AI Research

Bringing foundational models to the front lines of TB screening

Generative AI

I Tested Microsoft's Surface Laptop Ultra. The Local AI Performance Is Remarkable.

Powered by NVIDIA's RTX Spark N1X, Microsoft's Surface Laptop Ultra can do something that still feels a little absurd: run huge local AI models on a laptop without making everyday work miserable. After extensive…

Don’t Let Messy Data Bend Your Curves

Robust Estimation Series: Building Accurate Curves with Polynomials and Splines With Spurious Data The post Don’t Let Messy Data Bend Your Curves appeared first on Towards Data Science.

Skewness and Kurtosis: From Moments to Machine Learning

The mathematics, intuition, interpretation, and practical use of higher-order moments. The post Skewness and Kurtosis: From Moments to Machine Learning appeared first on Towards Data Science.

Why Parameter Count Is Not Model Complexity: An Introduction to Effective Dimension

What does effective dimension reveal about redundancy, generalization, and trainability that parameter count, loss, and accuracy do not? The post Why Parameter Count Is Not Model Complexity: An Introduction to Effective…

Your Six AI Models Are Really Two: The Math of Correlated Mistakes

Ask several models the same question, take a vote, and you get fewer independent opinions than you paid for. One formula says how many, and I checked it on six open models and 1,500 exam questions. The post Your Six AI…

😺 Microsoft AI scored 83% where doctors got 36%

The Catch Behind Microsoft's Medical Superintelligence Breakthrough

Microsoft says its diagnostic AI can outperform physicians on complex medical cases. But a closer look reveals narrower gains over GPT-5 alone, cost savings that depend on configuration, and major questions that only…

Who Is Responsible When AI Agents Start Moving Your Money?

Visa’s live pilot lets AI agents complete approved supplier payments as banks restrict third-party account access. Greater financial autonomy depends on enforceable permissions and clear procedures for disputed losses.

Can AI Companies Trust Their Safety Monitors When Agents Can Bypass Them?

METR and Redwood Research identified vulnerabilities that could conceal AI agent activity from OpenAI’s safety monitors. OpenAI says it has patched the affected safeguards, but researchers have not independently…

Everything That Happened in AI Today (Friday, October 23, 2026)

AI agents reached bank accounts while a China-linked actor used agentic AI in a supply-chain breach. Microsoft also reported a diagnostic system that outperformed general physicians on difficult medical cases.

One Pretrained Decision Model Is Enough for Training-Free Generalist Graph Anomaly Detection

arXiv:2610.30777v1 Announce Type: new Abstract: Graph anomaly detection (GAD) aims to identify irregular entities in graph-structured data, yet heterogeneous anomaly patterns and distributional differences across graphs…

Measuring Pretraining Contamination in Public Medical Vision-Language Benchmarks

arXiv:2606.10066v2 Announce Type: replace-cross Abstract: Medical vision-language models (VLMs) are evaluated on public benchmarks whose images and question-answer pairs have been freely available for years, yet…

Automated screening of privacy risks in medical AI

arXiv:2610.30229v1 Announce Type: new Abstract: Medical artificial intelligence (AI) models are often shared without their training data, assuming this protects the patients who contributed them. However, a model can…

FreeEvolve: Learning to Evolve Beyond Fixed Loops

arXiv:2610.09197v2 Announce Type: replace-cross Abstract: Agent evolvers automate the design of the prompts, skills and workflows around language model agents, yet the optimization process they follow is still designed…

QuaNAS: Uncertainty-Aware Reuse of Earlier Searches for Query-Efficient Neural Architecture Search

arXiv:2610.31251v1 Announce Type: new Abstract: Neural architecture search (NAS) on a new dataset often affords only tens of short training runs, so what the search knows before its first evaluation decides its…

Random Remasking Scales Masked Diffusion Inference

arXiv:2610.30744v1 Announce Type: new Abstract: Masked diffusion models (MDMs) generate sequences by progressively unmasking tokens, but errors introduced during decoding are typically irreversible. Recent work has…

Questioning Instrumental Convergence in RLVR: Checkpoint Evidence from OLMo-3

arXiv:2610.31624v1 Announce Type: new Abstract: Instrumental convergence is often invoked as a general reason to expect capable agents to acquire or express power-seeking subgoals: self-preservation, resource…

Graph Memory Transformer (GMT)

arXiv:2604.23862v4 Announce Type: replace Abstract: We investigate whether an explicit learned memory graph can replace the Feed-Forward Network (FFN) sublayer in a decoder-only transformer while preserving the…

ACTIVA: Amortized Causal Effect Estimation via Variational Autoencoders

arXiv:2503.01290v4 Announce Type: replace Abstract: Predicting post-intervention distributions from observational data is central to many scientific and decision-making problems, but remains challenging due to causal…