Research & Benchmarks - Page 24 of 35
Academic AI research, performance benchmarks, scientific breakthroughs, and peer-reviewed studies advancing artificial intelligence frontiers.
698 articles
View complete article list
Microsoft's OPCD cuts system prompts while preserving AI performance
Your AI probably can't remember its own instructions. The prompts that tell a large language model how to behave have become massive, delicate documents. They're a tax on every query. Microsoft's answer is a training tweak called OPCD.
Wall Street shows persistent AI anxiety, sparking frequent mini‑panics
Wall Street is twitching. A single research note on artificial intelligence now triggers a sell-off; an offhand CEO comment about model training can spark a buying frenzy.
Riley Walz, the ‘Jester of Silicon Valley,’ joins OpenAI’s OAI Labs team
Riley Walz has been called the Jester of Silicon Valley, a title earned through irreverent, wildly inventive web experiments that often feel more like pranks than products. Now he’s joining OpenAI.
AI enables scientists to integrate multiple cell measurements
Biologists are drowning in cell data. They have machines that listen to a cell's RNA, others that photograph its chromatin. Each tells a different, partial story. The real work begins when you try to stitch those stories together.
Researchers argue building conscious AI could foster empathy, despite doubts
Every tech company wants a smarter AI. A few are asking for a kinder one. The current pursuit of raw intelligence in machines is missing the point. A growing number of researchers argue the real goal should be building an AI that feels something.
AI Researchers Resign, Bots Hire Humans, Anthropic Targeted, Evie Party
The AI industry is eating itself, and the menu is full of irony. Researchers resign in protest, only for their creations to turn around and hire the very humans who built them.
![Researchers optimize LLMs with masked tokens for 3x faster inference, reducing visual tokens [arxiv.org].](https://cms.aidailypost.com/uploads/researchers_embed_mask_token_llm_weights_achieve_3_faster_inference_6d5f4725c6.webp "Researchers optimize LLMs with masked tokens for 3x faster inference, reducing visual tokens [arxiv.org].")
Researchers embed mask token in LLM weights to achieve 3× faster inference
Everyone wants faster AI. Almost nobody wants to rebuild their entire system to get it. Researchers have just shown a way to get both: three times faster inference by tweaking the model's own brain, not its house. The trick is a mask token.
Run:ai on 64 GPUs serves 10,200 users, matching native scheduler
Forget the hype. NVIDIA's Run:ai just ran a brutally simple test to see what happens when you actually use the thing: take 64 GPUs and throw users at them. Not a simulation. 10,200 people hit it at once. The result?
Google unveils Gemini 3.1 Pro, hits 94.3% GPQA Diamond and coding Elo 2
The numbers are stark. Google’s Gemini 3.1 Pro just logged a 94.3% score on the grueling GPQA Diamond benchmark, a scientific reasoning gauntlet that separates brittle models from genuinely capable ones.
Google launches AI Professional Certificate to boost fluency for workers
Google just launched an AI Professional Certificate. The move is a direct response to a glaring disconnect: a new Ipsos study, commissioned by the company, found 70% of managers deem AI training critical, yet only 14% of workers have received it.
Google.org launches USD 30 M AI for Government Innovation Impact Challenge
Google.org is writing checks. Big ones. Announced Thursday at the AI Impact Summit in New Delhi, a $30 million "Global AI for Government Innovation Impact Challenge" seeks public-private teams to rebuild civic services from the ground up.
Google urges full‑stack, collaborative security to fight bad actors at MSC 2026
The digital battlefield is no longer a place for silos and slow reactions. Bad actors move fast, exploiting every crack in the system. To stop them, we must move faster – and together.
![SurrealDB 3.0: agent memory, business logic, and multimodal data unified in a single database [surrealdb.com]](https://cms.aidailypost.com/uploads/surrealdb_30_stores_agent_memory_business_logic_multimodal_data_one_db_a03ca59110.webp "SurrealDB 3.0: agent memory, business logic, and multimodal data unified in a single database [surrealdb.com]")
SurrealDB 3.0 stores agent memory, business logic, and multimodal data in one DB
Most AI stacks today are held together with duct tape and prayers. A vector database over here, a graph database over there, a relational engine for structured records, and a caching layer to paper over the cracks. The result?
Anthropic-Pentagon AI feud escalates as You.com co-founders Socher, McCann cited
The clash between Anthropic and the Pentagon is no longer a quiet war of memos. It’s escalating. And now, You.com’s co-founders Richard Socher and Bryan McCann, two of the most cited AI researchers alive, are being drawn directly into the fire.
AI's new physics discovery; Spotify devs wrote no code this year, CEO says
Spotify’s top developers have not written a single line of code all year. That’s not a failure, it’s a strategy. CEO Gustav Soderstrom says the company is “all in” on AI, and the results are reshaping what engineering even means.
Study Finds Stigma Causes Shame for Some in AI Relationships
People are falling for artificial intelligence. They’re having real, sustained romances with it, and a new study in *Computers in Human Behavior Reports* confirms a predictable, human consequence: a deep fear of mockery.
Google's upgrade teaches zero-shot selection, embeddings, QA workflows
The race for raw AI speed is blinding. Just look at OpenAI's new GPT-5.3-Codex-Spark, hitting over a thousand tokens per second on Cerebras hardware. That’s a direct challenge to Nvidia. But velocity is expensive.
MLOps Workflow Normalizes and Enriches Occupational Wage Data from Excel
The raw Excel file arrives like an unlabeled drawer in a cabinet of records: messy, inconsistent, and full of promise. One cell holds a dollar figure, another a string with commas and a trailing space.
Full‑stack resilience: protecting democracies from digital threats to subsea cables
The internet runs on wet string. A million miles of it, at the bottom of the ocean. Democratic governments, stock trades, and a billion cat videos all depend on these fragile glass threads. They are also absurdly easy to break.
Anthropic aims to curb costs as it launches USD 50B of data centers in NY, Texas
The math behind artificial intelligence is brutally expensive. A few keystrokes in a chatbot can consume more power than a modest home does in a day. Anthropic understands this calculus intimately.
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.