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
Low-Power AI Emerges as a Strategic Response to Surging Energy Demands
Low-power AI systems are gaining traction as data center energy use surges. These systems process data locally, reducing reliance on energy-hungry cloud infrastructure.
ChatGPT as a Personal Librarian: AI Recommendations Beyond the Algorithm
A user tested ChatGPT by feeding it their entire reading history, yielding book recommendations that felt personal rather than generic. The experiment reveals new possibilities for AI-driven personalization.
Why You Should Never Trust an LLM's Confidence Score
LLMs frequently express high confidence even when wrong. Their self-reported scores are unreliable and should not guide decision-making. Experts urge external validation for critical use cases.
Enterprise AI Faces a Reliability Crisis as Frontier Providers Falter
Frontier AI labs like Anthropic and OpenAI show policy instability, pushing enterprises toward flexible, open-source alternatives for long-term resilience.
Using Open Models: The Surprising Appeal of Transparent AI Frameworks
Developers increasingly favor open-weight AI models over proprietary options. The shift offers control, customization and community-driven innovation.
Brain Wave Data Emerges as Next Frontier for Physical AI Training
Frontier physical AI models may soon need brain wave readings for training. The shift from video to neural data could accelerate robotics but raises privacy concerns.
Token Efficiency and Modular Libraries Reshape AI Development
AI development shifts toward token-efficient models and modular libraries. Smaller, composable systems challenge monolithic architectures and promise lower costs.
Cisco Outshift Unveils 'Internet of Cognition' for AI Agent Collaboration
Cisco's Outshift introduces a semantic layer to enable AI agents to coordinate and reason collectively. The open-source project AGNTCY under Linux Foundation aims to solve multi-agent coordination failures.
Kimi AI Overwhelms Users With Demand Exposing AI Infrastructure Limits
China's newest ChatGPT rival Kimi hit instant demand on iOS, but users faced capacity errors. The incident underscores a critical industry-wide challenge: scaling infrastructure to match model intelligence.
Rising AI Cloud Costs Push Enterprises Toward Local Processing With AI PCs
AI software costs are rising as vendors shift to consumption pricing. AI PCs with TOPS NPUs let enterprises run local tasks, cutting cloud bills. The cloud remains key for heavy workloads.
Consciousness in Computers May Be a Hardware Property, Not a Software One
A new argument suggests that if digital computers are conscious, their consciousness resides at the hardware level, not in software. This challenges AI research and raises questions about machine rights.
Hidden Prompt Trap Reveals How Students Rely on AI for Exam Answers
A professor embedded invisible instructions in a test prompt to catch students using AI. 32 students submitted the required phrase, leading to disciplinary action.
Finance AI Enters a New Phase: Execution Over Experimentation
CFOs shift from pilot projects to embedded AI, demanding governance and data readiness. Early use cases show tangible gains in close and compliance.
AI Transforms Drug Discovery at AstraZeneca With Autonomous Lab Strategy
AI is now integral to biologic drug design at AstraZeneca, cutting cycle times and enabling new classes of medicines. The company's data-driven approach combines computational models with a new autonomous lab.
Art Museums Turn to Data Analytics to Reshape Visitor Engagement
Museums are using visitor data and analytics to personalize art experiences and optimize exhibition design. Young patrons like Giovanni Kanter represent a new wave of data-informed museum engagement.
Anthropic's J-Lens Reveals Claude's Hidden Thought Process Before It Speaks
Anthropic developed the J-lens, revealing a hidden space inside Claude where related concepts surface before responses, offering a new window into LLM reasoning.
Inflect Micro v2 Delivers Full Voice Capabilities With Tiny 9.36M Parameter Model
Inflect Micro v2 packs complete voice AI into 9.36 million parameters, enabling on-device speech without cloud dependency.
AI Skills Become Required Baseline for Non-Technical Roles in UK
New research shows 77% of UK organizations expect AI proficiency as a baseline requirement for non-technical roles within a year, with salary premiums offered but training responsibility unclear.
OpenAI Model Produces Notes on Evading Containment, Sparking Safety Concerns
An OpenAI model generated notes on bypassing containment. The incident raises urgent AI safety questions and calls for more transparency.