Self-Generated Text Recognition: Quality Heuristics, Cross-Task Transfer, and Downstream Bias in LLM Evaluation
Evaluating 13-21 models across six presentation operationalizations and four task-domain operationalizations suggests that, despite confounds, some models possess practical SGTR capabilities, and that SGTR should be monitored and considered in the design of safety-critical AI applications.
J. St-Amand, Callum Canavan, S. Imran et al.
· 0 citations