arXiv Machine Learning
Aug 5
Evaluation Blindness: How Silent Measurement Failures Corrupt AI Systems from Training to Deployment
arXiv:2608. 02786v1 Announce Type: new Abstract: AI systems can fail silently.
From Subjective Judgments to Auditable Standards:Protocol-Guided AI Auditing of Website Redundancy
The paper introduces CORA (Counterfactual, Observable Redundancy Audit), a protocol for auditing website redundancy by measuring repetition load, normal-use tax, and failure-domain recovery reserve. Each audit run records screenshots, stable element identities, and task traces, while a versioned vision‑language model generates annotations that are validated and released only if they meet calibrated criteria. Experiments on a transparent mechanistic testbed show that CORA’s factorized representation separates reserve from normal-use tax and predicts perturbed success more accurately than scalar-load baselines, but it withholds automated scores when instruments fail to meet release requirements, indicating that CORA is an auditable candidate procedure for the studied benchmark rather than a universal standard.
Related stories
arXiv AI
Jun 2
Early Diagnosis of Wasted Computation in Multi-Agent LLM Systems via Failure-Aware Observability
arXiv:2606. 01365v1 Announce Type: new Abstract: Tool-using multi-agent large language model (LLM) systems spend computation through model tokens, tool calls, retries, and code execution before producing an answer.
arXiv AI
Aug 26
Confidently Wrong, Silently So: Auditing Undetectable Failures of a Deployed On-Device Language Model
The paper audits a developer‑accessible on‑device language model, revealing that it can confidently produce incorrect answers while refusing benign prompts, a phenomenon termed task‑asymmetric miscalibration. The model’s confident outputs are surface‑indistinguishable, with classifiers based on user‑visible features failing to separate correct from wrong responses. The authors propose a model‑agnostic audit protocol, a surface‑indistinguishability test, and a black‑box consistency wrapper that improves reliability without requiring model access.
arXiv Machine Learning
Aug 26
PinSieve: Production Selective VLM Serving and a Governed Memory Flywheel for Enterprise Content-Quality Triage
PinSieve is a production system that selectively serves vision‑language models (VLMs) for enterprise content‑quality triage, operating only on the grey‑zone cases that lightweight models cannot resolve. The deployed VLM Serving Agent filters 2.05× more non‑actionable items, improves review productivity by 25.7%, cuts operating costs by 16.2%, and delivers signals the same day instead of the next. A governed memory flywheel with selective feedback, audit sampling, and a bounded proposal‑verifier loop further reduces false‑negative rates from 17.73% to 13.29% over six months, while a reasoning review agent audits teacher‑generated rationales for keep/repair/drop decisions. whyItMatters":"The system demonstrates how selective VLM serving and governed feedback loops can substantially improve efficiency, cost, and accuracy in enterprise AI content‑quality pipelines."
arXiv AI
3d ago
APIFlow-Bench: Measuring Whether Agents Survive Long, Dependent API Workflows
arXiv:2608.29128v1 Announce Type: new Abstract: Tool-using agents are commonly evaluated by a single bit: whether an end-to-end workflow completed. This metric fails to distinguish failures that matt...
arXiv AI
Jul 24
Routing Subspaces: Auditing Evaluation-to-Deployment Mismatch in Fine-Tuned Language Models
arXiv:2607. 20436v1 Announce Type: cross Abstract: Safety evaluations often assume that behavior observed during testing reflects behavior in ordinary use, but fine-tuning can break this assumption.
arXiv Machine Learning
Jun 9
Decoy-Calibrated Failure Audits for Language Models
arXiv:2606. 09046v1 Announce Type: new Abstract: Useful audits reveal not only how often a model fails, but also where its failures concentrate.
arXiv AI
Aug 5
Measurement Without Validity: The Compounding Reliability Problem in Agentic AI Evaluation
arXiv:2608. 00794v2 Announce Type: replace Abstract: Agentic AI evaluation pipelines produce benchmark scores that justify deployment decisions, safety certifications, and regulatory compliance claims.
arXiv AI
Aug 12
Auditing Automated Evaluation, Error Propagation, and Runtime Mitigation in Tool-Using Language Agents
arXiv:2604. 16706v2 Announce Type: replace Abstract: Automated evaluation of tool-using large language model (LLM) agents is widely assumed to be reliable, yet this assumption is rarely validated against human annotation.
arXiv Computation and Language
4d ago
Fidelity Is Not Enough: Dispatch-Level Instrumentation for Agentic Datasheet Extraction
The paper reports that a model can pass fidelity checks—verifying that extracted values match the source—without actually opening a datasheet, due to a hidden constraint that disables tool use. To address this, the authors log every tool call in an agentic benchmark and develop two instruments: a rule‑based failure‑attribution classifier and a silent‑failure detector that flags runs based solely on which tools were invoked. While the detector shows low false positives on clean extractions and recovers all planted faults, its recall against correct tool usage but incorrect answers remains unmeasured, and a partial causal chamber confirms only a subset of claims, highlighting limitations in physical verification.
arXiv AI
Aug 24
No Judgment Without a Reason: Counterfactual Receipts for Versioned AI Evaluators
The paper introduces a framework for evaluating AI systems that not only checks final labels but also tracks the reasoning behind them through three core sources—grounds, norms, and authority—forming an eight-cell counterfactual judgment cube. It defines minimal source replacement sets, called judgment receipts, to explain changes in verdicts and provides certification cost bounds for black-box evaluators. The authors present ReasonBench, a benchmark with 19,520 cases, and demonstrate that while high standard accuracy can mask robustness issues, receipt accuracy reveals significant gaps in reasoning consistency across different models.
arXiv Computer Vision
2d ago
Accuracy and Cost Claims Do Not Survive Re-Execution in Agentic VideoQA
arXiv:2607.01469v3 Announce Type: replace Abstract: Agentic Video Question Answering (VideoQA) systems produce answers through adaptive reasoning and tool-use trajectories, yet standard practice eval...