Simone Bombari asked us whether the 1-bit quantized random vector Y = sgn(Wx) has subgaussian norm bounded by a universal constant. Here W is an n n random Gaussian matrix, and x is an independent standard normal random vector in Rn. The question is nontrivial since the coordinates of Y are not independent. We give a strong positive answer to this question - for any bounded map instead of sgn() - using AI: AIDiscovery and Generalization (Theorem 1): To handle coordinate dependence, Gemini 3.5 Flash1 proposed decomposing the Gaussian vector into independent parts, using one part to "smooth" the sign function, and then applying Gaussian concentration for Lipschitz functions.

Eyes in the tech world have turned toward Mountain View, California this week. The San Francisco Bay Area city is where Google's headquarters is located, making it a logical place to hold the company's annual developer conference. That's right, gang, Google I/O 2026 kicked off on Tuesday with the usual opening keynote, which is where the company reveals what's arguably the event's most relevant info for consumers. Google made a ton of Android announcements last week, so its mobile ecosystem wasn't really on the agenda. But what else could the onus possibly have been on if not AI? We heard the word Gemini more times than I could possibly care to count, and the company had many updates to share on that front. Search, Google's longtime bread and butter, was a big focus of the event. The company talked up a new Ask YouTube feature as well as changes to AI subscription pricing and Workspace features like Docs and Gmail.

Google says Gemini 3.5 Flash rivals'large flagship models' for coding and agentic tasks Google says Gemini 3.5 Flash rivals'large flagship models' for coding and agentic tasks It can complete tasks in a fraction of the time of other frontier models, Google claims. Google has unveiled Gemini 3.5, starting with the Gemini 3.5 Flash model that promises to outperform Gemini 3.1 Pro in real-world agentic and coding tasks. Announced at Google I/O 2026, this will be Google's default AI model (not to be confused with Flash-Lite), designed to deliver better speed than the current Gemini Pro models at a more affordable price. The tradeoff is lower performance than the 3.5 Pro model (coming next month) in tasks that require deep reasoning and high-context understanding. However, Google has reduced the compromise between the Pro and Flash models, saying Gemini 3.5 Flash delivers intelligence that rivals large flagship models on multiple dimensions.

The CEO of Google DeepMind tells WIRED that companies should use the productivity gains of AI to do more, not lay people off. Demis Hassabis, the CEO of Google DeepMind, is keen to talk about the coding skills of his company's newest model, Gemini 3.5 Flash. The model has been trained to perform complex agentic coding tasks: translate large code bases from one language to another; find and fix bugs lurking deep in knotty code; and even write entire operating systems from scratch. Hassabis does not, however, think this spells doom for software developers. "I have no idea why people are going around talking with certainty about that," Hassabis tells WIRED ahead of the new model reveal at today's Google's I/O event .

Domain adaptation faces a fundamental paradox in the cold-start regime. When target data is scarce, statistical methods fail to distinguish relevant source domains from irrelevant ones, which often leads to negative transfer. In this paper, we address this challenge by leveraging expert textual descriptions of the target domain, a resource that is often available but overlooked. We propose a probabilistic framework that translates these semantic descriptions into a choice model, namely a Language-Induced Prior (LIP), that learns the preferences from a pretrained Large Language Model (LLM). The LIP is then integrated into an Expectation-Maximization algorithm to identify source relevance. Methodologically, this framework is compatible with any parametric model where a likelihood is available. It allows the LIP to guide the selection of sources when target signals are weak, while gradually refining these choices as samples accumulate. Theoretically, we prove that the estimator roughly matches an oracle cold-start MSE under a correct prior, while remaining asymptotically consistent regardless of the quality of the LIP. Empirically, we validated the framework on a descriptive (Gaussian estimation), a predictive (C-MAPSS dataset), and a prescriptive task (MuJoCo hopper).

Apple could unveil Gemini-powered Siri in Feb. Gemini 3 is now Google's default model for AI Overviews Plus, you can start an AI Mode conversation directly from a summary. The Google logo and lettering can be seen on the façade of the company's Munich headquarters building in Munich (Bavaria). Google has begun rolling out two upgrades for Search. Starting today, Gemini 3 is the default model powering AI Overviews. When the company debuted its new family of AI systems last November, it first deployed Gemini 3 in AI Overviews through a router that was programmed to direct the most difficult questions to the new system.

Why the World's Best AI Systems Are Still So Bad at Pokémon Pillay is an editorial fellow at TIME. Pillay is an editorial fellow at TIME. Right now, live on Twitch, you can watch three of the world's smartest AI systems-- GPT 5.2, Claude Opus 4.5, and Gemini 3 Pro --doing their best to beat classic Pokémon games. At least by human standards, they are not very good. The systems are slow, overconfident, and often confused.

Google launched Gemini 3 Flash as its new default AI model, offering up to three times faster performance than Gemini 2.5 Flash while being more cost-effective. PCWorld reports the model excels in multimodal tasks, scoring 81.2% in MMMU-Pro benchmarks and performing comparably to Gemini 3 Pro and OpenAI's GPT-5.2. This upgrade enhances Google's AI products with improved visual understanding, making advanced AI capabilities more accessible for everyday workflows and data analysis. Google has now launched Gemini 3 Flash, a faster and more cost-effective AI model based on Gemini 3. According to Google, Gemini 3 Flash is up to three times faster than Gemini 2.5 Flash, and it outperforms previous Flash models in all internal tests. In several benchmark tests, Gemini 3 Flash performed on par with both Gemini 3 Pro and OpenAI's GPT-5.2. In the multimodal test MMMU-Pro, it even topped the list with a result of 81.2 percent. The Flash model is supposed to be adapted for fast and repetitive workflows.

Google's Gemini 3 Flash model outperforms GPT-5.2 in some benchmarks Gemini 3 Flash is now rolling out to the Gemini app and AI Mode in Search. Almost exactly a month after the debut of Gemini 3 Pro in November, Google has begun rolling out the more efficient Flash version of its latest AI model. According to the company, the new system offers similar pro-grade reasoning performance as its flagship model at a fraction of the cost, making it ideal for everyday use. In benchmarks, the new system performed significantly better than Google's previous generation models, including Gemini 2.5 Pro. More notably, in Google's testing it managed to trade blows with GPT-5.2, the model OpenAI rushed out to counter Gemini 3 Pro.

Don't expect the next Clair Obscur, though. Google's at it again, once more insisting that AI is something people need or want more of in their lives. The latest move comes from YouTube Gaming, which announced an open beta for a project called Playables Builder. This allows select YouTube Creators to use a prototype web app built using Gemini 3 to make bite-sized games, no coding required. YouTube is launching a closed Beta test for Playables Builder, a prototype web app built using Gemini 3 where users create games with short text, video or image prompts. YouTube was testing the addition of small-scale games to its desktop and mobile platforms back in 2023, then added multiplayer capability to Playables last year.