RESEARCH · RESEARCH · #1111
arXiv:2609.30563v1 — Intuitive prompting boosts LLM agents' fidelity in simulating individual social-media reactions
This paper (arXiv:2609.30563v1) evaluated four language models predicting reactions of eight profiled Serbian participants to 68 social-media posts under five prompt conditions. Results show that providing attitudinal profile content greatly outperformed demographic backstories, and instructing models to respond 'intuitively' (fast, immediate responses) yielded the highest fidelity, reduced compression of individual differences, and generalized better to topics not covered in the profiling.
KEY POINTS
- This paper (arXiv:2609.30563v1) evaluated four language models predicting reactions of eight profiled Serbian participants to 68 social-media posts under five prompt conditions.
- Results show that providing attitudinal profile content greatly outperformed demographic backstories, and instructing models to respond 'intuitively' (fast, immediate responses) yielded the highest fidelity, reduced compression of individual differences, and generalized better to topics not covered in the profiling.
- Demonstrates that prompt style and attitudinal profile content materially affect LLM agent fidelity, with implications for platform policy testing and risks of manufactured public opinion.
WHY IT MATTERS
Demonstrates that prompt style and attitudinal profile content materially affect LLM agent fidelity, with implications for platform policy testing and risks of manufactured public opinion.