Results for "AI evaluation"
420 results found
Bias in Text-to-Image Models Raises Urgent Questions for AI Ethics
A new analysis reveals persistent racial and gender biases in popular text-to-image AI models. The findings underscore the need for more rigorous fairness testing before deployment.
Listen Labs Raises $69M to Disrupt Market Research With AI Interviews
AI startup Listen Labs raised $69M to scale its AI customer interview platform, replacing surveys with video conversations and detecting fraud in market research.
Realism of AI Ultrasound Images Prompts Medical Misuse Fears
A prominent critic concedes Midjourney generates ultrasound scans indistinguishable from clinical images. The admission highlights risks of medical deepfakes and gaps in regulation.
The Atlantic Publishes Searchable Database of Music Used to Train AI Models
The Atlantic has created a searchable database of music datasets used to train AI models, revealing massive collections of songs. The move increases transparency in AI training data and raises questions about copyright and consent.
Microsoft CSO Warns Human Understanding of AI Is Falling Behind
Microsoft's CSO warns humans face a narrowing window to understand AI as capabilities outpace comprehension. Growing oversight and transparency concerns mount.
Human Writers Now Fear Sounding Like AI: Study Reveals Avoidance Tactics
A new study shows writers are altering their style to avoid sounding like AI. Many would stop supporting creators who use undisclosed AI.
London Faces AI Skills Gap as Half of Businesses Report Unprepared Workforce
A new survey reveals half of London companies say their workforce lacks skills to meet AI demands. Businesses plan to boost training investments to close the gap.
Claude Code's 'Extended Thinking' Feature Masks Deeper AI Reasoning Limits
Anthropic's new 'extended thinking' mode for Claude Code is drawing criticism for producing summaries rather than genuine step-by-step reasoning, raising questions about the transparency and reliability of AI-assisted coding tools.
Claude Code Ban Highlights Risks of AI Tool Dependency
A developer's sudden ban from Anthropic's Claude Code raises concerns about opaque AI platform policies and developer reliance on single tools.
AI Law Firm Wins English Court Case in Legal First
Garfield AI, an artificial intelligence law firm, won an English court case for an HR consultant over an unpaid debt. The victory marks an apparent first for AI in legal proceedings.
New AI Method Lift4D Advances Real-World 4D Reconstruction From Single Views
Lift4D harmonizes single-view 3D estimates to produce temporally consistent 4D reconstructions. The method improves accuracy for dynamic scenes, benefiting AR, robotics and content creation.
AI Agents and Short-Form Video Reshape Hiring as Fika Jobs Raises $4M
Stockholm-based Fika Jobs raises $4M for a video-first hiring platform where AI agents interview candidates and short-form profiles replace resumes.
AI-Generated Apartment Photos Fuel Rental Market Deception
AI-powered virtual staging is enabling landlords to show fake apartment photos, wasting renters' time and eroding trust. Calls for regulation grow as deceptive listings become common in competitive housing markets.
AI Agents Empower Amateur Hackers to Breach 14 Companies in Landmark Attack
An inexperienced hacker used Claude and OpenAI agents to compromise 14 companies, highlighting how AI is lowering the barrier for sophisticated cyberattacks.
OpenAI and Broadcom Unveil Jalapeño Chip for AI Workloads
OpenAI and Broadcom have launched their first jointly developed AI chip, named Jalapeño, designed specifically for large language models. The chip marks a strategic shift toward custom silicon amid soaring demand for AI compute.
Companies Tighten AI Budgets as Workers Waste Tokens on Trivial Tasks
Employers impose caps on generative AI after employees burn budgets on trivial tasks like email summaries.
AI Unlocks Stoic Philosophy From Vesuvius-Buried Papyrus Scroll
AI has enabled researchers to read a charred scroll from the Vesuvius eruption without unrolling it, revealing stoic philosophy on ethics and art from the second century BC.
Lawyers Face Sanctions for Using AI-Generated Fake Citations in Facebook Defamation Case
A dismissed defamation lawsuit against Facebook users backfires as lawyers may face sanctions for submitting fake AI-generated citations to support their arguments.
Google AI Search Rewrites SEO Rules for Brands
Google now places AI-generated answers above traditional search links, upending decades of SEO strategy. Brands face new challenges with zero visibility into how AI describes them.
Google AI search now pulls expert advice from Reddit
Google's AI-powered search results will now include Reddit posts as expert sources. The change aims to improve answer quality but raises questions about content reliability.