All Articles
18336 articles total
A Knowledge Theory of Capital:The Value of Natural and Artificial Intelligence
A theoretical exploration of how productive capacity in modern economies is shifting toward knowledge-bearing stock like software and AI models.
Sodium Bicarbonate for In-Hospital Cardiac Arrest – A Randomized Clinical Trial
A randomized clinical trial exploring the use of sodium bicarbonate for patients experiencing in-hospital cardiac arrest.
NASA picks Eric Schmidt’s rocket company for Mars mission, setting up a race with SpaceX
NASA has selected Eric Schmidt's rocket company, Relativity Space, for a Mars mission, creating a competitive dynamic with SpaceX.
QSignAI: Quantum-Randomness-Seeded Identity Signatures at the Intersection of AI for Science and Science for AI
Introduction of QSignAI, a platform that integrates quantum-randomness-seeded identity signatures into an AI-driven social platform.
Dynamic In-Group Persona Generation for Enhancing Human-AI Rapport
Research on using synthetic in-group personas in LLM chatbots to improve human-AI rapport and user engagement.
From Memorization to Creation: Evaluating the Cognitive Depth of LLM-Generated Educational Questions
An evaluation of LLMs' ability to generate educational questions based on Bloom's Taxonomy, introducing metrics like CogShift to measure cognitive depth.
Examining Human-Like Behaviors in LLMs: A Multi-Dimensional Analysis of Model Behaviors, User Factors, and System Prompts
A multi-dimensional analysis of human-like behaviors in LLMs and how system prompting can be used to control them for responsible design.
Caring Without Feeling: Affective Dynamics as the Control Layer of Human-AI Agent Collaboration
A review proposing a framework that treats affective dynamics in AI agents as a coordination layer for trust and delegation in human-AI collaboration.
How Well Do Large Language Models Capture Human Personality?
Study on 'persona manifold collapse', where increasing the complexity of persona prompts in LLMs actually reduces behavioral diversity and simulation fidelity.
Simulating Hate Speech Cascades with Multi-LLM Agents: Empirical Grounding, Modeling Fidelity, and Intervention Strategies
Use of multi-LLM agents to simulate hate speech propagation on social platforms like Bluesky to test moderation intervention strategies.
Synthetic Resonance: A Framework for Growth-Oriented Human-AI Relationships
Proposed conceptual framework 'synthetic resonance' to describe meaningful human-AI relationships without requiring anthropomorphism or subjective experience.
Apple boss Tim Cook says prices to rise due to memory chip costs
Apple CEO Tim Cook indicates that product prices may increase due to the rising costs of memory chips.
Towards an Agent-First Web: Redesigning the Web for AI Agents
A research paper proposing a fundamental redesign of the web to better accommodate AI agents, introducing the Agent Text Markup Language (ATML) and new economic models.
Analysing drivers and interdependencies in European electricity markets using XAI
An analysis of European electricity markets using deep neural networks combined with explainable AI (XAI) and SHAP to understand price drivers.
Human-AI Coevolution Dynamics: A Formal Theory of Social Intelligence Emergence Through Long-Term Interaction
Introduction of the Human-AI Coevolution Dynamics Framework (HACD-H) to model the emergence of social intelligence through long-term interaction.
Beyond Safe Data: Pretraining-Stage Alignment with Regular Safety Reflection
A proposal for 'Safety Reflection Pretraining' to integrate self-monitoring into LLMs during the pretraining stage, reducing unsafe behaviors generalized from safe data.
User as Engram: Internalizing Per-User Memory as Local Parametric Edits
A new method called 'User as Engram' that stores per-user memory as local parametric edits to a hash-keyed memory table, significantly reducing memory footprint compared to LoRA.
TxBench-PP: Analyzing AI Agent Performance on Small-Molecule Preclinical Pharmacology
The introduction of TxBench-PP, a verifiable benchmark for evaluating AI agents' performance in preclinical pharmacology and drug discovery.
X+Slides: Benchmarking Audience-Conditioned Slide Generation
X+Slides is introduced as a benchmark for audience-conditioned slide generation, measuring how well LLMs tailor content to specific target audiences.
NeSyCat Torch: A Differentiable Tensor Implementation of Categorical Semantics for Neurosymbolic Learning
Introduction of NeSyCat Torch, a differentiable tensor implementation for neurosymbolic learning that subsumes multiple inductive definitions of truth.