Tarea 5.1 - Analyzing the Behavioral Limits of a Large Language Model

Instructions:

1. Choose one LLM (e.g., GPT, Claude, Gemini, LLaMA).

2. Design 5 controlled prompts, each testing a different property:

• One reasoning task

• One hallucination test

• One paraphrase robustness test

• One bias-sensitive prompt

• One factual knowledge question

3. For each case:

• Provide the prompt

• Provide the model output

• Analyze the behavior

• Classify the limitation (epistemic, architectural, social)

Deliverable:

A technical report.