#model fairness

New research shows that adding context to ambiguous user queries significantly improves AI model evaluation, revealing biases and even reversing model rankings.