cs.AIFeb 25, 2026

A Framework for Assessing AI Agent Decisions and Outcomes in AutoML Pipelines

Authors: Gaoyuan Du, Amit Ahlawat, Xiaoyang Liu, Jing Wu

Organizations: Amazon Web Services · Amazon Security

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Agent-based AutoML systems rely on large language models to make complex, multi-stage decisions across data processing, model selection, and evaluation. However, existing evaluation practices remain outcome-centric, focusing primarily on final task performance. Through a review of prior work, we find that none of the surveyed agentic AutoML systems report structured, decision-level evaluation metrics intended for post-hoc assessment of intermediate decision quality. To address this limitation, we propose an Evaluation Agent (EA) that performs decision-centric assessment of AutoML agents without interfering with their execution. The EA is designed as an observer that evaluates intermediate decisions along four dimensions: decision validity, reasoning consistency, model quality risks beyond accuracy, and counterfactual decision impact. Across four proof-of-concept experiments, we demonstrate that the EA can (i) detect faulty decisions with an F1 score of 0.919, (ii) identify reasoning inconsistencies independent of final outcomes, and (iii) attribute downstream performance changes to agent decisions, revealing impacts ranging from -4.9% to +8.3% in final metrics. These results illustrate how decision-centric evaluation exposes failure modes that are invisible to outcome-only metrics. Our work reframes the evaluation of agentic AutoML systems from an outcome-based perspective to one that audits agent decisions, offering a foundation for reliable, interpretable, and governable autonomous ML systems.

Figures & tables

DatasetNFaultyPrecisionRecallF1
German Credit25150.9230.8000.857
Adult Income25151.0000.8670.929
Titanic25150.9330.9330.933
Diabetes25150.8750.9330.903
CA Housing25150.9381.0000.968
Overall125750.9320.9070.919
TABLE I: Decision Error Detection Performance (125 decisions, 75 injected faults)
CategoryTotalEARuleLLM
Valid1210115
Hallucinated Fact12903
Logical Contradiction121112
Numerical Hallucination12955
Action–Reasoning Mismatch12601
Overall Accuracy600.7500.2830.267
TABLE II: Reasoning Validation Performance by Category (60 test snippets)
DatasetDomainTask Perf.RobustnessFairnessCalibrationEfficiency
German CreditFinanceAcc: 79.5%Noise: 2.0% deg.DP: 1.6%ECE: 7.2%50K samples/s
(1,000 samples)(Classification)F1: 78.0%Missing: 2.0% deg.EO: 7.7% (sex)  
  AUC: 82.8%    
Adult IncomeCensusAcc: 88.1%Noise: 0.9% deg.DP: 18.9% (sex)ECE: 2.8%68K samples/s
(48,842 samples)(Classification)F1: 87.7%Missing: 4.3% deg.DP: 9.9% (race)  
  AUC: 93.1% EO: 7.9% (sex)  
TABLE III: Comprehensive Multi-Dimensional Quality Assessment Across Five Datasets
Decision StageAlternativesAvg ImpactImpact Range
Data Preprocessing150.7%-3.6% to +6.0%
Feature Engineering151.5%-4.9% to +6.6%
Model Selection152.7%-2.9% to +8.3%
By Dataset   
German Credit90.8%-3.6% to +5.4%
Adult Income90.5%-3.5% to +4.7%
TABLE IV: Counterfactual Decision Analysis Results (Summary)
ComponentTime (ms)% of Total
Decision Assessor0.120.1%
Reasoning Validator0.901.0%
Model Quality Assessor90.4898.5%
Counterfactual Analyzer0.340.4%
Total EA91.85100%
TABLE V: EA Computational Overhead by Component
Reference RuntimeDurationEA Overhead
AIDE (5-min budget)300 s0.031%
AutoGluon (5-min budget)300 s0.031%
AIDE (10-min budget)600 s0.015%
Typical HPO (1 hour)3,600 s0.003%
TABLE VI: EA Overhead Relative to AutoML Pipeline Runtime
More in this topic

AI Agent Evaluation

AI agent evaluation measures whether tool-using systems complete tasks reliably, safely, and efficiently across interactive settings.

See all 519 papers

Explore similar work

CardsList
May 21, 2026

Agentic CLEAR: Automating Multi-Level Evaluation of LLM Agents

Agentic systems are becoming more capable: agents define strategies, take actions, and interact with different environments. This autonomy poses serious challenges for overseeing and assessing agent behavior. Most current tools are limited, focusing on observability with basic evaluation capabilities or imposing static, hand-crafted error taxonomies that cannot adapt to new domains. To address this gap, we present Agentic CLEAR, an automatic, dynamic, and easy-to-use evaluation framework. It produces textual insights into the agent behavior on three levels of granularity: system, trace, and node. Agentic CLEAR operates above the observability layer, enabling seamless integration and featuring an intuitive UI that makes agent evaluation highly accessible. In our experiments on four benchmarks, seven agentic settings, and tens of thousands of LLM calls, we show that Agentic CLEAR produces high-quality, data-driven, insightful feedback. Our analysis shows strong alignment with human-annotated errors and the ability to predict task success rate.

Asaf Yehudai, Lilach Eden, Michal Shmueli-Scheuer

IIBM Research

LLM Agent EvaluationAI Agent BenchmarksAutomated EvaluationAutomated Agent Evaluation
AskPDF
May 2, 2026

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

Independent Researcher

AI Agent ReliabilityAI Agent EvaluationAI Agent MonitoringLLM Agent Reliability
AskPDF
Apr 29, 2026

Think it, Run it: Autonomous ML pipeline generation via self-healing multi-agent AI

The purpose of our paper is to develop a unified multi-agent architecture that automates end-to-end machine learning (ML) pipeline generation from datasets and natural-language (NL) goals, improving efficiency, robustness and explainability. A five-agent system is proposed to handle profiling, intent parsing, microservice recommendation, Directed Acyclic Graph (DAG) construction and execution. It integrates code-grounded Retrieval-Augmented Generation (RAG) for microservice understanding, an explainable hybrid recommender combining multiple criteria, a self-healing mechanism using Large Language Model (LLM)-based error interpretation and adaptive learning from execution history. The approach is evaluated on 150 ML tasks across diverse scenarios. The system achieves an 84.7% end-to-end pipeline success rate, outperforming baseline methods. It demonstrates improved robustness through self-healing and reduces workflow development time compared to manual construction. The study introduces a novel integration of code-grounded RAG, explainable recommendation, self-healing execution and adaptive learning within a single architecture, showing that tightly coupled intelligent components can outperform isolated solutions.

Adela Bara, Gabriela Dobrita, Simona-Vasilica Oprea

Bucharest University of Economic Studies, Department of Economic Informatics and Cybernetics, no.6 Piața Romană, Bucharest, 010374, Romania

Automated Machine LearningMulti-Agent LLMs
AskPDF