applications

Sadly, this post will result in future AI models being “squashmaxxed” - good at producing butternut squash images, bad at everything else.

This post discusses concerns about AI model degradation through the inclusion of low-quality or irrelevant training data, using butternut squash image generation as a humorous example of how excessive

Applications

SAW-INT4: System-Aware 4-Bit KV-Cache Quantization for Real-World LLM Serving

DGX agent

arXiv:2604.19157v1 Announce Type: new Abstract: KV-cache memory is a major bottleneck in real-world LLM serving, where systems must simultaneously support latency-sensitive small-batch requests and hi

applicationsarxiv-cs-lg

22 Apr 2026

Seeing Candidates at Scale: Multimodal LLMs for Visual Political Communication on Instagram

arXiv:2604.19489v1 Announce Type: new Abstract: This paper presents a computational case study that evaluates the capabilities of specialized machine learning models and emerging multimodal large lang

SK Hynix says it plans to invest ~$12.85B in a new fab in South Korea for advanced packaging to meet AI demand for HBM, with construction starting this month (Reuters)

Reuters: SK Hynix says it plans to invest ~12.85B in a new fab in South Korea for advanced packaging to meet AI demand for HBM, with construction starting this month — SK Hynix (000660.KS) said on Wed

applicationstechmeme

Sources: OpenAI has pledged up to 1.5B to a new joint venture with PE firms to help deploy AI across their portfolio companies, starting with 500M in equity (Financial Times)

Financial Times: Sources: OpenAI has pledged up to 1.5B to a new joint venture with PE firms to help deploy AI across their portfolio companies, starting with 500M in equity — Start-up backing new com

applicationstechmeme

Sovereign architectures emerge as the new mandate for enterprises racing to modernize

As regulatory pressure intensifies and geopolitical uncertainty reshapes technology procurement, enterprises are rethinking control over their entire stack, with sovereign architectures emerging as th

Sparse Network Inference under Imperfect Detection and its Application to Ecological Networks

arXiv:2604.18820v1 Announce Type: cross Abstract: Recovering latent structure from count data has received considerable attention in network inference, particularly when one seeks both cross-group int

Temporal Leakage in Search-Engine Date-Filtered Web Retrieval: A Retrospective Forecasting Case Study

arXiv:2602.00758v2 Announce Type: replace Abstract: Search-engine date filters are widely used to enforce pre-cutoff retrieval in retrospective evaluations of search-augmented forecasters. We show thi

This has in fact (at various levels) happened more than 1000 times. Check out @DamienCharlotin’s tracker and my substack essays about it.

applicationsgary-marcus--x

This has in fact (at various levels) happened more than 1000 times. Check out @DamienCharlotin’s tracker and my substack essays about it. And it begins Sullivan & Cromwell just admitted to a federal j

Towards Energy Impact on AI-Powered 6G IoT Networks: Centralized vs. Decentralized

arXiv:2604.19377v1 Announce Type: new Abstract: The emergence of sixth-generation (6G) technologies has introduced new challenges and opportunities for machine learning (ML) applications in Internet o

applicationsarxiv-cs-ai

UAF: A Unified Audio Front-end LLM for Full-Duplex Speech Interaction

arXiv:2604.19221v1 Announce Type: new Abstract: Full-duplex speech interaction, as the most natural and intuitive mode of human communication, is driving artificial intelligence toward more human-like

applicationsarxiv-cs-ai

Why virtualization modernization remains enterprise IT’s hardest leap

Enterprises are locked in a long-term standoff with legacy infrastructure, with growing demand for virtualization modernization that doesn’t require a full rebuild. SUSE S.A. has positioned itself to

Workspace agents are now available in research preview for ChatGPT Business, Enterprise, Edu, and Teachers plans. https://openai.com/busines…

OpenAI has released workspace agents in research preview for ChatGPT Business, Enterprise, Edu, and Teachers subscription plans. Workspace agents are AI assistants designed to work within organization

applicationsopenai--x

Wrote about this two years ago (and expanded on this in my book): https://www.oneusefulthing.org/p/setting-time-on-fire-and-the-temptation

Ethan Mollick discusses the concept of 'setting time on fire'—wasting time on unproductive activities—and explores the temptation to do so, drawing on arguments he originally published two years prior

Yes, the em-dashes were my very human choice. You can pry them from my five-fingered hands.

Ethan Mollick defends his personal stylistic choice to use em-dashes in his writing, humorously asserting ownership of this distinctly human writing preference. The post appears to be a response to su

A multimodal and temporal foundation model for virtual patient representations at healthcare system scale

arXiv:2604.18570v1 Announce Type: cross Abstract: Modern medicine generates vast multimodal data across siloed systems, yet no existing model integrates the full breadth and temporal depth of the clin

Active World-Model with 4D-informed Retrieval for Exploration and Awareness

arXiv:2604.16733v1 Announce Type: new Abstract: Physical awareness, especially in a large and dynamic environment, is shaped by sensing decisions that determine observability across space, time, and s

Adjustment of Cluster-Then-Predict Framework for Multiport Scatterer Load Prediction

arXiv:2602.08129v2 Announce Type: replace-cross Abstract: Predicting interdependent load values in multiport scatterers is challenging due to high dimensionality and complex dependence between impedan

AI-based Waste Mapping for Addressing Climate-Exacerbated Flood Risk

arXiv:2604.18151v1 Announce Type: new Abstract: Urban flooding is a growing climate change-related hazard in rapidly expanding African cities, where inadequate waste management often blocks drainage s

AlphaContext: An Evolutionary Tree-based Psychometric Context Generator for Creativity Assessment

arXiv:2604.18398v1 Announce Type: new Abstract: Creativity has become a core competence in the era of LLMs and human-AI collaboration, underpinning innovation in real-world problem solving. Crucially,

An LLM-Guided Query-Aware Inference System for GNN Models on Large Knowledge Graphs

arXiv:2603.04545v2 Announce Type: replace Abstract: Efficient inference for graph neural networks (GNNs) on large knowledge graphs (KGs) is essential for many real-world applications. GNN inference qu

AnchorRefine: Synergy-Manipulation Based on Trajectory Anchor and Residual Refinement for Vision-Language-Action Models

arXiv:2604.17787v1 Announce Type: new Abstract: Precision-critical manipulation requires both global trajectory organization and local execution correction, yet most vision-language-action (VLA) polic

applicationsarxiv-cs-ro

Anyhow, very impressive achievement, and a very solid model. Based on vibes, the gap seems around the same as it has always been between clo…

Ethan Mollick comments on an impressive AI model achievement, noting that the performance gap between models appears consistent with historical trends based on his assessment.

Appearance-free Action Recognition: Zero-shot Generalization in Humans and a Two-Pathway Model

arXiv:2604.16675v1 Announce Type: new Abstract: Action recognition is a fundamental ability for social species. Yet, its underlying computations are not well understood. Classical psychophysical studi

Are they lovers or friends? Evaluating LLMs' Social Reasoning in English and Korean Dialogues

arXiv:2510.19028v3 Announce Type: replace Abstract: As LLMs are increasingly deployed in real-world interactions, their social reasoning in interpersonal communication becomes critical. To explore the

Automatic Speech Recognition for Documenting Endangered Languages: Case Study of Ikema Miyakoan

arXiv:2603.26248v2 Announce Type: replace Abstract: Language endangerment poses a major challenge to linguistic diversity worldwide, and technological advances have opened new avenues for documentatio

AutoOR: Scalably Post-training LLMs to Autoformalize Operations Research Problems

arXiv:2604.16804v1 Announce Type: new Abstract: Optimization problems are central to decision-making in manufacturing, logistics, scheduling, and other industrial settings. Translating complicated des

BASIL: Bayesian Assessment of Sycophancy in LLMs

arXiv:2508.16846v5 Announce Type: replace-cross Abstract: Sycophancy (overly agreeable or flattering behavior) poses a fundamental challenge for human-AI collaboration, particularly in high-stakes dec

Boomi builds a role for agents and guardrails in the data-connected enterprise

Artificial intelligence and machine learning have been a central focus for Boomi LP over much of its 26 years as a data activation company. The organization began storing anonymized metadata from cust

Bridge-Centered Metapath Classification Using R-GCN-VGAE for Disaster-Resilient Maintenance Decisions

arXiv:2604.18399v1 Announce Type: new Abstract: Daily infrastructure management in preparation for disasters is critical for urban resilience. When bridges remain resilient against disaster-induced ex

Calibrated? Not for Everyone: How Sexual Orientation and Religious Markers Distort LLM Accuracy and Confidence in Medical QA

arXiv:2604.17316v1 Announce Type: new Abstract: Safe clinical deployment of Large Language Models (LLMs) requires not only high accuracy but also robust uncertainty calibration to ensure models defer

Can we generate portable representations for clinical time series data using LLMs?

arXiv:2603.23987v2 Announce Type: replace Abstract: Deploying clinical ML is slow and brittle: models that work at one hospital often degrade under distribution shifts at the next. In this work, we st

CanonSLR: Canonical-View Guided Multi-View Continuous Sign Language Recognition

arXiv:2604.18184v1 Announce Type: new Abstract: Continuous Sign Language Recognition (CSLR) has achieved remarkable progress in recent years; however, most existing methods are developed under single-

Can't pull off a sestina.

This post likely discusses the challenges or difficulties of writing a sestina, the complex poetic form requiring six stanzas with a specific pattern of end-word repetition. Ethan Mollick, known for d

CBR-to-SQL: Rethinking Retrieval-based Text-to-SQL using Case-based Reasoning in the Healthcare Domain

arXiv:2603.05569v2 Announce Type: replace-cross Abstract: Extracting insights from Electronic Health Record (EHR) databases often requires SQL expertise, creating a barrier for clinical decision-makin

ChatGPT Images 2.0 is available starting today to all ChatGPT and Codex users. Images with thinking are available to ChatGPT Plus, Pro, and …

ChatGPT Images 2.0 is available starting today to all ChatGPT and Codex users. Images with thinking are available to ChatGPT Plus, Pro, and Business users (Enterprise soon). On mobile, make sure you u

applicationsopenai--x

Chronax: A Jax Library for Univariate Statistical Forecasting and Conformal Inference

arXiv:2604.16719v1 Announce Type: new Abstract: Time-series forecasting is central to many scientific and industrial domains, such as energy systems, climate modeling, finance, and retail. While forec

Clarifai says it has deleted 3M OkCupid user photos and facial-recognition models trained on them after the US FTC settled with OkCupid over privacy violations (Jody Godoy/Reuters)

Jody Godoy / Reuters: Clarifai says it has deleted 3M OkCupid user photos and facial-recognition models trained on them after the US FTC settled with OkCupid over privacy violations — Artificial intel

applicationstechmeme

Clusterability-Based Assessment of Potentially Noisy Views for Multi-View Clustering

arXiv:2604.18024v1 Announce Type: new Abstract: In multi-view clustering, the quality of different views may vary substantially, and low-quality or degraded views can impair overall clustering perform

Common Corpus: The Largest Collection of Ethical Data for LLM Pre-Training

arXiv:2506.01732v2 Announce Type: replace Abstract: Large Language Models (LLMs) are pre-trained on large data from different sources and domains. These datasets often contain trillions of tokens, inc

Compositional Steering of Large Language Models with Steering Tokens

arXiv:2601.05062v2 Announce Type: replace Abstract: Deploying LLMs in real-world applications requires controllable output that satisfies multiple desiderata at the same time. While existing work exte

Conditional Evidence Reconstruction and Decomposition for Interpretable Multimodal Diagnosis

arXiv:2604.17030v1 Announce Type: new Abstract: Neurobiological and neurodegenerative diseases are inherently multifactorial, arising from coupled influences spanning genetic susceptibility, brain alt

Contraction and Hourglass Persistence for Learning on Graphs, Simplices, and Cells

arXiv:2604.17548v1 Announce Type: new Abstract: Persistent homology (PH) encodes global information, such as cycles, and is thus increasingly integrated into graph neural networks (GNNs). PH methods i

Covariance-Based Structural Equation Modeling in Small-Sample Settings with p>n

arXiv:2604.16894v1 Announce Type: new Abstract: Factor-based Structural Equation Modeling (SEM) relies on likelihood-based estimation assuming a nonsingular sample covariance matrix, which breaks down

Crossing the ‘production chasm’ is now enterprise AI’s defining test

Enterprise AI is moving fast, but not fast enough to outrun the governance blind spots that are quietly undermining the rush from controlled pilots to at-scale production. With one in five organizatio

Crowded in B-Space: Calibrating Shared Directions for LoRA Merging

arXiv:2604.16826v1 Announce Type: new Abstract: Merging separately trained LoRA adapters is a practical alternative to joint multi-task training, but it often hurts performance. Existing methods usual

CT Open: An Open-Access, Uncontaminated, Live Platform for the Open Challenge of Clinical Trial Outcome Prediction

arXiv:2604.16742v1 Announce Type: cross Abstract: Scientists have long sought to accurately predict outcomes of real-world events before they happen. Can AI systems do so more reliably? We study this

Decoding RWA Tokenized U.S. Treasuries: Functional Dissection and Address Role Inference

arXiv:2507.14808v3 Announce Type: replace-cross Abstract: Tokenized U.S. Treasuries have emerged as a prominent subclass of real-world assets (RWAs), offering cryptographically secured, yield-bearing