TAISE Practice Questions - CSA Trusted AI Safety Expert (TAISE) Exam

The model generating outputs that are factually incorrect or entirely fabricated with apparent confidence A visual distortion artifact caused by adversarial image perturbations A mode of training where the model imagines counterfactual scenarios to improve robustness An overfit condition in which training loss is extremely low but validation loss is high