Latest Articles

DeepMind's Latest Research: Agents Are World Models!
AI ResearchDeepMindMachine LearningArtificial IntelligenceWorld Models
...
Cursor 1.0 Debuts: BugBot Intelligent Correction, Background Agents Fully Open
Software DevelopmentCode EditorProduct UpdateDebugging ToolsAI Assistant
...
Fellou 2.0 Fully Upgraded, Success Rate from 31% to 80%! AI Enables Batch Production
AI BrowserAI AutomationSoftware UpdateAgentic AIProductivity Tools
...
A Deep Dive: Measuring the Human Brain, Consciousness, and AI Through the 'Nature of Time'?
NeurosciencePhilosophy of TimePhysicsConsciousnessArtificial Intelligence
...
Closer to AGI? Running Google's AlphaEvolve and UBC's DGM for Just 0.31 Yuan?
AI Self-ImprovementAlgorithmic OptimizationAI ResearchArtificial General IntelligenceLarge Language Models
...
Terry Tao Reveals: AlphaEvolve Breaks 18-Year Unsolved Problem Three Times in a Month, Completely Rewriting Rules of Mathematical Research
MathematicsArtificial IntelligenceAlphaEvolveHuman-AI CollaborationScientific DiscoveryTerry TaoCombinatorics
...
Google Open-Sources Gemini-Level AI Research Capabilities: Is Deep Research Becoming Commoditized?
Artificial IntelligenceAI AgentMachine LearningLangGraphGoogle GeminiOpen Source
...
The Smarter the Model, the Less Obedient? MathIF Benchmark Reveals AI Obedience Vulnerabilities
Artificial IntelligenceLarge Language ModelsAI ResearchMathematical ReasoningInstruction Following
...
First Genomic Reasoning AI Emerges! Accuracy Soars to 97%, Revolutionizing Genomics Research
Genomic AIArtificial IntelligenceExplainable AILarge Language ModelsGenomics
...
Process Supervision > Outcome Supervision! Huawei City University Reconstructs RAG Inference Training, 5k Samples Outperform 90k Model
RAG TrainingProcess SupervisionAI ResearchLarge Language ModelsReinforcement Learning
...
Reviewing the Progress of RL-Reasoning
Reinforcement LearningLarge Language ModelsNatural Language ProcessingMachine LearningReasoning
...
OPA-DPO: An Efficient Solution for the Hallucination Problem in Multimodal Large Models
Artificial IntelligenceMachine LearningMultimodal AIDirect Preference OptimizationComputer VisionLarge Language Models
...
AI Learns Reasoning Solely by "Confidence": Zhejiang University Alumnus Replicates DeepSeek's Long Chain-of-Thought Emergence, Reinforcement Learning Needs No External Reward Signals
Artificial IntelligenceLarge Language ModelsNatural Language ProcessingMachine LearningReinforcement Learning
...
Peking University Alumna Lilian Weng's Latest Blog Post: Why We Think
AI ReasoningLarge Language ModelsTest-Time ComputationReinforcement LearningChain-of-Thought
...
Microsoft and Others Propose 'Chain-of-Model' New Paradigm, Comparable to Transformer Performance with Better Scalability and Flexibility
AI ArchitectureLarge Language ModelsDeep Learning ResearchModel ScalingTransformer Models
...
将心智理论视为思维的思维语言：融合了贝叶斯网络/因果文法模型和编程模式模型的优点 DSL
Theory of MindProgramming LanguagesComputational ModelsArtificial IntelligenceCognitive Science
...
Building an AI Software Engineer in Two Years! OpenAI Codex Authors Unveil a New Paradigm for Human-AI Pair Programming
Artificial IntelligenceSoftware DevelopmentHuman-AI CollaborationProgrammingAI Agents
...
Will the Vision of LSTM's Father from 22 Years Ago Come True? AI 'Self-Evolution' Papers Concentratedly Released in One Week, Is a New Trend Emerging?
AI Self-ImprovementLarge Language ModelsGödel MachineMulti-modal AIReinforcement Learning
...
Deep Dive | Granola Founder (AI Note-Taking Valued at $250M): AI Habits are Reshaping Our Intuition; AI's Role Should Be to Augment, Not Replace Humans
Artificial IntelligenceAI ToolsStartup InnovationProductivity SoftwareFuture of Work
...
AI Math Ability Skyrockets 100%, Self-Evolution Nears RL Limits! CMU's New Work Overturns Perceptions
Artificial IntelligenceLarge Language ModelsMathematical ReasoningSelf-ImprovementReinforcement Learning
...