lynn

[5] Using AI to transform the ECB’s Corporate Telephone Survey

European Central Bank

[6] Survey Analysis: AI Methods, Automated Reporting & Tool

Sopact

[7] AI-readiness for Official Data and Statistics

UN Statistics Division

[8] How to deal with the survey-taking AI agents that threaten to upend social science

Nature

Contact: bakodramane@gmail.com