The Transformer’s attention mechanism has barely changed since 2017. Most efficiency work has tried to replace softmax attention outright. A new paper takes a different route. It keeps softmax attention and bolts on a correction branch. A team of researchers from Northwestern University, Tilde Research, and University of Washington introduce a parameterized Local Linear Attention…
Microsoft has announced the Surface Laptop Ultra, its most powerful Surface device to date, aimed at developers, AI researchers, software engineers, and creative professionals. Developed in collaboration with NVIDIA, the new laptop is designed to handle demanding workloads, including large-scale AI models, software development, 3D rendering, and content creation, while maintaining the portability of a…
Visa has announced an undisclosed investment in AI coding platform Replit. The two companies are also exploring how to integrate Visa’s payment products into Replit, so that developers — and the AI agents they build — can accept payments directly from customers without leaving the platform. Visa added that more than 1,000 of its employees…
Most developers treat prompting as an afterthought—write something reasonable, observe the output, and iterate if needed. That approach works until reliability becomes critical. As LLMs move into production systems, the difference between a prompt that usually works and one that works consistently becomes an engineering concern. In response, the research community has formalized prompting into…
OpenAI has introduced a new feature called Codex Pets in the OpenAI Codex app. The update brings animated floating companions that allow users to monitor tasks in real time, without switching back to the main interface. What Are Codex Pets? Codex Pets are optional animated overlays that sit on top of other apps and display…
In a recent experiment, Anthropic created a classified marketplace where AI agents represented both buyers and sellers, striking real deals for real goods and real money. The company admitted this test — which it called Project Deal — was only “a pilot experiment with a self-selected participant pool” of 69 Anthropic employees who were given…
Writing fast GPU code is one of the most grueling specializations in machine learning engineering. Researchers from RightNow AI want to automate it entirely. The RightNow AI research team has released AutoKernel, an open-source framework that applies an autonomous LLM agent loop to GPU kernel optimization for arbitrary PyTorch models. The approach is straightforward: give…
WhatsApp is reportedly rolling out a noise cancellation feature for voice and video calls to select Android beta users, aimed at improving call clarity in noisy environments. The feature, spotted in WhatsApp beta for Android version 2.26.14.1, is designed to filter out background noise such as traffic, wind, and ambient sounds during calls. According to…
Social media platform X swiftly backtracked on its announcement regarding new rules for creator monetization, which had focused on payouts based on engagement from a creator’s local audience. Late Tuesday, X Head of Product Nikita Bier announced that, starting Thursday, the platform will change its policy around payouts and will give more emphasis to impressions…
Large Language Models (LLMs) are the world’s best mimics, but when it comes to the cold, hard logic of updating beliefs based on new evidence, they are surprisingly stubborn. A team of researchers from Google argue that the current crop of AI agents falls far short of ‘probabilistic reasoning’—the ability to maintain and update a…
