Acceptance Cards is a new four-diagnostic standard for safe fine-tuning defense claims that requires statistical reliability, fresh semantic generalization, mechanism alignment, and cross-task transfer; under this protocol SafeLoRA fails the full-card pass on Gemma-2-2B-it.
hub Mixed citations
Model Cards for Model Reporting
Mixed citation behavior. Most common role is background (60%).
70
Pith papers citing it
57
external citations · external index
Background
60% of classified citations
hub tools
citation-role summary
background 14
extension 1
citation-polarity summary
representative citing papers
The khipu problem frames a governance failure in distributed AI where interpretive continuity is lost even when traces remain, requiring infrastructure to preserve reading practices rather than only data retention.
Face-Feature Tuning is a label-free logit remapping method that reduces FPR/TPR gaps across groups in deepfake detection while preserving overall accuracy.
Fair regression with demographic parity penalty is recast as optimal transport, yielding optimal maps under Wasserstein-2 and total variation penalties that work in both aware and unaware regimes.
Empirical study finds strong heterogeneity in LLM process alignment across models and organizations; process alignment predicts output accuracy in legal decisions but is low and resistant in credit decisions where higher alignment may not be desirable.
Causal state binding is introduced as a framework that predicts action control in language agents, validated across large benchmarks and SWE-bench Lite where adding the measure raised issue-to-file hit@3 AUC from 0.873 to 0.935.
LLMs default to responses more similar to opinions from the USA and some European and South American countries; prompting for a country shifts alignment but can introduce stereotypes, while translation does not reliably match language speakers.
The paper defines five AI system categories for public administration and reports that 55% of 91 recent papers leave the system type underspecified while 31% study one type but motivate with another.
Introduces decision-alignment to evaluate uncertainty metrics against downstream decision utilities and proposes prior-weighted proper scoring rules that align better in benchmarks and case studies.
Simulations show standard CI methods underperform for classifier metrics in small and nested datasets, while Agresti-Coull, Wilson, Clopper-Pearson, and a new pseudo-count regularized bootstrap perform better, with specific adjustments needed for nested structures.
Empirical audit of LAION-2B-en and LAION-2B-multi finds overrepresentation of young adults, White people, and males plus stereotypical emotion associations across two attribute classifiers.
Introduces a gradient-based multilingual audit framework for LLM moral decisions in robot assistance scenarios and reports persistent culturally asymmetric gradient tracking failures not fixed by prompting.
Prompts for public-sector LLMs encode value-laden decisions and should be governed through community-maintained Prompt Commons repositories with provenance, licensing, and moderation.
Facial recognition enacts computational epistemicide by progressively reducing faces to standardized numerical vectors, rendering reformist ethical AI insufficient and requiring abolition of vectorized identity as a basis for rights.
CS researchers show pragmatic skepticism toward LLM leaderboards, using them despite distrust while preferring peer networks, arena leaderboards, and cost transparency as key missing feature.
StructuredSemanticSearch uses table discovery operators and orientation-aware integration on model-card tables to improve evidence coverage and diversity in model recommendation queries over a semantic baseline.
AI deployment in high-stakes areas requires domain-scoped calibrated verification with monitoring and revocation, using a proposed six-component Verification Coverage standard instead of mechanistic interpretability.
Agent benchmarks can report evidence-supported score bounds instead of single misleading success rates by adding a layer that checks required artifacts for outcome verification.
CIVeX maps agent tool calls to structural causal queries, checks identifiability, and issues auditable verdicts to prevent false executions while preserving utility on confounded benchmarks.
A conceptual framework is introduced that links activist needs to decentralized social network features and is applied to compare Mastodon and Bluesky plus example communities.
The paper names 'agentic literacy debt' as the structural deficit from deploying autonomous AI agents without literacy infrastructure for users to govern them.
DAISY is a structured form tool that generates more complete AI disclosure statements for research papers without reducing author comfort levels.
Introduces a parameter-driven framework for data attribution in LLMs that enables negotiation among creators, users, and intermediaries to meet stakeholder goals within the data economy.
The paper proposes the IARC-TS protocol that combines drift monitoring, uncertainty quantification, and stress tests to generate reproducible robustness evidence for industrial time series models mapped to EU AI Act obligations.
citing papers explorer
-
Acceptance Cards:A Four-Diagnostic Standard for Safe Fine-Tuning Defense Claims
Acceptance Cards is a new four-diagnostic standard for safe fine-tuning defense claims that requires statistical reliability, fresh semantic generalization, mechanism alignment, and cross-task transfer; under this protocol SafeLoRA fails the full-card pass on Gemma-2-2B-it.
-
The Khipu Problem: Institutional Legibility Under Distributed Cognition
The khipu problem frames a governance failure in distributed AI where interpretive continuity is lost even when traces remain, requiring infrastructure to preserve reading practices rather than only data retention.
-
Toward Calibrated, Fair, and accurate Deepfake Detection
Face-Feature Tuning is a label-free logit remapping method that reduces FPR/TPR gaps across groups in deepfake detection while preserving overall accuracy.
-
Geometry of Relaxed Fair Regression: A Unified Framework for Aware and Unaware Settings
Fair regression with demographic parity penalty is recast as optimal transport, yielding optimal maps under Wasserstein-2 and total variation penalties that work in both aware and unaware regimes.
-
Whose Alignment? Comparing LLM Process Alignment Across Diverse Organizational Decision Contexts
Empirical study finds strong heterogeneity in LLM process alignment across models and organizations; process alignment predicts output accuracy in legal decisions but is low and resistant in credit decisions where higher alignment may not be desirable.
-
Causal state binding predicts action control in language agents
Causal state binding is introduced as a framework that predicts action control in language agents, validated across large benchmarks and SWE-bench Lite where adding the measure raised issue-to-file hit@3 AUC from 0.873 to 0.935.
-
Towards Measuring the Representation of Subjective Global Opinions in Language Models
LLMs default to responses more similar to opinions from the USA and some European and South American countries; prompting for a country shifts alignment but can introduce stereotypes, while translation does not reliably match language speakers.
-
A Technical Typology of AI Systems in Public Administration
The paper defines five AI system categories for public administration and reports that 55% of 91 recent papers leave the system type underspecified while 31% study one type but motivate with another.
-
Decision-Aligned Evaluation of Uncertainty Quantification
Introduces decision-alignment to evaluate uncertainty metrics against downstream decision utilities and proposes prior-weighted proper scoring rules that align better in benchmarks and case studies.
-
Estimating Uncertainty in Classifier Performance with Applications to Large Language Models and Nested Data
Simulations show standard CI methods underperform for classifier metrics in small and nested datasets, while Agresti-Coull, Wilson, Clopper-Pearson, and a new pseudo-count regularized bootstrap perform better, with specific adjustments needed for nested structures.
-
Unmasking LAION-5B: Age, Gender, Race, and Emotion Biases in Large-Scale Image Datasets
Empirical audit of LAION-2B-en and LAION-2B-multi finds overrepresentation of young adults, White people, and males plus stereotypical emotion associations across two attribute classifiers.
-
Auditing LLM-Governed Social Robots with Culture-Specific Moral Gradients
Introduces a gradient-based multilingual audit framework for LLM moral decisions in robot assistance scenarios and reports persistent culturally asymmetric gradient tracking failures not fixed by prompting.
-
Prompts for Public-Sector LLMs Should Be Governed as Commons
Prompts for public-sector LLMs encode value-laden decisions and should be governed through community-maintained Prompt Commons repositories with provenance, licensing, and moderation.
-
Frankenstein in the Pipeline: Computational Epistemicide in Facial Recognition
Facial recognition enacts computational epistemicide by progressively reducing faces to standardized numerical vectors, rendering reformist ethical AI insufficient and requiring abolition of vectorized identity as a basis for rights.
-
The Trust Paradox: How CS Researchers Engage LLM Leaderboards
CS researchers show pragmatic skepticism toward LLM leaderboards, using them despite distrust while preferring peer networks, arena leaderboards, and cost transparency as key missing feature.
-
Diversed Model Discovery via Structured Table Discovery
StructuredSemanticSearch uses table discovery operators and orientation-aware integration on model-card tables to improve evidence coverage and diversity in model recommendation queries over a semantic baseline.
-
The Open-Box Fallacy: Why AI Deployment Needs a Calibrated Verification Regime
AI deployment in high-stakes areas requires domain-scoped calibrated verification with monitoring and revocation, using a proposed six-component Verification Coverage standard instead of mechanistic interpretability.
-
Can Agent Benchmarks Support Their Scores? Evidence-Supported Bounds for Interactive-Agent Evaluation
Agent benchmarks can report evidence-supported score bounds instead of single misleading success rates by adding a layer that checks required artifacts for outcome verification.
-
CIVeX: Causal Intervention Verification for Language Agents
CIVeX maps agent tool calls to structural causal queries, checks identifiability, and issues auditable verdicts to prevent false executions while preserving utility on confounded benchmarks.
-
The Activist's Guide to the Decentralized Social Universe: A Framework for Exploring How Decentralized Social Networks Can Support Collective Action
A conceptual framework is introduced that links activist needs to decentralized social network features and is applied to compare Mastodon and Bluesky plus example communities.
-
Agentic Literacy Debt: A Structural Problem the AI Literacy Field Has Not Yet Named
The paper names 'agentic literacy debt' as the structural deficit from deploying autonomous AI agents without literacy infrastructure for users to govern them.
-
AI Disclosure with DAISY
DAISY is a structured form tool that generates more complete AI disclosure statements for research papers without reducing author comfort levels.
-
A Human-Centric Framework for Data Attribution in Large Language Models
Introduces a parameter-driven framework for data attribution in LLMs that enables negotiation among creators, users, and intermediaries to meet stakeholder goals within the data economy.
-
Industrial AI Robustness Card for Time Series Models
The paper proposes the IARC-TS protocol that combines drift monitoring, uncertainty quantification, and stress tests to generate reproducible robustness evidence for industrial time series models mapped to EU AI Act obligations.
-
Inference Scaling Laws: An Empirical Analysis of Compute-Optimal Inference for Problem-Solving with Language Models
Empirical analysis shows scaling inference compute via strategies like tree search can be more efficient than scaling model parameters, with 7B models plus novel search outperforming 34B models.
-
BLOOM: A 176B-Parameter Open-Access Multilingual Language Model
BLOOM is a 176B-parameter open-access multilingual language model trained on the ROOTS corpus that achieves competitive performance on benchmarks, with improved results after multitask prompted finetuning.
-
PaLM: Scaling Language Modeling with Pathways
PaLM 540B demonstrates continued scaling benefits by setting new few-shot SOTA results on hundreds of benchmarks and outperforming humans on BIG-bench.
-
CTRL: A Conditional Transformer Language Model for Controllable Generation
CTRL is a large conditional transformer language model that uses naturally occurring control codes to steer text generation style and content.
-
Context-Aware Synthesis of Optimization Pipelines for Warehouse Optimization
CASOP is a new framework that uses a modular algorithm repository, semantic descriptions, and a problem taxonomy to synthesize and evaluate over a million valid optimization pipelines for warehouse order fulfillment across seven benchmarks.
-
Editorial Alignment: A Participatory Approach to Engaging Editorial Expertise in LLM-mediated Knowledge Dissemination
The paper introduces 'editorial alignment' as a participatory design practice that treats editorial standards as design artifacts to guide LLM behavior in knowledge dissemination, shown through workshops at one Nordic institution.
-
AI Sandboxes: A Threat Model, Taxonomy, and Measurement Framework
The paper presents a threat model, taxonomy, and six-dimension measurement framework for AI sandboxes to clarify valid testing claims for safety, security, and regulatory assurance.
-
Traxia: A Framework for Verifiable, Agent-Native Scientific Publishing
Traxia is a proposed agent-native scientific publishing framework with five formalised components: agent identity registry, verifiable publishing layer, four-tier peer review, reputation engine, and knowledge graph with contradiction detection.
-
VET: A Framework for Analyzing AI Discourse
Introduces the VET framework to categorize and critique polarized AI narratives including hype, doom, denial, and normalcy.
-
ClawHub Security Signals: When VirusTotal, Static Analysis, and SkillSpector Disagree
Analysis of 67,453 OpenClaw skills shows three scanners overlap on at most 10.4% of combined positives, with 81.9% flagged by only one scanner and distinct profiles for malicious versus suspicious skills.
-
Referential Security as a New Paradigm for AI Evaluations
Proposes referential security as a paradigm for AI evaluations that reframes model identity as verifiable to support reproducible audits and regulatory decisions despite system changes.
-
Ontological Knowledge Blocks: Executable Compliance and Profile-Based Validation for Trustworthy AI Systems
Ontological Knowledge Blocks formalize regulatory obligations as 5-tuples linking RDF/OWL schemas, SHACL rules, evidence requirements and provenance, with a compiler enabling profile-based validation demonstrated in an HPC allocation scenario.
-
NIMROD-to-IMAS workflow for extended-magnetohydrodynamic data with reusable datasets and implications for IMAS schema development
A NIMROD-to-IMAS conversion workflow preserves equilibrium, profile, perturbation and grid data from an edge harmonic oscillation simulation and identifies gaps in the IMAS schema for extended MHD.
-
The Quiet Path from Seemingly Minor Design Errors to Workplace AI Incidents
Empirical analysis of 1,524 AI incident reports shows 83% arise from worker-AI trait misalignments, with 74% of those traceable to developers prioritizing efficiency over precision or personalization.
-
The Agentic Economy: Humans, AI Agents, Robots, and the Measurable Transition toward Distributed Economic Action
The agentic economy features distributed economic action across humans, AI agents, robots, protocols, and energy systems, with quantitative diagnostics from public data indicating accelerating AI adoption, robot capacity, and task reallocation rather than labor disappearance.
-
Beyond Model Readiness: Institutional Readiness for AI Deployment in Public Systems
Introduces the Institutional Alignment Readiness (IAR) framework with five dimensions to evaluate institutional deployment readiness for AI in public systems, motivated by two anonymized education-sector cases.
-
Voices in the Loop: Mapping Participatory AI
Authors build a harmonized, geolocated atlas of participatory AI projects from existing and new sources, documenting geographic concentration and participation mostly at problem formulation and evaluation stages while providing update and governance mechanisms.
-
Mechanism Plausibility in Generative Agent-Based Modeling
Introduces the Mechanism Plausibility Scale, a four-level framework separating generative sufficiency from mechanistic plausibility in LLM-based agent-based models.
-
Exploring CoCo Challenges in ML Engineering Teams: Insights From the Semiconductor Industry
Interviews in a semiconductor company reveal 16 collaboration and communication challenges in ML engineering teams, with unclear roles and responsibilities as the top issue, and list effective mitigation practices under hardware-driven constraints.
-
Recommender Systems as Control Systems
Modeling recommender systems as control systems shows that time-optimized fairness interventions can improve overall long-term performance rather than merely trading off against utility.
-
Governing What the EU AI Act Excludes: Accountability for Autonomous AI Agents in Smart City Critical Infrastructure
The EU AI Act narrows accountability for multi-agent AI in critical infrastructure by excluding safety components from key explanation and impact assessment rights, and the paper proposes AgentGov-SC, a three-layer architecture with 25 measures to address this through traceability to existing AI and
-
Fairness-First Design Thinking for Software Architecture
A fairness-first Design Thinking method is proposed and tested in software architecture education to systematically address hidden fairness issues in digital systems.
-
Reckoning with the Political Economy of AI: Avoiding Decoys in Pursuit of Accountability
AI accountability efforts are undermined by five decoys that create illusions of progress while co-constituting the extractive political economy of the AI Project.
-
Towards A Framework for Levels of Anthropomorphic Deception in Robots and AI
A conceptual framework classifies anthropomorphic deception into four levels using humanlikeness, agency, and selfhood to guide ethical and practical decisions in HCI and HRI.
-
AI of the People, by the People, for the People: A Social Choice Approach to Collective Control of Artificial Intelligence
Proposes applying social choice theory as a modeling language and axiomatic tool for incorporating collective input across the ML development pipeline.
-
The Imbalanced User-AI Relationships as an Ethical Failure of Front-End Design in Healthcare AI
Imbalanced user-AI relationships form a distinct front-end ethical failure in healthcare AI that design choices such as restricted inputs and suppressed uncertainty can undermine agency and that reciprocity offers a path to more balanced interactions.