Self-driven AI model interaction (independent study) for consistency and hallucination analysis
Conducted independent study by testing public generative AI tools to analyze prompt structures and identify logical inconsistencies in outputs. Drafted comparative summaries to highlight hallucination and superficial reasoning patterns in generated responses. This constitutes LLM evaluation/red-teaming style work centered on assessing response quality and reliability. • Analyzed prompt structures and output logic for inconsistency. • Identified cases of data hallucination in automated responses. • Assessed reasoning depth and coherence in AI-generated text. • Produced comparative summaries to document failure modes.