The paper, called AI and synthetic data in market research, said that synthetic data has several advantages over real data, including enhanced privacy due to the lack of actual personal identifiable information in data sets.

Other advantages include that small segments and/or hard-to-reach groups can be extrapolated to boost sample sizes for analysis, and also that well-trained models show strong correlations with the real-world data sets that they are based on.

However, the paper argued that claims synthetic data will completely replace primary data collection are unlikely, partly because synthetic data, like most new technologies when they first appear, “is currently over-hyped”.

Ethical and regulatory concerns are also an issue, as is the likelihood that synthetic data will replicate biases in its training data and algorithm.

Other issues include the fact that synthetic data is retrospective and can never be current or show emerging trends, making it unsuitable for longitudinal research.

The fact that people are complex also makes replicating that nuance with synthetic data difficult with current technology.

The paper identifies four core challenges in the market research industry in the age of AI, and said generative AI and synthetic data could help to alleviate some of these challenges and provide new opportunities for the industry in areas such as enhancing data sets by adding unmeasured variables.

The full report can be read here: https://www.mrs.org.uk/pdf/MRS_SyntheticData+AI_Paper_Mar2025.pdf