AI/ML arXiv cs.AI

Black-Box Inference of LLM Architectural Properties with Restrictive API Access

The NightVision attack demonstrates that LLM architectural properties like depth and parameter count can be inferred even from restrictive black-box APIs.

Other Hacker News

Gun Mistakes in Fiction Writing: Handgun Edition

A discussion on common mistakes writers make when depicting handguns in fiction.

Software Engineering Hacker News

Commodore 64 Basic for PostgreSQL

A technical curiosity implementing Commodore 64 Basic functionality for PostgreSQL.

Hardware/Chips The Verge

A behind-the-scenes look at Midjourney’s medical scanner leaves many questions unanswered

Midjourney reveals a medical ultrasound scanner utilizing hacked ultrasound probes and Raspberry Pis, though proof of effectiveness is limited.

AI/ML arXiv cs.AI

AI Assistance for Human Review of Default Judgments

Researchers developed a 'Default Assistant' using LLMs to help US courts review default judgments more accurately and quickly.

AI/ML arXiv cs.AI

Artificial Intelligence-Enabled Accounting Information Systems and Fraud Detection in Nigeria's Financial Services Sector: The Moderating Role of Natural Language Processing

A study on how AI-enabled accounting systems and NLP improve fraud detection in Nigeria's financial services sector.

Hardware/Chips arXiv cs.AI

The Rising Unsustainability of AI Graphics Cards Production

An analysis of the escalating environmental costs and sustainability issues associated with producing AI graphics cards.

AI/ML arXiv cs.AI

Benchmarking Federated Learning and Knowledge Distillation for Point Cloud Classification

A benchmark evaluating federated learning and knowledge distillation for 3D point cloud classification on edge hardware.

AI/ML arXiv cs.AI

Domain Knowledge Based Temporal-Spatial Graph Convolution Network for ECG Recognition

Introduction of a domain knowledge-based graph convolution network to improve ECG recognition and interpretability in healthcare.

AI/ML arXiv cs.AI

Scaling Laws for Grid-Based Approximate Nearest Neighbor Search in High Dimensions

A study on scaling laws for grid-based approximate nearest neighbor search, showing advantages in high-dimensional settings.

AI/ML arXiv cs.AI

Adaptive Companionship for Group-Following Robots: Handling Dynamically Changing Group Formations

A method for group-following robots using VLMs and MPPI controllers to maintain natural social distances and formations.

AI/ML arXiv cs.AI

Prompt Framing Distorts Count-Based Evaluation of LLM Error Detection: Evidence from Numeric Anchoring

Researchers identify 'F1 Inflation,' where prompt framing can artificially boost LLM error-detection scores without improving actual localization accuracy.

AI/ML arXiv cs.AI

Mapping Text to Multiplex Graph: Prompt Compression as L\'evy Walk-Guided Graph Pruning

Introducing RAGP, a prompt compression method that treats text as a multiplex graph and uses Lévy walks for efficient redundancy-aware pruning.

AI/ML arXiv cs.AI

ExPerT: Personalizing LLM Responses to Users' Domain Expertise via Query-Wise Semantic and Keystroke Behavioral Cues

ExPerT is a framework that personalizes LLM responses by inferring a user's domain expertise through both semantic query text and keystroke behavioral cues.

AI/ML arXiv cs.AI

Office Comprehension Benchmark

The Office Comprehension Bench (OCB) is a new public benchmark for evaluating LLM performance on native Word, Excel, and PowerPoint file formats.

AI/ML arXiv cs.AI

LLMs as Teaching Assistants for Mathematics Exam Grading: Reliability, and Practical Usability

A study evaluates LLMs as teaching assistants for grading discrete mathematics exams, finding that 'liberal' partial-credit prompting improves agreement with human graders.

AI/ML arXiv cs.AI

A Practice Auditing Framework for Large Language Model Use: Collective Empiricism, Pseudo-Rational Cognition, and Governance of AI-Generated Content

The paper proposes a practice auditing framework to govern AI-generated content and mitigate risks like 'pseudo-rational cognition' and memory pollution.

AI/ML arXiv cs.AI

Structuring the Space of Sociotechnical Alignment

This research argues for a more systematic, human-centered framework to specify and evaluate 'social desirability' in sociotechnical AI alignment.

AI/ML arXiv cs.AI

Collaborative Disagreement Resolution for Scalable Oversight

Researchers propose 'disagreement resolution,' a collaborative truth-seeking paradigm for AI oversight that outperforms adversarial debate in judging accuracy.

AI/ML arXiv cs.AI

How Indian Dermatologists are Utilizing Artificial Intelligence for Clinical Practice and Workflow Management: A Nationwide Survey with a Special Focus on atopic dermatitis

A survey of Indian dermatologists reveals that general-purpose AI is mainly used for administrative tasks rather than specialized clinical diagnostic workflows.