DailyArxiv - AI Research Podcast

Today's papers: - Efficient Text-to-Image Generation: An Adaptive Step Schedule Controller for Diffusion Models: https://arxiv.org/abs/2609.16572v1 - Pick Your Poison: Learning to Select Poison Sets for Stronger LLM Backdoor Attacks: https://arxiv.org/abs/2609.15029v1 - A Framework for Generating Valid Context-Specific Benchmarks through Expert Guidance: https://arxiv.org/abs/2609.16592v1 - OpenAI4S: Code as Action, Science as Sessions: https://arxiv.org/abs/2609.15096v1 - SeqMaestro: From nucleotide sequences to biological hypotheses through interpretable machine learning: https://arxiv.org/abs/2609.14882v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Concept-Grounded Reasoning with Prompt-Driven Localization for Interpretable Structured Report Generation: https://arxiv.org/abs/2609.15334v1 - Question's Gambit: The First Move Matters in Agentic Deep Search: https://arxiv.org/abs/2609.14412v1 - Causal multi-modal AI for personalized chemosensitivity prediction: https://arxiv.org/abs/2609.13567v1 - Building Legal Reward Models for Grounding and Abstention: https://arxiv.org/abs/2609.14739v1 - Generative AI and Extended Reality in Collaborative Architectural Design Education: An Exploratory Studio Study: https://arxiv.org/abs/2609.13494v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Agent-Based ML-LLM Fusion with Self-Optimizing Prompts for Plateau Weather Alerts: https://arxiv.org/abs/2609.10135v1 - Domain-Specific Hallucination Detection in Large Language Models: https://arxiv.org/abs/2609.11878v1 - A Comparative Evaluation of Pre-trained Convolutional Neural Networks for Melanoma Detection: https://arxiv.org/abs/2609.11550v1 - Characterizing Job Power Elasticity for Power-Flexible AI Training: https://arxiv.org/abs/2609.11542v1 - DriftNet: A Dual-Head Trajectory Transformer for Detecting and Localizing Prompt Injection in LLM Agents: https://arxiv.org/abs/2609.10892v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - WorldAgen: Unified State-Action Prediction with Test-Time World Model Training: https://arxiv.org/abs/2609.08162v1 - Evidence-Aligned Entity Verification for Hallucination Detection in Retrieval-Augmented Generation: https://arxiv.org/abs/2609.08267v1 - Elastoformer: Enabling Dynamic Adaptivity via Elastic Model Transformation: https://arxiv.org/abs/2609.10018v1 - SQLMorph: Query Mutation and Fine-Grained Metrics for Text-to-SQL Evaluation: https://arxiv.org/abs/2609.08950v1 - Time-Varying Data as Sheaves: an Invitation to Narratives: https://arxiv.org/abs/2609.09056v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - NOAH: Learning the Full Patient Journey. A Longitudinal Multimodal Time-Aware Model for Representation and Forecasting: https://arxiv.org/abs/2609.09140v1 - Agentic ML Exploration (A-MLE) for Ads Ranking: https://arxiv.org/abs/2609.08248v1 - Bag of Tricks or Bag of Myths? Reducing Modeling Complexity with Task Knowledge in Explainable Suicide Risk Assessment: https://arxiv.org/abs/2609.07766v1 - SAFEGuard: Detect Optimization-Based Jailbreak Attacks Through Harmful Semantic Analysis and Fluency Measurement: https://arxiv.org/abs/2609.05850v1 - AgentDrift: A Step-Labeled Benchmark of Injection-Hijacked LLM Agent Trajectories: https://arxiv.org/abs/2609.06972v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Leveraging Imperfect Restoration for Data Availability Attack: https://arxiv.org/abs/2609.04627v1 - Lightweight Vision Transformer Compression for On-Device Plant Disease Detection in Resource-Constrained Agricultural Field Conditions: https://arxiv.org/abs/2609.05334v1 - The Mirror Agent Model: a Bayesian Architecture for Interpretable Agent Behavior: https://arxiv.org/abs/2609.05190v1 - Sound-based Multi-Person 3D Pose Estimation: https://arxiv.org/abs/2609.04902v1 - Dynamic Heterogeneous Graph Representation Learning: A Survey: https://arxiv.org/abs/2609.04779v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Commonsense Reasoning in Computer Vision: Foundations, Recent Advancements, and Future Directions: https://arxiv.org/abs/2609.05257v1 - Cross-modal triage network: a multimodal deep learning framework for severity-based triage and visual explainability in chest radiographs: https://arxiv.org/abs/2609.04357v1 - A Schema Bounded Language Model for Refining Robot Policies Without Destabilizing Local Learning: https://arxiv.org/abs/2609.05133v1 - Local Updates, Global Learning (LUGL): Playing Games with non-incremental Learners: https://arxiv.org/abs/2609.03660v1 - Investigating the Ability of Large Language Models to Analyze Recipes for Diabetes: https://arxiv.org/abs/2609.03967v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - C$^{3}$T: Counterfactual Causal Reasoning for Sentiment Shifts in Social-Media Conversation Trees: https://arxiv.org/abs/2609.02131v1 - GraFT: A Training-Free Framework for Spatial Reasoning in Multimodal Large Language Models via 3D Scene Graphs: https://arxiv.org/abs/2609.03892v1 - ENEAS: Embedding-guided Neural Ensemble for Adaptive Segmentation: https://arxiv.org/abs/2609.03756v1 - LLaDA-Image: Building Strong Image Generators with Fully Open Training Recipes: https://arxiv.org/abs/2609.03796v1 - Spectral Convergence of Random Feature Method in Multiple Dimensions: https://arxiv.org/abs/2609.03401v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Benchmarking Vision-Language Models for Automated Pathology Diagnosis and Report Generation: https://arxiv.org/abs/2609.00866v1 - Differentially Private Paired Table-Image Multimodal Synthesis: https://arxiv.org/abs/2609.00708v1 - HiLRP: Toward One Trustworthy Explanation for Vision Transformer: Conservation-Valid Attribution via Attention Primitives: https://arxiv.org/abs/2609.01282v1 - Who Judges the Judges? A Chinese Safety QA Benchmark for Evaluating LLM Responses and Safety Judges: https://arxiv.org/abs/2609.01210v1 - VoiceLongMemEval: Do Assistants Remember How You Sounded?: https://arxiv.org/abs/2609.00570v2 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - CateKV: On Sequential Consistency for Long-Context LLM Inference Acceleration: https://arxiv.org/abs/2608.30295v1 - HBQ: Hierarchical Scaling Block Quantization with Hardware-Efficiency-Aware Design for Accurate LLM Inference: https://arxiv.org/abs/2609.00450v1 - Deploying and Evaluating a Smart-Agriculture Agentic Engine for Full-Season Soybean Farm Operations: https://arxiv.org/abs/2609.00106v1 - Fine-Grained Multi Image Object Hallucination Benchmark: https://arxiv.org/abs/2608.30653v1 - VoiceLongMemEval: Do Assistants Remember How You Sounded?: https://arxiv.org/abs/2609.00570v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Spatial-Semantic Reasoning using Large Language Models for Efficient UAV Search Operations: https://arxiv.org/abs/2608.28270v1 - Token-Efficient Data Reasoning Agents via Adaptive Structuring of Unstructured Data: https://arxiv.org/abs/2608.31082v1 - MAP: A Benchmark on Multimodal Accessibility Planning for Real World Places: https://arxiv.org/abs/2608.28384v1 - Regime-Aware Portfolio Management via Retrieval-Augmented LLM-Guided Expert Switching: https://arxiv.org/abs/2608.28252v1 - Generating Clinical Vignettes that Preserve Cognitive Formulations: https://arxiv.org/abs/2608.29995v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Accelerating Scientific Research with Gemini in the Real-World: https://arxiv.org/abs/2608.26701v1 - Learning a Continuous Sepsis Severity Score Without Hour-by-Hour Supervision: A Two-Site Retrospective Study: https://arxiv.org/abs/2608.27421v1 - CURA: Certified Runtime Alarms for Computer-Use Agents: https://arxiv.org/abs/2608.27808v1 - PCFBench: A Diagnostic Benchmark for Product Carbon Footprint Estimation: https://arxiv.org/abs/2608.27716v1 - OpenStamp: A Watermark for Open-Source Language Models: https://arxiv.org/abs/2608.27899v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - VBVR-Pro: A Scalable and Verifiable Suite for Native Visual Reasoning: https://arxiv.org/abs/2608.26105v1 - Skill Issue: Are Skills Language-Invariant in LLMs?: https://arxiv.org/abs/2608.25832v1 - Omni-Interactive Universal Embedder: https://arxiv.org/abs/2608.27044v1 - $R^3$: Training Robots to Reason in Natural Language via Reinforcement Learning: https://arxiv.org/abs/2608.26053v1 - AgentJudgeBench: A Multi-Difficulty Benchmark for Evaluating LLM Judges on Agentic Tool-Calling: https://arxiv.org/abs/2608.26623v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Parason: Revealing Subtask and Trial Parallelism in LLM Reasoning: https://arxiv.org/abs/2608.24658v1 - MLLMCLIP: Feature-Level Distillation of MLLM for Robust Vision-Language Representations: https://arxiv.org/abs/2608.25575v1 - Selective Regenerative Decoding: Trajectory-Level Intervention for Inference-Time Reasoning: https://arxiv.org/abs/2608.24338v1 - Mahalanobis-Based Multi-Head Attention for Complex State Propagation: https://arxiv.org/abs/2608.24462v1 - Hydra: Phase-Aware Workload Characterization of LLM Inference across Edge SoC Generations, Backends, and Quantization Levels: https://arxiv.org/abs/2608.25053v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Gated Activation Steering for Reducing Sycophancy & Hallucination in Medical Question Answering: https://arxiv.org/abs/2608.23666v1 - MatReplace: A Reference-Free, Conditioning-Aligned Benchmark for Material Replacement in Interior Scenes: https://arxiv.org/abs/2608.24107v1 - Prime Agent: A Self-Improving RLM Harness: https://arxiv.org/abs/2608.23552v1 - Beyond Confidence: Test-Time Scaling for Multi-Turn Search Agents via Retrieval Grounding: https://arxiv.org/abs/2608.24024v1 - Serving Masked Diffusion LLMs: Characterization and Design Principles from Real Hardware: https://arxiv.org/abs/2608.23807v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Power-Performance Characterization of TinyML Systems: https://arxiv.org/abs/2608.21646v1 - GuardPaint:SpeculativeSafetyDecodingforText-to-ImageGeneration: https://arxiv.org/abs/2608.21869v1 - Teaching LLMs How ICU Physicians Approach Clinical Reasoning Through OMOP-Aligned Retrieval Improves Reasoning Across Clinical Domains: https://arxiv.org/abs/2608.22622v1 - SAEM: Stage-Aware Expert Management for Memory-Efficient MoE Inference in Chain-of-Thought Reasoning: https://arxiv.org/abs/2608.21614v1 - Development and Feasibility Evaluation of an Edge AI as Medical Device System for Breast Cancer Multidisciplinary Team Meetings: https://arxiv.org/abs/2608.22108v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Consilience: Conformally Calibrated Communication Control for Hidden-Profile Multi-Agent Reasoning: https://arxiv.org/abs/2608.20564v1 - Volumetric Radiology AI in the Era of Multimodal Large Language Models: https://arxiv.org/abs/2608.20549v1 - Designing Human-mediated AI Guidance: Ready Together for Personalized Family Emergency Preparedness: https://arxiv.org/abs/2608.19950v1 - No PUN Intended: Plausible Unknown Names for Person-Centred LLM Evaluation: https://arxiv.org/abs/2608.21206v1 - SABET-QA: Temporal Knowledge Graph Question Answering: https://arxiv.org/abs/2608.20083v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - SPADE: Self-Play in Adaptive Synthetic Executable Environments: https://arxiv.org/abs/2608.19197v1 - Evaluating and Explaining Prompt Sensitivity of LLMs Using Interactions: https://arxiv.org/abs/2608.18539v1 - HYDRA: A Heterogeneous Chiplet DSE Framework for Serving Dynamic Hybrid LLM Workloads: https://arxiv.org/abs/2608.19395v1 - PEA-DPO: Perception-Enhanced Alignment Direct Preference Optimization for MLLMs Alignment: https://arxiv.org/abs/2608.19598v1 - In Two Minds about Lifelong Learning: Exploring Hemispheric Redundancy and Specialisation in Neural Models: https://arxiv.org/abs/2608.19514v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - MotoSafety: Edge-AI with Learned Temporal Importance for Two-Wheeler Collision Risk Assessment Under Time Pressure: https://arxiv.org/abs/2608.17823v2 - Adaptive surrogate modeling for high-dimensional spatio-temporal output: https://arxiv.org/abs/2608.17250v1 - Evaluating the Diversity of AI-Generated Content with Diversity Profiles: https://arxiv.org/abs/2608.17731v1 - From Corpora to Co-Evolving Capabilities: Capability-Centric Data Design for Generalist Image Generation: https://arxiv.org/abs/2608.18076v1 - Accuracy and Robustness of Model Cascades Under Data Perturbations: https://arxiv.org/abs/2608.17711v2 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - KnowSim: Evaluating Information Calibration in LLM Assistants with User Simulators that Learn: https://arxiv.org/abs/2608.17150v1 - Without journalists, there is no journalism: the social dimension of generative artificial intelligence in the media: https://arxiv.org/abs/2608.17017v1 - Delegation Asymmetry in Agentic Recommender Systems: Measuring Two-Sided Receptivity in Online Dating: https://arxiv.org/abs/2608.18058v1 - MotoSafety: Edge-AI with Learned Temporal Importance for Two-Wheeler Collision Risk Assessment Under Time Pressure: https://arxiv.org/abs/2608.17823v1 - The 10th AI City Challenge: https://arxiv.org/abs/2608.17044v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Agent Gym: A Framework for Continuous Evaluation and Evolution of LLM Agents Through Human-in-the-Loop Feedback: https://arxiv.org/abs/2608.15591v1 - Emergent Misaligned Communication in Long-Horizon Multi-Agent LLM Commerce: https://arxiv.org/abs/2608.14825v1 - Self-Supervised Visual On-Policy Distillation: https://arxiv.org/abs/2608.14144v1 - FZ-VLM: A Two Stage Florence-Zephyr Vision Language Model Framework for Pulmonary Nodule Characterization and Clinical Decision Making: https://arxiv.org/abs/2608.15004v1 - BGA: A noise-immune neural distillation framework for malicious signature extraction in high-entropy encrypted flows: https://arxiv.org/abs/2608.14126v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Can Language Models Understand mmWave Data? Benchmarking Large Language Models for mmWave Radar-Based Human Understanding: https://arxiv.org/abs/2608.14179v1 - Amplified Does Not Mean Predictive: Reasoning Behaviors in Thinking Models: https://arxiv.org/abs/2608.13760v1 - When Personal Memory Has No Single Answer: Evaluating LLM Agents under Irreducible Conflict: https://arxiv.org/abs/2608.13921v1 - A Pathway to General-Purpose Scientific AI: Multimodal Comprehension of Scientific Images: https://arxiv.org/abs/2608.14075v1 - Meteorology-driven Causal Nowcasting of Fugitive Landfill Emissions Enables Proactive Public Health Response: https://arxiv.org/abs/2608.14254v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - FlashDrive: Flash Vision-Language-Action Inference for Autonomous Driving: https://arxiv.org/abs/2608.12932v1 - The Role of Natural Language Understanding in Multimodal Video-Based Dengue Diagnosis: https://arxiv.org/abs/2608.12677v1 - EgoMonth: A Month-Level Egocentric Video Benchmark for Long-Term Spatiotemporal Memory: https://arxiv.org/abs/2608.13113v1 - On the Expressive Power of Transformers: https://arxiv.org/abs/2608.12671v1 - AI and Consumer Rights in India Working Paper: https://arxiv.org/abs/2608.12863v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - From Interpretability to Control: Insights from Six Years of the TrustNLP Workshop - Reinforcing Step-level Reasoning for Effective Self-Correction in LLMs - Rule of Thumb: Explaining Artificial Intelligence Systems using Partial Information - VERDICT: Training-Free Step-Wise Verification of Multimodal Reasoning via Disagreement-Aware Consensus - Rubric Dropout: A Simple Way to Mitigate Reward Hacking in Rubric-as-Reward RL This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Towards Expert-level Medical AI for Real-time Video Consultations: https://arxiv.org/abs/2608.09861v1 - TCS-BENCH: Benchmarking State-of-the-Art Generative AI Theoretical Computer Science Research Ability: https://arxiv.org/abs/2608.09538v1 - Multimodal Model Diffing for Feature Discovery and Control: https://arxiv.org/abs/2608.09928v1 - Motif 3: Technical Report: https://arxiv.org/abs/2608.09119v1 - P$^{3}$: Joint Program-and-Proof Planning for Verified Code Generation: https://arxiv.org/abs/2608.09277v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - ResidencyRL: Reinforcement Learning in Simulated Clinical Environments: https://arxiv.org/abs/2608.07418v1 - Characterizing the Quality Profile of AI-Generated C++ in Production: https://arxiv.org/abs/2608.06640v1 - Reducing belief in conspiracy theories as they unfold using large language models: https://arxiv.org/abs/2608.06151v1 - Taxonomy-Driven Analysis of Open-Source AI Risk Mitigation Tools: https://arxiv.org/abs/2608.07446v1 - Omni-modal decomposition autoencoders learn full-stack wearable disentangled representations: https://arxiv.org/abs/2608.07385v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - The First EgoCross Challenge at EgoVis 2026: Cross-Domain Egocentric Video Question Answering: https://arxiv.org/abs/2608.04589v1 - Chained Recursive Language Models for Multi-Iteration Reasoning: https://arxiv.org/abs/2608.05124v1 - Easy to Complete, Hard to Choose: Investigating LLM Performance on the ProverbIT Benchmark: https://arxiv.org/abs/2608.04670v1 - Trident : How to Break Deep Reinforcement Learning Cyber Defenses (Agentic): https://arxiv.org/abs/2608.04317v1 - Domain-Grounded Candidate Selection for Agentic Image Editing: A Shadow Removal Case: https://arxiv.org/abs/2608.06075v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - TextNCA: Neural Cellular Automata for Language Modeling via Hierarchical Local Attention: https://arxiv.org/abs/2608.02050v1 - A Graph Signal Processing Perspective on Numerical Sequence Representations in LLM In-Context Learning: https://arxiv.org/abs/2608.03015v1 - A Taxonomy of Cognitive Capability Gaps in Generative and Agentic AI: https://arxiv.org/abs/2608.02553v1 - OliveGemma: A 3 Billion Visual Language Model for Recognising the Mediterranean & European Diet: https://arxiv.org/abs/2608.03428v1 - Weights or Skills? A Survey of Robot-Learning Techniques: from Action-Predicting Weights to Robots that Write their Own Skills: https://arxiv.org/abs/2608.01851v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Dense Temporal Contrast Synthesis via Conditioned Latent Transport: https://arxiv.org/abs/2607.29394v1 - TerraNova: A Foundation Model for the Anthropocene: https://arxiv.org/abs/2607.29527v1 - DiffusionGemma Technical Report: https://arxiv.org/abs/2608.00146v1 - FactorJEPA: Factorizing Monolithic Futures into Layout-Agent-Interaction Channels for Crowded and Chaotic Global South Urban Worlds: https://arxiv.org/abs/2608.01049v1 - Verifiable Checks for Business Rule Consistency: https://arxiv.org/abs/2608.00396v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - ORCA-bench: How Ready Are Language Model Agents for Oncall?: https://arxiv.org/abs/2607.28545v1 - Using Large Language Models for Idea Generation in Innovation: https://arxiv.org/abs/2607.27553v1 - AISPA: User-Centric System Prompt Auditing for Large Language Model Applications: https://arxiv.org/abs/2607.28617v1 - EMBL AI Librarian: Life-Sciences Knowledge Layer for AI Agents: https://arxiv.org/abs/2607.28229v1 - OSReward: Instituting Standardized Evaluation for Cross-Platform Computer-Use Reward Models: https://arxiv.org/abs/2607.28609v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - WhisperRec: Latent Reasoning for Efficient Foundation Recommendation Models: https://arxiv.org/abs/2607.26621v2 - MultivationBench: A Benchmark for Multimodal Sequential Motivation Reasoning: https://arxiv.org/abs/2607.26465v1 - See2Think: Do Multimodal Models Really Use Intermediate Visual States?: https://arxiv.org/abs/2607.26769v1 - Hearsay: Vision-Language Medical Diagnoses Without an Image: https://arxiv.org/abs/2607.26886v1 - Can AI agents conduct open-ended AI research? Early evidence from two case studies: https://arxiv.org/abs/2607.27191v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - GPT-Red: Automated Red Teaming via Self-Play at Scale: https://arxiv.org/abs/2607.26115v1 - Raven: High-Recall Sequence Modeling with Sparse Memory Routing: https://arxiv.org/abs/2607.25357v1 - WhisperRec: Latent Reasoning for Efficient Foundation Recommendation Models: https://arxiv.org/abs/2607.26621v1 - Evaluating Multi-Turn Multimodal Diagnostic Reasoning on Challenging Real-World Clinical Cases: https://arxiv.org/abs/2607.25933v1 - Beyond Counts: A Distributional Robustness Margin For Pathology Foundation Models: https://arxiv.org/abs/2607.25497v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - EgoPlay: Event-Triggered Video Editing for Egocentric Streams: https://arxiv.org/abs/2607.24560v1 - Learning from 53.6K Real-World Developer Edits of AI-Generated Code: https://arxiv.org/abs/2607.25130v1 - CADER: Confidence-Aware Dynamic Evidence Reasoning for Long-Video Understanding: https://arxiv.org/abs/2607.24582v1 - CoTinyVLA: Chain-of-Thought Distillation for a Sub-Billion-Parameter Vision-Language-Action Model: https://arxiv.org/abs/2607.25487v1 - KANEx: Translating Kolmogorov-Arnold Networks' Interpretability to Medical Explainability: https://arxiv.org/abs/2607.24730v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - A Roadmap to Impactful Pluralistic Alignment Research: https://arxiv.org/abs/2607.22305v1 - Learning to Prepare Molecular Ground States with Transformer Models: https://arxiv.org/abs/2607.22468v1 - Training Language Models to Cooperate with Inference-Time Controllers: https://arxiv.org/abs/2607.23771v1 - Building AI That Works: ESnet's Pragmatic Approach to AI-Driven Operational Excellence: https://arxiv.org/abs/2607.22948v1 - IDEAgent: Agentic Quality-Diversity Search for Research Idea Generation: https://arxiv.org/abs/2607.22375v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - OpenForgeRL: Train Harness-native Agents in Any Environment: https://arxiv.org/abs/2607.21557v2 - Filling Before Advancing: Capability-Gap-Driven Post-Training for Scenario-Specialized Remote Sensing MLLMs: https://arxiv.org/abs/2607.22205v1 - Code Monitor Red Teaming for Public-Test-Passing Code: https://arxiv.org/abs/2607.20852v1 - TRaM-VSR: Importance-Aware Token Routing and Merging for One-Step Diffusion Video Super-Resolution: https://arxiv.org/abs/2607.22231v1 - Beyond Independent Optimization: Compression, MoE Routing, and Quantization Interactions in Multimodal Edge Intelligence: https://arxiv.org/abs/2607.20981v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - PoTRE: Test-Time Reasoning inspired by Cognitive Heterogeneity: https://arxiv.org/abs/2607.20268v1 - Overview of FinMMEval 2026 Task 2: Multilingual Financial Short-Answer Question Answering: https://arxiv.org/abs/2607.19867v1 - Did Alice Do Wrong? Cross-Cultural Differences in Student Perceptions of Generative AI Use in University Computing Education: https://arxiv.org/abs/2607.19699v1 - Post-Training in Time Series Foundation Models: A Unifying Framework: https://arxiv.org/abs/2607.20002v1 - slang.gr as a Large-Scale Crowdsourced Resource for Non-Standard Greek: https://arxiv.org/abs/2607.21255v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Mage-Flow: An Efficient Native-Resolution Foundation Model for Image Generation and Editing: https://arxiv.org/abs/2607.19064v2 - Attributes Should Come from Images, Not Class Names: Distribution-Conditioned Attribute Selection for Vision-Language Models: https://arxiv.org/abs/2607.18695v1 - Small, Free, and Effective: Orchestrating Open-Weight Small Language Models to Outperform Single LLM for Malware Analysis: https://arxiv.org/abs/2607.20216v1 - Overview of FinMMEval 2026 Task 1: Multilingual Financial Multiple-Choice Question Answering: https://arxiv.org/abs/2607.19856v1 - Data Leakage Prevention in Agentic Applications via Preemptive Hardening: https://arxiv.org/abs/2607.18847v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - HyCoRec: Hypergraph-Enhanced Multi-Preference Learning for Alleviating Matthew Effect in Conversational Recommendation: https://arxiv.org/abs/2607.17461v1 - Coarse-to-fine Framework for Generative MEF via Implicit Neural Representation: https://arxiv.org/abs/2607.17611v1 - Simple Domain Generalization for Strong Pixel-Level Image Tampering Detection in Modern VLMs: https://arxiv.org/abs/2607.18230v1 - Find Before You Fine-Tune: A Diagnostic Study of Small LLMs for Cybersecurity QA: https://arxiv.org/abs/2607.18725v1 - Mage-Flow: An Efficient Native-Resolution Foundation Model for Image Generation and Editing: https://arxiv.org/abs/2607.19064v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Frontier AI performance across the business disciplines: a case-grounded benchmark of knowledge work and analytical reasoning: https://arxiv.org/abs/2607.16057v1 - ThAME: 3D Memory-Enabled Heterogeneous Accelerator for LLM Mixture of Experts: https://arxiv.org/abs/2607.17074v1 - Hard Rules, Soft Preferences: Bridging Reasoning, Learning, and Optimization for Personalized Packing Checklist Generation: https://arxiv.org/abs/2607.15562v1 - When Do Multi-Agent Systems Help? An Information Bottleneck Perspective: https://arxiv.org/abs/2607.16133v1 - Knowledge-Centric Agents for Workflow Generation: https://arxiv.org/abs/2607.15845v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Verbalizable Representations Form a Global Workspace in Language Models: https://arxiv.org/abs/2607.15495v1 - Orbis 2: A Hierarchical World Model for Driving: https://arxiv.org/abs/2607.15898v1 - RoboTTT: Context Scaling for Robot Policies: https://arxiv.org/abs/2607.15275v1 - In-Place Tokenizer Expansion for Pre-trained LLMs: https://arxiv.org/abs/2607.15232v1 - Test-Time Noise Guided Adaptation for Realistic Autoregressive Video Generation: https://arxiv.org/abs/2607.15849v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Towards a Unified Multidimensional Explainability Metric: Evaluating Trustworthiness in AI Models: https://arxiv.org/abs/2607.14315v1 - RxBrain: Embodied Cognition Foundation Model with Joint Language-Visual Reasoning and Imagination: https://arxiv.org/abs/2607.14187v1 - Action QFormer: Structured Representation Shaping under Action Supervision in Vision-Language-Action Models: https://arxiv.org/abs/2607.14635v1 - CatalogAgent: A Supervisor-mediated Self-Learning System Enabling Context Engineering for GenAI Models: https://arxiv.org/abs/2607.14396v1 - Instant NuRec: Feed-Forward 3D Gaussian Reconstruction for Driving Scene Simulation: https://arxiv.org/abs/2607.14203v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - MxGPS: Multiplex Graph Transformers for a Power Grid Foundation Model: https://arxiv.org/abs/2607.13763v1 - Improving Molecular Property Prediction in Small Language Models Using Graph-based Tools: https://arxiv.org/abs/2607.13115v1 - From Observation to Insight: Mechanistic World Models and the Quest for Autonomous Discovery: https://arxiv.org/abs/2607.12474v2 - The Computational Basis of Confidence in Large Language Models: https://arxiv.org/abs/2607.12447v1 - Grounded world models in biological organisms and future embodied AI: https://arxiv.org/abs/2607.13560v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Evidence-Backed Video Question Answering: https://arxiv.org/abs/2607.11862v1 - Technical Report on the CVPR 2026@AdvML Workshop Challenge: https://arxiv.org/abs/2607.11560v1 - Toward Trustworthy Autonomous Science: A Two-Year Community Roadmap: https://arxiv.org/abs/2607.12113v1 - Xiaomi-Robotics-U0: Unified Embodied Synthesis with World Foundation Model: https://arxiv.org/abs/2607.11643v1 - MMA-Former: Multi-Window Mixture-of-Head Attention Transformer for Adaptive PNI Prediction in 3D MRI: https://arxiv.org/abs/2607.10988v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Video Generation Models are General-Purpose Vision Learners: https://arxiv.org/abs/2607.09024v1 - Phone Segmentation and Recognition through Phonological Activation Mapping: https://arxiv.org/abs/2607.09020v1 - Integrating Large Language Models and Graph Convolutional Networks for Semi-Supervised Image Classification: https://arxiv.org/abs/2607.09104v1 - Extending LLM Context via Associative Recurrent Memory: https://arxiv.org/abs/2607.11614v1 - SynthDocBench: Controlled Benchmark for Long-Context Visual Document Understanding: https://arxiv.org/abs/2607.10400v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Reinforcing the Generation Order of Multimodal Masked Diffusion Models: https://arxiv.org/abs/2607.08056v1 - ProofCouncil: An LLM Agent for Solving Open Mathematical Problems: https://arxiv.org/abs/2607.09474v1 - WebSwarm: Recursive Multi-Agent Orchestration for Deep-and-Wide Web Search: https://arxiv.org/abs/2607.08662v1 - Prompt-Driven Exploration: https://arxiv.org/abs/2607.08837v1 - GATS: Graph-Augmented Tree Search with Layered World Models for Efficient Agent Planning: https://arxiv.org/abs/2607.08894v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Multimodal Unlearning Across Vision, Language, Video, and Audio: Survey of Methods, Datasets, and Benchmarks: https://arxiv.org/abs/2607.07907v1 - COBART: Controlled, Optimized, Bidirectional and Auto-Regressive Transformer for Ad Headline Generation: https://arxiv.org/abs/2607.08071v1 - Accurate, Interdisciplinary and Transparent Structure-property Understanding with Deep Native Structural Reasoning: https://arxiv.org/abs/2607.07708v1 - Towards Efficient Large Language Model Serving: A Survey on System-Aware KV Cache Optimization: https://arxiv.org/abs/2607.08057v1 - From Solvers to Research: Large Language Model-Driven Formal Mathematics at the Research Frontier: https://arxiv.org/abs/2607.07779v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Ad Headline Generation using Self-Critical Masked Language Model: https://arxiv.org/abs/2607.06818v1 - SPEAR: A Simulator for Photorealistic Embodied AI Research: https://arxiv.org/abs/2607.06701v1 - TopoBrick: Agentic Topology Sampling of Exogenous Variables for Zero-Shot Building IoT Forecasting: https://arxiv.org/abs/2607.06349v1 - Search, Fail, Recover: A Training Framework for Correction-Aware Reasoning: https://arxiv.org/abs/2607.07492v1 - Beyond Static Evaluation: Building Simulation Environments for Scalable Agentic Reinforcement Learning: https://arxiv.org/abs/2607.05773v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Unified Audio Intelligence Without Regressing on Text Intelligence: https://arxiv.org/abs/2607.05196v2 - LLM-as-a-Verifier: A General-Purpose Verification Framework: https://arxiv.org/abs/2607.05391v2 - DSpark: Confidence-Scheduled Speculative Decoding with Semi-Autoregressive Generation: https://arxiv.org/abs/2607.05147v1 - Wasserstein Residuals: Learning Gradient Flows from Population Dynamics: https://arxiv.org/abs/2607.04738v1 - TILDE: TILt-based Distributional Erasure for Concept Unlearning: https://arxiv.org/abs/2607.06432v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - Where do LLMs Fall Short in CBT-Guided Affective Reasoning?: https://arxiv.org/abs/2607.02885v1 - DynaVieW: Schema-Guided World Modeling for Understanding Hierarchical Visual Dynamics: https://arxiv.org/abs/2607.04112v1 - Nemotron-Labs-3-Puzzle-75B-A9B: Compressing Hybrid MoE LLMs: https://arxiv.org/abs/2607.04371v1 - Unsupervised Features Mining via Activation Geometry: https://arxiv.org/abs/2607.04222v1 - Gemma 4 Technical Report: https://arxiv.org/abs/2607.02770v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.

Today's papers: - PACE: A Proxy for Agentic Capability Evaluation: https://arxiv.org/abs/2607.02032v1 - Mastermind: Strategy-grounded Learning for Repository-Scale Vulnerability Reproduction: https://arxiv.org/abs/2607.01764v1 - AIriskEval-edu: New Dataset for Risk Assessment in AI-mediated K-12 Educational Explanations: https://arxiv.org/abs/2607.01934v1 - UA-ChatDev: Uncertainty-Aware Multi-Agent Collaboration for Reliable Software Development: https://arxiv.org/abs/2607.02186v1 - Overview of Risk Assessment and Management for Intelligent Systems under the AI Act and Beyond: https://arxiv.org/abs/2607.02197v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.