The paper proposes an EU GPAI Evaluation Standards Taskforce to develop adaptive standards for AI evaluations, based on four desiderata: internal validity, external validity, reproducibility, and portability.
We need a science of evals
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CY 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
GPAI Evaluations Standards Taskforce: Towards Effective AI Governance
The paper proposes an EU GPAI Evaluation Standards Taskforce to develop adaptive standards for AI evaluations, based on four desiderata: internal validity, external validity, reproducibility, and portability.