🧪 AI Failure Lab

❌ Hallucination

Model confidently invents facts

❌ Reasoning failure

Incorrect logic despite having correct information

❌ Context confusion

Model misunderstands what matters in a long conversation

❌ Instruction conflict

Model follows the wrong instruction among competing instructions

❌ Overconfidence

Model sounds certain when it shouldn't

❌ False consensus

Model presents one interpretation as universally true

❌ Inconsistent answers

Same question produces conflicting answers

❌ Knowledge cutoff failure

Model lacks recent information

❌ Math/calculation errors

Model makes numerical mistakes

❌ Citation fabrication

Model invents sources or references