~ai-venturebeat | Bookmarks (535)

GUEST: Intelligence is pervasive, yet its measurement seems subjective. At best, we approximate its measure through tests and benchmarks. Think of college entrance exams: Every year, countless students sign up, memorize test-prep tricks and sometimes walk away with perfect scores. Does a single number, say a 100%, mean those who got it share the same intelligence […]

Are we unlocking new frontiers in AI reasoning, or simply stretching the limits of token memory without meaningful improvements?

Reacting to continuing stock market woes and perhaps tech industry lobbyin, Trump backed off on tariffs for electronics late last night.

It achieved an 8.0% higher win rate over DeepSeek R1, suggesting that its strengths generalize beyond just logic or math-heavy challenges.

Get ready. GamesBeat Summit 2025 will take place on May 19 to May 20 at the Marriott Marina del Rey in Los Angeles.

DeepCoder-14B competes with frontier models like o3 and o1—and the weights, code, and optimization platform are open source.

Ai2's new open-source OLMoTrace tool allows enterprises to directly trace LLM outputs back to original training data, bringing transparency to AI decision-making and addressing trust barriers.

Writer unveils AI HQ platform to transform enterprise work with autonomous agents that execute complex workflows across systems, potentially reducing workforce needs while delivering measurable ROI on AI investments.

NTT Research announced at an event that it has started a new AI basic research group, dubbed the Physics of Artificial Intelligence Group.

Sophisticated attacks must be tracked and contained in a business's core security infrastructure, managed from its SOC.

Devs and non-devs can use the cloud-based, Gemini-powered platform to build, launch, iterate and monitor apps, APIs, backends and frontends.

Anthropic launches new Claude Max subscription tiers at $100 and $200 monthly, challenging OpenAI's premium offerings while targeting power users who need expanded AI assistant capabilities.

The business version of Gemini builds on the core AI capabilities already familiar to Android developers but introduces enhanced features for organizations managing sensitive codebases and workflows.

Google unveils Ironwood, its seventh-generation TPU chip delivering 42.5 exaflops of AI compute power — 24x more than the world's fastest supercomputer — ushering in the "age of inference."

Google Cloud releases new Google Unified Security platform to simplify cybersecurity, along with new semi-autonomous AI security agents.

Google expanded its agent offering with Agent Development Kit where enterprises can build and deploy AI agents all in one place.

Google unveils Ironwood TPUs, Gemini 2.5 "thinking models," and Agent2Agent protocol at Cloud Next '25, challenging Microsoft and Amazon with a comprehensive AI strategy that enables multiple AI systems to work together across platforms.

Reward models holding back AI? DeepSeek's SPCT creates self-guiding critiques, promising more scalable intelligence for enterprise LLMs.

The initial model lineup includes five base sizes: 3 billion, 8 billion, 14 billion, 32 billion, and 70 billion parameters.

Wells Fargo’s generative AI assistant, Fargo, surpassed 245 million interactions in 2024 using a model-agnostic architecture powered by Google’s Flash 2.0. The bank’s privacy-forward orchestration approach offers a blueprint for regulated industries looking to scale AI safely and efficiently.