AI models give overly repetitive confidence scores, skewing reliability tests

Original: Evaluation Pitfalls and Sparsity Limitations in LLM-based Confidence Estimates for Classification

Source: arxiv.org ↗