Large Language Models as Subpopulation Representative Models: A Review

Research Questions

1. Can LLMs be used as Subpopulation Representative Models (SRMs)?
2. What behavioral steering techniques and existing SRM applications are currently available for LLMs?
3. How should the SRM lifecycle—design, development, and operation—be structured?
4. What are the benefits and risks associated with this approach?

Results

• LLMs can serve as powerful tools for capturing public opinion and representing subpopulations.
• SRMs provide opportunities to overcome data scarcity when survey response rates are low.
• SRM applications are being explored in politics, sociology, and commercial domains.
• However, significant risks exist, including misinformation, bias, privacy violations, and potential misuse.
• Tasks range widely—from low-complexity classification to multi-turn, open-ended dialogue.

Findings

• Advantages:

  • LLMs enable low-cost, open-ended analysis and can generate human-like representatives of subpopulations.

• Historical Parallel:

  • SRMs echo early efforts such as the 1960s “People Machine,” signaling renewed interest in public opinion modeling.

• Application Areas:

• Growing use in:

  • Election forecasting
  • Consumer sentiment collection
  • Brand perception tracking
  • Broader public opinion research

• Risks:

  • Key concerns include misinformation generation, poor performance for marginalized groups, and the potential for social manipulation.

• Evaluation Framework:

• A five-criteria framework is proposed:

  • Fidelity
  • Necessity
  • Robustness
  • Sensitivity
  • Fairness

• LLM Models: 5

• Synthetic Data: 4

• Method: 5

• Speed: 3

• Ethics: 5

• Accuracy: 4

• Demographics: 5

If you would like to explore this research in more detail, click here to read the full paper.