Walkthrough: Comparing Human and Machine Explanations
July 20, 2025
What happens when we compare GPT’s interpretation of a text with a scholar’s?
In this walkthrough, we use a sample paragraph from literature and:
- Run it through a transformer model with attention visualizations
- Collect human commentary from an expert annotator
- Align insights using intertextual markup
We find that models often miss historical nuance or rhetorical cues—but excel in surface-level structure. Human input, grounded in broader texts and traditions, helps contextualize AI behavior.
This is where InterXAI thrives.
🧠📖🤖