The role of AI tools in survey research

Artificial Intelligence (AI) advancements present both opportunities and challenges for research practices and workflows.

In survey research, for years the implications have been primarily confined to the use of analytic AI methods in data treatment, augmentation, and analysis, often done by quantitatively-oriented methodologists and practitioners. Since late 2022, the introduction of generative AI, especially those based on large language models (LLMs), has brought much broader interests in incorporating AI tools into survey research.

By now, there is a substantial and growing body of literature on various AI applications for surveys and an even larger amount of POVs, commentaries, claims, and anecdotes shared on social media and professional networks by individuals and businesses.

What do these all mean to survey workflow and quality? Can we understand and approach this plethora of ideas, evidence, and claims in a principled and systematic manner? And how do we, as human researchers, and practitioners, need to adapt and evolve to thrive in this new reality while ensuring the value and positive impact of our work?

In this one-hour seminar, Mario Callegaro (Callegaro Research Ltd.) and Yongwei Yang (Google DeepMind) will attempt to give a structured review of the landscape of AI applications for surveys. They will keep a balance between concrete examples and higher-order mental models.

Specifically, they will describe a human-complementary perspective on using AI for research and show how it can lead to ideas about applications. They organise such ideas along the survey research workflow and illustrate them with recent research from the survey community, with discussions of both pros and cons. Lastly, they will share recommendations on how survey researchers and practitioners may want to consider reskilling, upskilling, and positioning themselves in this changing environment.