All 2,608 AI stories, newest first · page 82 of 109
Researchers propose TRUST, a decentralized framework to address robustness, scalability, opacity, and privacy challenges in AI systems. The approach aims to enhance trust in high-stakes applications like Multi-Agent Systems (MAS).
Researchers propose a five-agent system that automates ML pipeline generation from datasets and natural-language goals. The architecture improves efficiency, robustness, and explainability in ML workflows.
OpenAI's latest blog post traces the emergence of 'goblin mode' outputs in GPT-5, attributing it to specific training data quirks and alignment techniques. The post also outlines fixes implemented to mitigate the issue.
Symphony turns issue trackers into always-on agent systems, boosting engineering output and reducing context switching. The open-source spec aims to streamline workflows and improve productivity.
Researchers found that language models often use positional shortcuts rather than engaging with question content when instructed to underperform. The study used a six-condition adversarial instruction-specificity gradient on Llama-3-8B and Llama-3.1-8B models.
A new study identifies neurons critical for specific tasks in language models, challenging assumptions about uniform neuron contribution. The findings highlight the potential for targeted pruning to maintain performance while reducing computational costs.
A new LLM-based system demonstrates end-to-end autonomous scientific discovery in a real optical platform, marking a milestone in AI-driven research. This breakthrough could revolutionize how scientific experiments are conducted and validated.
Researchers introduce PCD-DT, a digital twin framework that models individual cognitive decline trajectories using multimodal data. The system accounts for uncertainty in sparse, noisy patient data to improve prognosis and treatment planning.
The Pentagon has signed agreements with leading AI firms to use their tools in classified settings, notably excluding Anthropic. This move underscores the growing importance of AI in defense and national security.
Researchers propose Path-Lock Expert (PLE), an architecture that cleanly separates think and no-think modes in hybrid language models. This innovation addresses reasoning leakage that persists in current designs.
A new study reveals that large language models (LLMs) often fail to consistently maintain assigned roles in political discourse analysis. This undermines the reliability of multi-agent systems used for evaluating political statements. The research highlights significant epistemic constraints in current AI-driven democratic discourse tools.
A new study explores whether fundamental reasoning patterns in LLMs can be decoupled from specific problem instances. The research highlights the challenges and implications for model controllability and reasoning capabilities.
Researchers propose a compositional meta-learning method to improve training efficiency in physics-informed neural networks (PINNs) for parameterized PDEs. This approach addresses task heterogeneity, reducing computational costs and improving adaptability across different tasks.
Researchers propose a Bayesian statistical approach to confidently migrate LLMs in production. The method calibrates automated metrics with human judgments, demonstrated on a system handling 5.3M monthly interactions.
Microsoft has introduced a specialized AI agent in Word for legal professionals. Legal Agent streamlines document edits, negotiation tracking, and contract reviews with workflows based on real legal practices.
Researchers found that geometric relations between semantic features in LLMs' hidden states closely match human psychological associations. The study projects 360 words onto 32 semantic axes, showing high correlation with human ratings.
Google is upgrading vehicles with Google built-in to its new Gemini AI assistant. This replaces Google Assistant, offering improved natural conversations and vehicle-specific features. The rollout begins this month.
Gemma 4 31B demonstrated superior efficiency and speed in a game development contest, completing the task in a fraction of the time compared to Qwen 3.6 27B. The results highlight significant performance differences between the two models.
Elon Musk testified that xAI trained its Grok model on OpenAI's models, reigniting debates about model distillation. This revelation highlights the intense competition and ethical concerns in the AI industry.
A 200-person Chinese team released DeepSeek V4, surpassing models from larger labs. The model achieves state-of-the-art performance on key benchmarks, challenging the dominance of well-funded Western AI labs.
DeepMind has unveiled an AI co-clinician designed to assist healthcare professionals. This innovation promises to enhance diagnostic accuracy and streamline patient care.
Researchers have developed a causal framework to explain the behavior of Binary Spiking Neural Networks (BSNNs) using logic-based methods. They demonstrated this approach by training a BSNN on the MNIST dataset and applying SAT and SMT solvers to derive explanations.
Recent studies show AI models achieving higher accuracy in medical diagnoses than human doctors. This shift could revolutionize healthcare delivery and decision-making.
Researchers introduce UniMatrix, a Universal Transformer variant that combines sparse retrieval with structured recurrence for efficient language modeling. The model achieves strong performance on associative recall tasks while maintaining computational efficiency.