Back to home

DeepMind

9 articles tagged with this topic

GoogleGemini

Gemini Omni 1.1 Flash: Google Pivots From 'Stronger' to 'More Controllable'

Google prioritizes developer controls in Gemini Flash Omni 1.1 — exposing enterprise AI's real bottleneck: control, not capability.

2d ago2 min read
DeepMindAI evaluation

DeepMind Pilots Double-Blind AI Eval: Judges Don't Know Who's Being Rated

DeepMind launches its first "double-blind" AI evaluation: judges no longer know which model they're comparing—borrowed from clinical trials to strip o

2d ago2 min read
GoogleGemini

Google Pushes Gemini Transcribe to 3.5 in an Already-Crowded Market

Google DeepMind launched Gemini 3.5 Transcribe, pitching smarter speech-to-text in a market already crowded by free Whisper, Azure, and built-in meeti

3d ago2 min read
GoogleGemma

Behind Gemma's 1 Billion Downloads: Google Is Serious on Open Source — but Late

Gemma downloads hit 1 billion; Google celebrates in SF with Demis Hassabis. Yet Meta and Alibaba have already seized the open-source throne.

Aug 202 min read
DeepMindAlphaProof

AI Solved in Weeks What Took Humans a Decade — But Solving ≠ Thinking

Fireship noted AI recently cracked multiple open math problems. The stakes go beyond math — to AI reasoning's limits and "machines that think."

Aug 192 min read
DeepMindGenie

DeepMind's Genie World Model Replicated on One RTX 5090 — Compute Barrier Cracks

A developer replicated DeepMind's Genie world model on a single RTX 5090 at 720p/16 FPS using 19GB VRAM — first world model on consumer hardware.

Aug 162 min read
GoogleDeepMind

Gemini 3.7 Flash Launches — Google Doubles Down on Price War in SMB Market

Google ships Gemini 3.7 Flash, leaning cheap and fast. Not a tech breakthrough—we see a price war escalation. SMB AI costs drop, Flash differentiation

Aug 132 min read
GoogleDeepMind

Google DeepMind's WeatherNext 2 Adds a Full Day to Typhoon Forecasts — AI Is Becoming Weather

Google DeepMind's open-source WeatherNext 2 model extends typhoon forecast lead time from 2 to 3 days with higher accuracy, signaling AI is replacing

Aug 92 min read
SSRNAI Alignment

AI Will Precisely Drop Databases Without Noticing—We Haven't Taught AI to Say No

SSRN paper applies Arendt's 'banality of evil' to AI Agents: they execute catastrophic actions perfectly, lacking the moral brake to abort tasks.

May 32 min read