cs.AIDate pending

Measurement Without Validity: The Compounding Reliability Problem in Agentic AI Evaluation

Authors: William Caban

Organizations: Red Hat, Inc., Raleigh, North Carolina, USA · Alma Mater Europaea University, Vienna, Austria

Abstract

Agentic AI evaluation pipelines produce benchmark scores that justify deployment decisions, safety certifications, and regulatory compliance claims. No formal framework has yet characterized how validity degrades across the stages of these pipelines. We present a three-layer compounding validity model, Vtotal≤V1×V2×V3V_{total} \leq V_1 \times V_2 \times V_3, that captures multiplicative degradation across task generation (V1V_1), human-simulator calibration (V2V_2), and automated judgment (V3V_3). Under empirically grounded estimates, a pipeline retaining 70% validity at each stage is at most 34% valid against the intended construct (range 0.17-0.54 across the empirical estimate bounds). We examine the model's predictions against a structured survey of 55 published agentic evaluation papers, finding that approximately 82% of papers in this purposive sample apply structurally mismatched, incomplete, or absent inter-rater reliability (IRR) metrics, a pattern consistent with systematic V3V_3 collapse. We further identify empirical evidence of V1V_1 failures (task validity flaws in 7 of 10 popular benchmarks) and V2V_2 miscalibration (up to 9 percentage points inter-simulator variance, with systematic demographic disparities for non-Standard American English speakers). We derive eight prescriptions grounded in psychometric science and domain-stratified reliability thresholds (ICC ≥\geq 0.70; α≥\alpha \geq 0.67/0.70/0.80 by consequence level) that practitioners and benchmark authors can apply immediately. The framework provides a tractable knowledge-based tool for diagnosing and correcting evaluation pipeline validity before deployment decisions are made.

Explore similar work

CardsList
Jul 31, 2026cs.AI

Beyond Component Testing: Validating Agentic AI Systems

Agentic AI systems act through multi-step trajectories that combine planning, tool use, memory, interaction, and adaptation. This behavior stretches validation practice beyond component testing and one-shot input--output evaluation, because acceptable system behavior now depends on how decisions unfold over time and under changing environmental conditions. This survey synthesizes 257 papers spanning agent evaluation, software assurance, cyber-physical systems, runtime monitoring, and regulatory guidance in order to characterize the validation problem for agentic systems. The review is organized around a five-dimension taxonomy covering behavioral, safety, temporal, regulatory, and multi-agent concerns, and uses that taxonomy to map current approaches and expose recurrent coverage gaps. The analysis shows that behavioral evaluation is comparatively mature, while temporal validity, runtime evidence maintenance, regulatory legibility, and open-ended multi-agent systems assurance remain under-developed. Three cross-domain case studies (medical care, industrial operations, smart-mobility systems) provide operational illustrations of how the five taxonomy dimensions recur in safety-critical settings, grounded in the failure patterns documented in the reviewed literature. The paper concludes with a lifecycle-oriented research agenda centered on bounded-autonomy specifications, adversarial trajectory generation, runtime monitoring, and audit-ready evidence structures. The central claim is that trustworthy deployment of agentic AI depends on validating trajectories in context rather than assessing isolated components alone.
Fabio Orazio Mirto, Luca D'Agati, Giuseppe Tricomi +4
Production Agentic SystemsArtificial Intelligence RiskValidityCyber-Physical SystemRuntime
AskPDF
May 2, 2026cs.AI

Evaluating Agentic AI in the Wild: Failure Modes, Drift Patterns, and a Production Evaluation Framework

Existing evaluation frameworks for large language models -- including HELM, MT-Bench, AgentBench, and BIG-bench -- are designed for controlled, single-session, lab-scale settings. They do not address the evaluation challenges that emerge when agentic AI systems operate continuously in production: compounding decision errors, tool failure cascades, non-deterministic output drift, and the absence of ground truth for long-horizon tasks. This paper makes three contributions. First, we present a taxonomy of seven failure modes unique to production agentic systems, each grounded in observations from systems operating at billion-event scale. Second, we demonstrate empirically where standard metrics -- ROUGE, BERTScore, accuracy/AUC, and the agentic benchmarks above -- fail to detect each failure mode. Third, we propose PAEF (Production Agentic Evaluation Framework), a five-dimension evaluation framework with an open-source reference implementation, designed for continuous evaluation on production traffic rather than episodic benchmark runs. Our analysis shows that standard metrics fail to detect four of the seven failure modes entirely and detect three others only after a lag of multiple evaluation cycles.
Mukund Pandey
Agentic EvaluationsFailure Modes
AskPDF
Jul 24, 2026cs.AI

Do Agent Benchmarks Measure Capability? Protocol Validity in the Age of Agentic AI

Agent benchmarks increasingly evaluate repository editing, web research, terminal use, and long-horizon interaction. Their scores support capability claims only when the evaluation protocol keeps the intended capability necessary for success. Recent reward-hacking benchmarks and system reports show that agents can instead recover public solutions, read evaluation artifacts, infer generator structure, manipulate feedback, or benefit from invalid scoring paths; existing responses do not provide a common procedure for attributing these shortcuts and quantifying their effect across benchmarks. We formulate protocol validity and introduce HackDetect, a post-hoc audit that identifies an exposure, determines how the agent used it, and assesses whether the resulting score is misleading. We quantify score inflation with the Mislead gap, defined as the exploit score minus the intended score. We audit 2,385 traces across 15 agent benchmarks and find evidence of exposures and reward hacking in 67.0% of Frontier Science traces and 66.7% of AutoLab tasks. Across paired comparisons, we measure score inflation of 0.45-1.00, showing that benchmark reports should provide evidence that scores reflect the intended capability.
Jiaqi Shao, Hanck Chen, Wei Zhang +2
Agentic BenchmarksExploitationAlgorithm AuditingValidity
AskPDF