Research & Benchmarks - Page 30 of 34
Academic AI research, performance benchmarks, scientific breakthroughs, and peer-reviewed studies advancing artificial intelligence frontiers.
666 articles
View complete article list
ServiceNow uses LangSmith, knowledge graph and MCP to orchestrate agents
The enterprise is finally getting serious about agents, and ServiceNow just raised the bar.
OpenAI researcher details new AI model using general RL, no code interpreters
Forget the specialized tools. OpenAI’s newest model doesn’t use a custom math engine or a separate code interpreter. It just uses reinforcement learning, the kind you’d train a game-playing bot with, and a lot of raw compute.
WeatherNext 2 data now in Earth Engine, BigQuery; Vertex AI early access opens
Weather forecasting just got a major upgrade. WeatherNext 2’s forecast data is now live in Earth Engine and BigQuery, and early access on Vertex AI is open for custom model inference. This isn’t a simple tweak.
Stereogum persists amid streaming, AI and dwindling ad revenue as ads dry up
The music blog Stereogum has been a digital mainstay for nearly two decades, a place where the conversation about new albums, obscure bands, and the culture of listening felt like a genuine hang. But the internet’s economic ground has shifted.
DeepEyesV2 Beats Larger Open-Source Models by Leveraging Search Tools
In the arms race of AI, size has long been the ultimate advantage, until now. DeepEyesV2, a smaller open-source model, is punching well above its weight class. How? Not by memorizing more data, but by knowing when to reach for a tool.
Google AI agents: consistency, context, short-term session history, long-term memory
Most AI conversations feel like talking to someone with short-term memory loss. You give it your name. It asks for it again two lines later. The promise of a continuous, intelligent dialogue keeps breaking against simple amnesia.
Researchers push Context Engineering 2.0 as AI moves from Era 2.0 to 3.0
Every AI has a memory limit, a point where it starts making things up. This is not a minor bug. It is the core technical lie behind every demo where a chatbot flawlessly analyzes a novel you just uploaded.
OpenAI finds sparse models aid debugging, may boost mechanistic interpretability
OpenAI has a new plan for cracking open the black box: make the box smaller. For years, the field of mechanistic interpretability, the quest to explain exactly why a neural network does what it does, has been stuck.
RDMA Cuts CPU Use in S3-Compatible Storage, Boosting AI Performance
AI needs data constantly, and that movement is surprisingly expensive. Every chunk of training data pulled from an object store like S3 traditionally requires the server's central processor to handle the networking chatter.
Indian language ID proves tough; authors release baseline ML models
Indian languages don’t play nice with standard NLP. They share scripts, bleed into each other through code-mixing, and trip over their own morphological complexity.
NVIDIA Blackwell Wins All MLPerf Training v5.1 Benchmarks with FP4 Accuracy
NVIDIA just ran the table. In the latest MLPerf Training benchmarks, every single result came from a Blackwell system. The win was total. More importantly, it proved a point about precision everyone else missed.
Upwork study finds AI agents outperform alone when paired with humans
AI agents are not good employees. Left to their own devices, they screw up. But give them a human supervisor, and they become useful. This is the simple, expensive lesson from a new study commissioned by Upwork.
Human-aligned AI models show greater robustness and reliability, study finds
An AI can be perfectly sure of itself and totally wrong. That’s a problem. Humans generally aren’t like that. We feel our uncertainty.
DeepMind AI agent explores new games, explains its actions better than SIMA 1
DeepMind’s new AI agent, SIMA 2, makes its predecessor look like it was playing with the controller upside down. This one doesn’t just execute commands. It explains itself. It learns new video games on the fly. It can even read your emojis.
OpenAI Codex CLI Works with ChatGPT Plan and VS Code Extension
Your $20 monthly ChatGPT subscription is mostly a toy. The real tool is the command line. OpenAI’s Codex CLI, a terminal-based coding assistant, works with that plan. It unlocks the professional-grade utilities hiding behind the chat interface.
Bengaluru Hosts The Best Firm Summit 2026 for HR and AI Leaders
Bengaluru has booked another corporate conference for 2026. This one is called The Best Firm Summit, and it's for people who manage humans and people who manage machines.
Google AI Advisors Let Users Probe Performance with Conversational “Why” Queries
Google wants to replace your data analyst with a chatbot. The company's new Analytics Advisor tool, announced this week, is another attempt to make sense of metrics by letting you ask questions in plain English. Forget the pivot tables.
RECAP tool shows Claude 3.7 reproduces ~3,000 words from The Hobbit and Harry Potter
AI models are supposed to summarize and synthesize, not recite. A new test suggests they might be doing a lot more of the latter. Researchers have built a tool, called RECAP, to force large language models to cough up what they've memorized.
ElevenLabs' Scribe v2 delivers real-time, negative-latency transcription
ElevenLabs has released a transcription tool that claims to hear the future. Their new Scribe v2 doesn't just convert speech to text.
Meta's SPICE framework beats baselines, boosts math and general reasoning
Teaching logic to an AI is a famously stubborn problem. Meta’s researchers just attacked it with a new, pugilistic framework called SPICE. The core mechanic is simple: pit two AIs against each other.
Featured Resources & Reviews
No Code MBA Course Review
Learn to build AI-powered apps without coding. Our comprehensive review of No Code MBA's course.
AI Tools & Resources
Curated collection of AI tools, courses, and frameworks to accelerate your AI journey.
Weekly AI Digest
Get the week's most important AI news delivered to your inbox every week.