Artificial Intelligence
Follow Artificial Intelligence in the Factlen app and get an alert when there is news. Get the app
More in Artificial Intelligence
ExplainerDiffusion Architecture
Why Generative AI Uses U-Net Architecture to Predict Noise in Reverse Diffusion
Generative AI Researchers 45%Computer Vision Traditionalists 35%Transformer Advocates 20%
ExplainerCompute-in-Memory
How Processing-in-Memory Architectures Bypass the Von Neumann Bottleneck in AI Accelerators
Commercial Memory Vendors 40%Hardware Architecture Researchers 30%Software Ecosystem Analysts 30%
ExplainerFrontier AI
The 10^26 FLOP Threshold: How the US Government Monitors Frontier AI
National Security Advocates 35%AI Safety Researchers 35%Commercial AI Developers 29%
Model Evaluation
The Perplexity Metric: How the Geometric Mean of Inverse Probabilities Quantifies Language Model Performance
Applied AI Researchers 40%Information Theorists 30%NLP Practitioners 30%
ExplainerAI Architecture
The Four Components of a Retrieval-Augmented Generation (RAG) System: Indexing, Retrieval, Generation, and Evaluation
Enterprise AI Architects 40%AI Researchers 40%Evaluation Specialists 20%
Reinforcement Learning
Bellman Equation Calculates Optimal AI Strategies Through Recursive State Values
Applied AI Researchers 60%Theoretical Mathematicians 40%
ExplainerModel Architecture
How Mixture of Experts Routing Networks Decouple LLM Parameter Count From Compute Cost
Hardware Efficiency Advocates 40%Memory Constraint Skeptics 30%Algorithmic Researchers 30%
ExplainerSim-to-Real Transfer
How Domain Randomization Bridges the Reality Gap in AI Robotics
Simulation-First Advocates 45%Industrial Deployment Engineers 35%Editorial Synthesis 20%
Lunar Exploration AIOpen Source Release
IBM and NASA Open-Source Lunar Foundation Model Trained on Decades of Moon Mission Data
Open-Source Advocates 35%Commercial Space Industry 35%Scientific Researchers 30%
ExplainerSelf-Attention
How Parallel Self-Attention Overcame the Sequential Bottleneck of Recurrent Neural Networks
Parallelization Advocates 40%Efficiency Skeptics 30%Hardware Developers 30%
Open-Weights ModelsBenchmark Milestone
DeepSeek Releases V4.1 Flash Model, Bringing Near-Frontier Performance to Open-Source Ecosystem
Open-Source Advocates 40%Independent Evaluators 40%Enterprise Integrators 20%
Deep DiveInference Economics
The Quadratic Cost vs. Linear Benefit: How Model Size Increases Inference Latency and Cost
Architectural Reformers 40%Economic Pragmatists 40%Compute Scaling Advocates 20%
ExplainerTokenization
How Byte Pair Encoding Translates Raw Text into Model Input Tokens
Subword Efficiency Advocates 60%Multilingual Equity Critics 30%Algorithmic Formalists 10%
ExplainerAI Regulation
Mapping the Compliance Burden of the EU AI Act's Four-Tiered Risk Framework
European Regulators 40%Enterprise Compliance Officers 35%Industry Analysts 25%
AI Infrastructure
How NVLink Fusion Connects d-Matrix Inference Chips to Nvidia Server Racks
Specialized Silicon Designers 35%Infrastructure Ecosystems 35%Data Center Operators 30%
Model OrchestrationProduct Launch
Sakana AI Releases Fugu Max Orchestration Model, Reducing Multi-Agent Routing Costs by 60 Percent
Enterprise AI Adopters 40%AI Orchestration Developers 35%Venture & Market Analysts 25%
ExplainerStructured Generation
How Constrained Decoding Forces Large Language Models to Output Valid JSON
Application Developers 40%Model Researchers 35%Security Teams 25%
Siri OverhaulProduct Launch
Apple Launches Rebuilt Siri Powered by Google Gemini Across All Major Operating Systems
Consumer Tech Reviewers 40%Industry Strategists 35%Privacy Monitors 25%
ExplainerAdversarial Machine Learning
Fast Gradient Sign Method Reverses Loss Gradients to Force AI Misclassifications
AI Researchers 40%Security Practitioners 35%Systems Engineers 25%
ExplainerVision-Language-Action
How Vision-Language-Action Models Translate Pixels Into Robotic Movement
Autoregressive Advocates 35%Continuous Control Proponents 35%Open-Source Researchers 30%
ExplainerPhysical AI
Skild AI Unveils S1 Foundation Model, Enabling Robots to Learn 10-Minute Tasks From a Single Video
Robotics Researchers 40%Industrial Integrators 35%Open-Source Advocates 25%
ExplainerAI Governance
Govern, Map, Measure, Manage: The Four Core Functions of the NIST AI Risk Management Framework
Enterprise Security Leaders 40%AI Governance Advocates 35%Critical Infrastructure Operators 25%
Time-Series AIModel Release
IBM Releases Granite PatchTST-FM-r2 Zero-Shot Time Series Model Under Commercial License
Enterprise Data Teams 40%AI Architecture Researchers 35%Streaming Infrastructure Providers 25%
ExplainerAI Infrastructure
How KV Caching Solves the LLM Latency Bottleneck
Inference Engineers 40%Hardware Vendors 30%Application Developers 30%
ExplainerAI Watermarking
How Invisible AI Watermarks Actually Work: Inside the Push to Tag Synthetic Text
AI Safety Researchers 40%Privacy Advocates 30%Regulatory Compliance Teams 30%
ExplainerAgent Architecture
Translating the OODA Loop: How Autonomous AI Agents Observe, Orient, Decide, and Act
Automation Advocates 35%Human-in-the-Loop Proponents 35%Systemic Risk Analysts 30%
ExplainerModel Alignment
How Human Preferences Train the Reward Model to Align AI Behavior
Alignment Researchers 40%Enterprise AI Providers 40%Optimization Theorists 20%
ExplainerVideo Generation
How Temporal Attention Layers Enforce Frame Consistency in AI Video Generation
Architectural Innovators 40%Efficiency-First Developers 35%Analytical Synthesis 25%
ExplainerAI Hardware
Why BF16's 8-Bit Exponent Prevents Numerical Overflow in LLM Training
AI Research Scientists 45%Hardware Architects 35%Storage Infrastructure Engineers 20%
Model DistillationSecurity Warning
U.S. Intelligence Agencies Accuse Chinese Firms of Industrial-Scale AI Model Distillation
U.S. National Security Apparatus 40%Chinese Government 30%Global Tech Industry Analysts 30%
ExplainerNeural Network Architecture
Translating Raw Scores Into Words: How the Softmax Layer Drives Large Language Models
Machine Learning Practitioners 40%Hardware Engineers 30%AI Interpretability Researchers 30%
ExplainerReinforcement Learning
Using Epsilon Decay to Transition AI Agents From Exploration to Exploitation
Algorithmic Pragmatists 45%Advanced Exploration Advocates 35%Multi-Agent Theorists 20%
ExplainerModel Optimization
The L2 Penalty: How Weight Decay and Dropout Prevent Neural Network Overfitting
Deep Learning Practitioners 45%Theoretical Computer Scientists 35%Optimization Researchers 20%
ExplainerMechanistic Interpretability
Translating the Black Box: How the Logit Lens Maps AI Computations to Human-Readable Text
Mechanistic Interpretability Researchers 40%AI Safety Advocates 35%Model Developers 25%
Medical AIPolicy Decision
FDA TEMPO Pilot Allows Pre-Approval Use of Generative AI Medical Devices
Digital Health Developers 40%Regulatory Agencies 40%Patient Safety Advocates 20%
Sovereign AIFunding Milestone
Mistral Secures €3 Billion in Europe's Largest Tech Funding Round, Led by Samsung
European Tech Sovereignty Advocates 40%Enterprise Security Leaders 35%Semiconductor Manufacturers 25%
ExplainerModel Interpretability
How Concept Cones and In-Parameter Erasure Stop AI Models From Over-Refusing Safe Prompts
Safety Alignment Researchers 40%Model Interpretability Teams 35%Open-Source Advocates 25%
ExplainerPrompt Engineering
The 28.2% Accuracy Gain: How Chain-of-Thought Prompting Unlocks Reasoning in Large Language Models
AI Researchers 40%Enterprise Adopters 35%Model Evaluators 25%
Genomics AITool Release
DeepMind Unveils AlphaGenome Atlas to Predict the Effects of 9 Billion DNA Variants
Computational Biologists 40%Translational Researchers 40%Cloud Infrastructure Providers 20%
Search Algorithms
How Alpha-Beta Pruning Doubles the Search Depth of Adversarial AI
Classical AI Theorists 40%Modern Heuristic Developers 35%Multi-Agent Researchers 25%
PyTorch EcosystemStrategic Alliance
Alibaba Cloud, Cambricon, and Ant Group Join PyTorch Foundation Governing Board
Open-Source Foundation Leadership 40%Global AI Industry Observers 30%Regional Technology Analysts 30%
Research BriefAI Governance
The Four Dimensions of AI Risk: Technical, Societal, Operational, and Catastrophic
Regulatory and Compliance Bodies 38%National Security Analysts 38%Open-Source AI Developers 25%
ExplainerNeural Architectures
How Convolutional Filters and Pooling Layers Extract Hierarchical Features in Computer Vision
Architecture Designers 40%Hardware Optimizers 30%Interpretability Researchers 30%
ExplainerModel Training
How Adam's Adaptive Learning Rates Solve the Deep Learning Convergence Bottleneck
Adaptive Optimization Proponents 65%Generalization Purists 35%
Drug DiscoveryMilestone Approval
China Approves Mprosevir, the First Class 1 Innovative Drug Developed With AI Assistance
AI Integration Advocates 60%Regulatory Traditionalists 40%
ExplainerModel Training
How Data Annotation Separates Supervised, Unsupervised, and Semi-Supervised AI
AI Research Community 40%Data Annotation Industry 30%Enterprise Adopters 30%
Explore more topics
Per page