Evaluating the Limitations of General-Purpose AI in Legal Question-Answering and Advocating for Open-Source, Domain-Specific Solutions

General-purpose AI models like ChatGPT exhibit significant limitations in legal question-answering tasks, including hallucinations, biases, and lack of diversity in responses. An open-source, domain-specific approach is needed to address these shortcomings and improve the accuracy, transparency, and narrative representation in legal AI systems.