LLMPollSimulationTask

This task simulates polls and surveys with AI personas.

Cognotik | LLMPollSimulationTask
Home > Tasks > Social

LLMPollSimulationTask

Simulate complex polls and surveys using diverse AI personas. Model demographic patterns, detect instrument bias, and generate synthetic research data with high fidelity.

Category: Social Side-Effect Safe GPT-4 Preferred
⚙️ ExecutionConfig.json

{
  "questions": [
    {
      "id": "remote_work",
      "text": "How has remote work affected your productivity?",
      "type": "LIKERT_SCALE",
      "min": 1, "max": 5
    }
  ],
  "respondent_profiles": [
    {
      "description": "Urban Tech Professional",
      "demographics": {"age": "25-34", "location": "Urban"},
      "characteristics": ["Tech-savvy", "Early adopter"]
    }
  ],
  "respondents_per_profile": 20,
  "bias_detection": true
}
→
📊 Session UI (Statistics Tab)

Descriptive Statistics

remote_work: How has remote work...
Mean: 4.22 Median: 4.00 Std Dev: 0.65

Bias Detection

⚠️ remote_work: Possible central tendency bias detected (mean=4.22, midpoint=3.0)

Live Results Showcase

Explore actual artifacts generated by this task, including statistical reports and persona transcripts.

Configuration Parameters

Field Type Description
questions * List<SurveyQuestion> List of survey questions to ask respondents. Supports Multiple Choice, Likert, Open-Ended, etc.
respondent_profiles * List<RespondentProfile> Templates defining demographics and characteristics for simulated personas.
respondents_per_profile Int Number of simulated respondents per profile. Default: 10.
cross_tabulation Boolean Generate analysis across demographic dimensions. Default: true.
bias_detection Boolean Detect response patterns like central tendency or primacy effects. Default: true.
sentiment_analysis Boolean Perform LLM-based sentiment analysis on open-ended responses. Default: true.
temperature Double Controls variance in persona responses (0.0 to 1.0). Default: 0.7.