1854 stories tagged AI · page 22 of 78
Researchers introduce interventional grounding audits, a black-box method that substitutes predicates in LLM chain-of-thought reasoning to test whether conclusions genuinely depend on stated premises, evaluated on the ProntoQA benchmark.
Researchers introduced Safe-Psych, a sequential benchmark that evaluates how well large language models handle evolving diagnostic uncertainty in clinical psychiatry. Unlike existing medical benchmarks that assume complete information upfront, Safe-Psych tests whether models request clarification or abstain when data is insufficient.
Researchers introduced OriginBlame, a record- and token-level data provenance system that lets data contributors precisely remove their specific content from AI training datasets without over-deleting unrelated data.
Researchers propose a meta-learning framework for RLHF and DPO that transfers preference data across languages, enabling effective alignment of large language models in low-resource languages with minimal data.
FixItFlow is an AI system that automatically generates troubleshooting guides from historical cloud incident data using large language models, aiming to replace manual, labor-intensive documentation and improve incident response consistency.
A VentureBeat survey of 107 enterprises reveals a critical AI compute gap: companies are accelerating AI infrastructure spending faster than they can track or measure its economics. Most rely on hyperscalers and model-provider APIs, but the next dollar is aimed at specialized compute few use today. A majority plan to switch or add providers within the year. Buying decisions hinge on integration and total cost of ownership rather than token price — fortunate, because most can't see their unit economics clearly, with GPUs sitting at half utilization.
A VentureBeat survey of 101 enterprises reveals that AI agents are producing confident but wrong answers due to missing or inconsistent context. The core issue is trust, not retrieval — and most organizations are still building the governed semantic layer needed to fix it.
A study of 157 enterprises reveals that half have shipped AI agents that passed internal evaluations but failed in production. Only 5% fully trust automated evaluations, yet two-thirds are moving toward fully automated deployments without human oversight. The core issue is not test coverage but a reality-alignment gap between evaluations and real-world outcomes.
A VentureBeat Pulse Research report of 101 enterprises finds that most deployed 'AI agents' are actually simple chatbot wrappers, not true autonomous agents. Anthropic's Claude leads agent orchestration adoption due to reliable multi-step execution, while enterprises deliberately build hybrid control planes to avoid vendor lock-in. Real-time fiscal control over token burn remains rare.
The Verge AI reports from a major law enforcement conference in Fort Worth, Texas, where AI companies are aggressively marketing predictive policing, facial recognition, and surveillance tools to police departments. The story examines the privacy, bias, and accountability risks these technologies pose to communities.
1Password has launched a browser integration for Claude that allows the Anthropic chatbot to access stored security credentials like usernames and passwords. Users can authorize Claude to complete multi-step tasks like booking travel and managing online accounts on their behalf without manually inputting login details.
Anthropic and Blackstone launch Ode, a startup that embeds forward-deployed AI engineers inside enterprises to accelerate real-world AI adoption, betting implementation is the next trillion-dollar opportunity.
Researchers propose a modular Context-Augmented Prompting framework that combines small language models (SLMs) with graph neural networks (GNNs) to overcome structural blindness in molecular property prediction. The method provides predictive hints and explanatory subgraphs at inference time, improving accuracy for drug discovery and materials science.
OpenAI outlines a 'reverse federalism' approach to AI governance, where state-level laws help build a national framework for safe, democratic AI. This strategy aims to make regulations more responsive to local needs while fostering innovation.
A new study introduces MAGE (Memory-Augmented Goal-directed Prompt Evolution), a framework that reveals a previously unknown phenomenon in how different components of AI prompt optimization interact, exposing critical trade-offs between stability and performance.
A new graph-based AI framework from ArXiv cs.CL detects and analyzes disinformation narratives on Telegram by clustering related claims and modeling their diffusion across channels. The tool targets coordinated misinformation campaigns, particularly in the Russia-Ukraine conflict.
OpenAI researcher Miles Wang is in funding talks for a new AI drug discovery startup, with discussions valuing the company at $2 billion. The move underscores surging investor interest in applying AI to accelerate life-saving drug development.
New York has become the first state to temporarily stop approving new large data centers, as Governor Kathy Hochul warns that the AI-driven building boom must not come at the expense of higher electricity costs, water supplies, or local control.
A new arXiv paper shows that point-in-time language models—trained only on data available up to each calendar date—can now perform nearly as well as unrestricted models, eliminating lookahead bias that previously compromised financial backtests and causal inference.
Researchers propose TAKE (Trajectory-Aware Knowledge Estimation), a text dataset distillation framework that reduces corpora to as little as 0.1% of their original size while preserving downstream task fidelity, using influence functions to select the most valuable samples.
Researchers fine-tuned LLaMA 3 (8B) with LoRA and 4-bit quantization to replace costly cross-encoders in RAG pipelines, achieving efficient real-time reranking for AI assistants and search tools.
Researchers at CNPO developed G-SHARE, a guideline-based structured reasoning framework that diagnoses human-factor events in nuclear power plants with greater reliability and logical consistency than existing AI approaches.
Hugging Face introduced VoiceEQ, a tool that evaluates how natural and human-like AI-generated voices sound by comparing them to real human speech. This helps developers build more convincing voice assistants, audiobooks, and customer service bots.
A new report from PBS NewsHour warns that Google's AI-powered search features pose an 'unacceptable risk' to children, citing exposure to inappropriate content and privacy concerns. Experts call for stronger safeguards and better parental controls.