About AI Comparison

Goal

This tool helps people understand that different AI systems can have different biases and interpretations when analyzing the same data.

How it works

  1. Upload a graph or plot image
  2. The same image is sent to multiple AI models with identical prompts
  3. Each AI interprets the plot independently
  4. Sentiment analysis shows the tone of each interpretation

The prompt

Every AI receives the exact same prompt:

"Interpret this plot."

Methodology

  • All three models run in parallel with identical inputs
  • Temperature is set to 0 on all models for consistent, deterministic responses
  • Sentiment is scored using VADER — it measures the emotional tone of each AI's language, not the data itself

Currently supported

  • Claude (Anthropic)
  • ChatGPT (OpenAI)
  • Grok (xAI)

Source

View on GitHub