AI Fails the History Test: Can LLMs Authentically Mimic Past Voices?
LLMs struggle to authentically reproduce historical language, even with fine-tuning. New research explores prompting, fine-tuning, and dedicated historical models, revealing limitations and trade-offs.