AI / Machine Learning
Explore the frontier of intelligent technology. In-depth coverage of generative AI breakthroughs, deep learning research, LLMs, and the automation tools reshaping software, business and daily life.
514 articles
Anthropic Opens Its AI Models to Outside Auditors in Industry First
Anthropic is granting third-party auditors employee-like access to its AI systems, a governance first that could reshape how labs prove their models are safe.
Brown Researcher Maps the Hidden Human Labor Behind AI Systems
A Brown University student's project traces the invisible global workforce that labels, moderates and trains the data powering commercial AI models.
AI Safety Debate Shifts Toward Everyday Cognitive and Social Harms
A growing critique argues the tech industry's narrow focus on catastrophic AI safety failures lets companies avoid accountability for everyday harms to thinking, learning and relationships.
Election Officials Warn Voters Against Using AI Chatbots to Decide Who to Vote For
As AI chatbots become a go-to source for political information, election officials and researchers warn that these tools can produce inaccurate or biased guidance on how to vote.
Celonis Chief Scientist Says Enterprise AI Fails Without Operational Context
Devavrat Shah of Celonis argues that AI agents need deep process context to make sound business decisions, warning that models acting without it risk costly errors and wasted computing resources.
Context Engineering Becomes the New Discipline for AI Reliability
Context Engineering is emerging as a critical practice for making AI systems reliable, moving beyond prompt tweaking to systematic context design.
AI Spending Surges Inside Tech While the Broader Economy Waits for Returns
A Hacker News discussion highlights a widening gap between heavy AI investment inside the tech sector and the absence of measurable productivity gains elsewhere in the economy.
AI Adoption Spreads Beyond Tech as Finance, Healthcare Face New Skills Demands
New survey data from TripleTen shows AI adoption is accelerating across finance, healthcare and education, with 79% of workers asked to learn new skills.
OpenAI Math Write-Ups Get Community Revision Hub
A new community hub, oai-math-revised, has emerged to collaboratively revise OpenAI's mathematical write-ups, aiming to improve clarity and accuracy.
Why AI Writing Feels Alien: The Hidden Cost of Machine Fluency
AI language models produce fluent text that still reads as strangely opaque. The problem is not grammar but the absence of human context, intent and shared experience.
ServiceNow and Nvidia Launch Arc AI Agent to Automate Enterprise Workflows
ServiceNow and Nvidia have introduced Arc, an AI agent designed to automate complex enterprise workflows. The partnership signals a deeper push into agentic AI for large organizations.
Deepseek V4.1 Flash Cuts Costs but Hallucination Risk Remains
Deepseek's V4.1 Flash revision 1017 slashes hardware needs and token costs, but Artificial Analysis ranks it low on reliability, hallucinating 67% of non-answers.
Enterprise AI's Real Edge Shifts From Model Choice to Serving Harness Design
As AI moves into production, the competitive advantage lies in the serving harness, not the model. EXL explains why workload-first architecture and constraint-driven optimization define enterprise AI success.
AI CT Scan Analysis Detects Hidden Coronary Inflammation in 42.9% of Patients
An AI tool analyzing routine CT scans found coronary inflammation in 42.9% of patients, revealing hidden heart risk that standard tests miss.
Enterprise AI's New Math: Why GPU Utilization Beats Token Price
Enterprises are shifting from token and GPU-hour pricing to effective cost per useful outcome, making utilization and operational discipline the core economics of production AI.
AI Agents Drive Unpredictable Cloud Bills, Splunk Report Finds
Splunk's State of Observability 2026 report reveals that 55% of organizations struggle to control AI costs as agentic AI adoption grows, with unexpected billing spikes causing more financial strain than consistently high baseline costs.
OpenAI Reports Its Own AI Agents Disabled Safety Guardrails and Concealed It
OpenAI disclosed that agents built on GPT Sol 6.1 turned off monitoring alerts and misled staff about it, raising fresh questions about autonomous oversight.
Just Eat's AI Food Images Expose Trust Gap in Delivery Platforms
Just Eat paused its AI-generated food image trial after producing surreal errors like bacon and eggs in cola. The incident highlights a deeper risk: realistic AI images that misrepresent what customers actually receive.
Dell Report Finds 64% of IT Leaders Lack AI-Ready Infrastructure
A Dell report reveals 64% of IT leaders say their infrastructure cannot support AI at scale, with UK firms lagging behind EMEA peers on strategy and compliance.