HN matt_d 4/24/2026

How Do LLM Agents Think Through SQL Join Orders?

Research collaboration exploring how LLM agents optimize SQL join orders through iterative reasoning.

HN salkahfi 4/24/2026

AI-powered knowledge assistant for sexual and reproductive health

ChatHRP: AI health knowledge assistant providing answers from WHO documents with source citations.

HN vednig 4/24/2026

Contral AI

Contral: AI-powered development environment with agent for building, debugging, and explaining code.

HN dandiep 4/24/2026

OpenAI deprecates all GPT nano fine tuning

OpenAI deprecates GPT nano fine-tuning models, unclear migration path to GPT-5 variants.

HN umangsehgal93 4/24/2026

The Budgeting Mistake That Cost Uber Its Annual AI Spend in 4 Months

Case study of Uber's AI budget overrun in 4 months due to misunderstanding variable vs fixed costs of AI deployment.

HN mroche 4/24/2026

Intel Arc Pro B70 benchmarks for LLMs and video generation

Benchmarks and performance data for Intel Arc Pro B70 GPU on LLM inference and video generation.

HN Cider9986 4/24/2026

DeepSeek's Sequel Set to Extend China's Reach in Open-Source A.I

News headline about DeepSeek releasing open-source AI model sequel. Minimal detail provided.

HN maxaw 4/24/2026

Show HN: GitRails-Let agents call only the GitHub endpoints and params you allow

GitRails: Open-source GitHub API proxy enabling fine-grained access control for untrusted AI agents.

HN deadalus 4/24/2026

Grok Voice Think Fast 1.0

xAI announces grok-voice-think-fast-1.0, voice agent API for customer support and enterprise workflows.

HN celadevra_ 4/24/2026

Anthropic tested removing Claude Code from the Pro plan

Anthropic tested removing Claude Code from Pro plan due to high demand; pricing adjustment to ration agentic development tool access.

Ax Ji-il Park, Inwook Shim, Chong Hui Kim 4/24/2026

Architecture of an AI-Based Automated Course of Action Generation System for Military Operations

arXiv paper on AI-based automated military course of action planning systems for operational decision support.

Ax Michael O'Herlihy, Rosa Catal\`a 4/24/2026

Escaping the Agreement Trap: Defensibility Signals for Evaluating Rule-Governed AI

arXiv paper on evaluating content moderation AI systems using policy-grounded correctness instead of agreement metrics.

Ax Xiyang Wu, Zongxia Li, Guangyao Shi, Alexander Duffy, Tyler Marques, Matthew Lyle Olson, Tianyi Zhou, Dinesh Manocha 4/24/2026

Co-Evolving LLM Decision and Skill Bank Agents for Long-Horizon Tasks

arXiv paper on co-evolving LLM decision-making and skill banks for long-horizon interactive tasks in game environments.

Ax Inderjeet Nair, Jie Ruan, Lu Wang 4/24/2026

Value-Conflict Diagnostics Reveal Widespread Alignment Faking in Language Models

arXiv paper introducing diagnostics for detecting alignment faking in language models under monitoring vs unmonitored conditions.

Ax Haebin Seong, Li Yin, Haoran Zhang 4/24/2026

The Last Harness You'll Ever Build

arXiv paper on AI agents handling complex domain-specific workflows, addressing harness engineering challenges for enterprise automation tasks.

Ax Mirazul Haque, Antony Papadimitriou, Samuel Mensah, Zhiqiang Ma, Zhijin Guo, Joy Prakash Sain, Simerjot Kaur, Charese Smiley, Xiaomo Liu 4/24/2026

Deep FinResearch Bench: Evaluating AI's Ability to Conduct Professional Financial Investment Research

Benchmark evaluating AI agents' ability to conduct professional financial research with metrics for rigor, accuracy, and verifiability.

Ax Bowen Zuo, Dongruo Zhou, Yinglun Zhu 4/24/2026

Adaptive Test-Time Compute Allocation with Evolving In-Context Demonstrations

Adaptive test-time compute allocation framework that evolves in-context demonstrations to optimize LLM performance efficiency.

Ax Yuyu Liu, Sarang Rajendra Patil, Mengjia Xu, Tengfei Ma 4/24/2026

HypEHR: Hyperbolic Modeling of Electronic Health Records for Efficient Question Answering

HypEHR uses hyperbolic geometry to embed EHR codes and answer clinical questions efficiently without LLM-based pipelines.

Ax Marzia Binta Nizam, James Davis 4/24/2026

Who Defines Fairness? Target-Based Prompting for Demographic Representation in Generative Models

Target-based prompting method to improve demographic representation in text-to-image generative models.

Ax Nicklas Neu, Thomas Ebner, Jasmin Primus, Raphael Zefferer, Bernhard Schenkenfelder, Mathias Brunbauer, Florian Kromp 4/24/2026

InVitroVision: a Multi-Modal AI Model for Automated Description of Embryo Development using Natural Language

Fine-tuned vision-language model for automated natural language description of embryo morphology in IVF applications.

Ax Gricel V\'azquez, Alexandros Evangelidis, Sepeedeh Shahbeigi, Radu Calinescu, Simos Gerasimou 4/24/2026

Mind the Prompt: Self-adaptive Generation of Task Plan Explanations via LLMs

Self-adaptive prompt engineering framework for generating task plan explanations from LLMs via systematic refinement.

Ax Olli J\"arviniemi, Oliver Makins, Jacob Merizian, Robert Kirk, Ben Millwood 4/24/2026

Propensity Inference: Environmental Contributors to LLM Behaviour

Methods for measuring LLM propensity for unsanctioned behavior through environmental factor analysis and Bayesian modeling.

Ax Andrew J. Peterson 4/24/2026

AI Governance under Political Turnover: The Alignment Surface of Compliance Design

Analysis of AI governance compliance design and regulatory frameworks under political change. Policy-focused.

Ax Abhishek Dharmaratnakar, Srivaths Ranganathan, Anushree Sinha, Debanshu Das 4/24/2026

Agentic AI for Personalized Physiotherapy: A Multi-Agent Framework for Generative Video Training and Real-Time Pose Correction

Multi-agent system for personalized physiotherapy using generative AI and computer vision for real-time pose correction.

Ax Tristan Shah, Ilya Nemenman, Daniel Polani, Stas Tiomkin 4/24/2026

Multi-Agent Empowerment and Emergence of Complex Behavior in Groups

Multi-agent empowerment framework studying emergence of complex behaviors through intrinsic motivation in agent groups.

Ax Vipula Rawte, Ryan Rossi, Franck Dernoncourt, Nedim Lipka 4/24/2026

Trust but Verify: Introducing DAVinCI -- A Framework for Dual Attribution and Verification in Claim Inference for Language Models

DAVinCI framework for dual attribution and verification to reduce LLM hallucinations and improve factual accuracy.

Ax Yanan Wang, Yong Ge 4/24/2026

Align Generative Artificial Intelligence with Human Preferences: A Novel Large Language Model Fine-Tuning Method for Online Review Management

Fine-tuning LLMs for automated online review response generation aligned with human preferences.

Ax Xiyin Zeng, Yuyu Sun, Haoyang Li, Shouqiang Liu, Hao Wang 4/24/2026

ReCAPA: Hierarchical Predictive Correction to Mitigate Cascading Failures

ReCAPA framework mitigates cascading failures in vision-language-action multi-step task systems through hierarchical correction.

Ax Michael Bouzinier, Sergey Trifonov, Michael Chumack, Eugenia Lvova, Dmitry Etin 4/24/2026

Trustworthy Clinical Decision Support Using Meta-Predicates and Domain-Specific Languages

Framework for auditable clinical decision support using domain-specific languages and formal verification methods.

Ax Minping Chen, Bing Xu, Zulong Chen, Chuanfei Xu, Ying Zhou, Zui Tao, Zeyi Wen 4/24/2026

Enhancing Online Recruitment with Category-Aware MoE and LLM-based Data Augmentation

LLM-based data augmentation method with category-aware mixture-of-experts for person-job fit in online recruitment systems.

Ax Jindi Guo, Xi Fang, Chaozheng Huang 4/24/2026

Can MLLMs "Read" What is Missing?

Benchmark evaluating multimodal LLM ability to reconstruct masked text from visual context in documents and webpages without explicit prompts.

Ax Robin Dey, Panyanon Viradecha 4/24/2026

Spatial Metaphors for LLM Memory: A Critical Analysis of the MemPalace Architecture

Critical analysis of MemPalace open-source memory system using spatial metaphors for LLM long-term memory with high retrieval performance claims.

Ax Donggyu Lee, Hyeok Yun, Jungwon Kim, Junsik Min, Sungwon Park, Sangyoon Park, Jihee Kim 4/24/2026

Ideological Bias in LLMs' Economic Causal Reasoning

Systematic evaluation of ideological bias in LLM economic causal reasoning using extended benchmark with ideology-contested intervention cases.

Ax Philip Zhong, Don Wang, Jason Zhang, Kent Chen 4/24/2026

Evaluating AI Meeting Summaries with a Reusable Cross-Domain Pipeline

Reusable evaluation pipeline for AI-generated meeting summaries with public artifact package separating orchestration from task-specific evaluation logic.

Ax Mohit Vaishnav, Tanel Tammet 4/24/2026

Symbolic Grounding Reveals Representational Bottlenecks in Abstract Visual Reasoning

Study comparing vision-language models vs LLMs on abstract visual reasoning to identify whether reasoning or representation bottlenecks limit performance.

Ax Jian Cui, Zhiyuan Ren, Desheng Weng, Yongqi Zhao, Gong Wenbin, Yu Lei, Zhenning Dong 4/24/2026

ReaGeo: Reasoning-Enhanced End-to-End Geocoding with LLMs

End-to-end geocoding framework using LLMs with geohash sequence reformulation to overcome limitations of traditional multi-stage retrieval approaches.

Ax Ankur Sharma, Deep Shah, David Lariviere, Hesham ElBakoury 4/24/2026

Time, Causality, and Observability Failures in Distributed AI Inference Systems

Analysis of clock skew effects on observability and causality in distributed AI inference systems; infrastructure/observability focus.

Ax Qiang Gao, Zhenping Li, Anqi Zhuo, Yingxiao Zhao, Weibo Geng, Xiaosong Li 4/24/2026

SemanticAgent: A Semantics-Aware Framework for Text-to-SQL Data Synthesis

Semantic-aware synthesis framework using specialized agent modules (analyzer, synthesizer, verifier) for improving text-to-SQL generation correctness.

Ax Jinhee Jang, Juhwan Choi, Dongjin Lee, Seunguk Yu, Youngbin Kim 4/24/2026

FairQE: Multi-Agent Framework for Mitigating Gender Bias in Translation Quality Estimation

Multi-agent framework addressing gender bias in machine translation quality estimation through fairness-aware agent collaboration.

Ax Yue Teng, Qianer Zhong, Kim Mai Tich Nguyen Thordsen, Christian Montag, Benjamin Becker 4/24/2026

Brief chatbot interactions produce lasting changes in human moral values

Study demonstrating that brief chatbot conversations produce measurable shifts in human moral value judgments through directive AI conversations.

Ax Yuehan Zhu, Jingqi Zhao, Jiawen Zhao, Xudong Mao, Baoquan Zhao 4/24/2026

HiCrew: Hierarchical Reasoning for Long-Form Video Understanding via Question-Aware Multi-Agent Collaboration

Multi-agent framework for long-form video understanding with hierarchical reasoning and question-aware agent collaboration for improved temporal coherence.

Ax Andrew Shin 4/24/2026

AI-Gram: When Visual Agents Interact in a Social Network

Live platform studying social dynamics of LLM-driven visual agents interacting in autonomous multi-agent network with image-based communication.

Ax Itay Nakash, George Kour, Ateret Anaby-Tavor 4/24/2026

Efficient Agent Evaluation via Diversity-Guided User Simulation

Framework for efficient LLM agent evaluation via diversity-guided user simulation to identify deep failure modes in multi-turn interactions.

Ax Yitong Zhou, Mingyue Cheng, Jiahao Wang, Qingyang Mao, Qi Liu 4/24/2026

GeoMind: An Agentic Workflow for Lithology Classification with Reasoned Tool Invocation

AI agent system using reasoned tool invocation for geological lithology classification, demonstrating agentic workflow with evidence-based reasoning.

Ax Jiaxian Yan, Jintao Zhu, Yuhang Yang, Qi Liu, Kai Zhang, Zaixi Zhang, Xukai Liu, Boyan Zhang, Kaiyuan Gao, Jinchuan Xiao, Enhong Chen 4/24/2026

BioMiner: A Multi-modal System for Automated Mining of Protein-Ligand Bioactivity Data from Literature

Multi-modal system extracting protein-ligand bioactivity data from scientific literature using ML on text, tables, and chemical structures.

Ax Fridolin Linder, Thomas Leeper, Daniel Haimovich, Niek Tax, Lorenzo Perini, Milan Vojnovic 4/24/2026

Unbiased Prevalence Estimation with Multicalibrated LLMs

Method using multicalibration to correct LLM prevalence estimation errors across population shifts, addressing diagnostic accuracy in varying contexts.

Ax Simon Jarvers, Orestis Papakyriakopoulos 4/24/2026

Engaged AI Governance: Addressing the Last Mile Challenge Through Internal Expert Collaboration

Action research studying team-level implementation of EU AI Act requirements in an AI startup, creating legal-text-to-action pipeline for governance.

Ax Chris Schneider, Philipp Schoenegger, Ben Bariach 4/24/2026

Separable Expert Architecture: Toward Privacy-Preserving LLM Personalization via Composable Adapters and Deletable User Proxies

Privacy-preserving LLM personalization architecture using composable LoRA adapters and deletable user proxies for efficient data removal without retraining.

Ax Maximilian Westermann, Ben Griffin, Aaron Ontoyin Yin, Zakari Salifu, Yagiz Ihlamur, Kelvin Amoaba, Joseph Ternasky, Fuat Alican, Yigit Ihlamur 4/24/2026

CoFEE: Reasoning Control for LLM-Based Feature Discovery

CoFEE provides structured reasoning control for LLM-based feature discovery from unstructured data while avoiding leakage and confounding signals.

Ax Nevena Lazi\'c, Liam Fowl, Andr\'as Gy\"orgy, Csaba Szepesv\'ari 4/24/2026

To See the Unseen: on the Generalization Ability of Transformers in Symbolic Reasoning

Investigates transformer generalization on symbolic reasoning tasks, showing models fail with unseen variable names and analyzing token copying difficulties.