
AI-Generated Responses Raise Fresh Concerns Over Trust in Market Research Data
The growing use of artificial intelligence in online surveys is creating fresh challenges for market researchers seeking reliable consumer insights, as AI-generated responses become increasingly difficult to distinguish from genuine human feedback.
The concern featured prominently at the 2026 Quirk’s Event in New York, where Kurt Knapton of Realm Resources and Steve Snell of Rep Data examined how changes in online research recruitment, automated responses and fraudulent participation have complicated efforts to maintain data quality.
According to Brandspur Brand News, the discussion highlighted the need for research companies to strengthen respondent verification and make their data collection processes more transparent as artificial intelligence becomes more sophisticated.
Knapton, who has followed the development of online research sampling for more than 25 years, explained that the industry had moved from relatively straightforward recruitment systems to a more complicated network involving sample aggregators, exchanges, multiple panels and independent traffic sources.
In the early days, researchers could often establish respondents’ identities through recruitment channels linked to airlines, hotels and car rental companies. These organisations already held customer information that helped verify participants.
However, the expansion of online research created demand for larger and more diverse respondent pools. Aggregators made it easier to reach those audiences but also reduced researchers’ visibility into the original sources of some participants.
That shift has made verification more demanding. Researchers now need to examine respondents before, during and after surveys, rather than relying solely on the reputation of the platforms supplying participants.
Knapton identified device information, browser activity, typing patterns, mouse movements and copy-and-paste behaviour among the signals that can help detect suspicious participation. He cautioned that individual indicators may not establish fraud on their own, making it necessary to assess several behavioural patterns together.
Artificial intelligence presents another complication because automated systems can produce responses that appear coherent, detailed and more consistent than ordinary human answers.
However, Knapton argued that synthetic respondents and AI representations of individuals could still have legitimate applications in certain research settings. Their usefulness depends on the subject being investigated, particularly where a study requires personal experiences or knowledge that an AI model may not possess.
The researchers also pointed to two persistent threats to reliable survey results: participants who misrepresent themselves to qualify for studies and those who complete questionnaires without paying sufficient attention.
Rep Data’s Steve Snell stressed that researchers must establish whether participants genuinely possess the identity, location, experience or knowledge required for a particular study. Their willingness to engage meaningfully with the questions is equally important.
For companies commissioning market research, the implications extend beyond the collection of survey responses. Poor-quality data can undermine the consumer insights used to guide product development, marketing campaigns and business decisions.
Knapton advised research buyers to demand greater transparency from their data suppliers, including information about respondent recruitment channels, the origin of sample sources and the number of participants removed during quality checks.
He also encouraged clients to request percentage breakdowns showing how many respondents were excluded for different quality concerns, arguing that businesses should understand how their research data is collected and validated.
Artificial intelligence is already being applied across several areas of the research process, including fraud detection, survey programming, open-ended response coding and data analysis.
However, Knapton warned that the speed at which new threats emerge makes it difficult for research companies to rely on existing safeguards indefinitely.
He maintained that technological progress must be matched by stronger verification systems, responsible recruitment practices and closer scrutiny of the companies supplying research participants.
The discussion reflected a wider challenge for the market research industry: producing larger volumes of consumer data is no longer enough. Researchers must also demonstrate that the people, or systems, behind those responses are appropriate for the decisions the findings are expected to support.





