Science ❯ Computer Science ❯ Machine Learning

Model Evaluation

Performance Metrics Benchmarking Performance Benchmarking Performance Analysis Bias in AI Environmental Assessment Benchmarking Performance Dataset Creation Empirical Results Empirical Validation Contextual Limitations Cognitive Models Human-Likeness Assessment Energy Consumption Evaluation Metrics Empirical Research Proxy Models Energy Metrics Analytic Systems Alignment Testing Phase Transition Benchmarking Techniques Statistical Analysis Performance Assessment Cross-Entropy Loss Stochastic Processes Agent Performance Factuality in AI Reliability and Validity Research Findings Benchmarking AI Models Factuality and Truthfulness Explainability Data Annotation

MIT Index Finds AI Agents Proliferating With High Autonomy and Thin Safety Disclosures

An MIT index finds widespread autonomy paired with scarce agent‑specific safety evidence.

Reasoning-Enabled AI Systems Emit Up to 50 Times More CO₂ Than Concise Models