ARDEZART

  The Chicago exhibition argues that bird-friendly façades are not a constraint to hide — they can be a source of architectural invention. A façade can be beautiful and dangerous Chicago sits on the Mississippi Flyway, one of North America’s major migratory bird routes. The combination of dense urban development and extensive glass creates a serious collision risk, turning an architectural material choice into an ecological issue. Flyway City: Architecture for a Flourishing Ecosystem, designed and co-curated by Studio Gang in collaboration with the Chicago Architecture Center, focuses on ways the built environment can reduce that risk. Bird safety as design intelligence The exhibition’s premise is more interesting than simply adding markers to glass after a building is designed. It asks how bird-safe strategies can become part of architectural thinking from the beginning — influencing façades, pattern, transparency, material selection and urban development. That shift matters...

  ChatGPT and Gemini are becoming product-discovery channels, forcing retailers to optimize for AI answers while protecting the customer data that powers their businesses. From search engine optimization to answer optimization Retailers are beginning to treat AI assistants as a serious discovery channel. Reuters reports that companies including Walmart, Ulta Beauty and Wayfair are changing how their sites expose product information so that their products are more likely to appear in answers from systems such as ChatGPT and Google Gemini. The traffic appears unusually valuable. Reuters cites Adobe Analytics data indicating that 41% of U.S. consumers used generative AI for online shopping in June, while visitors referred by AI services generated 41% higher revenue per visit than shoppers coming through more traditional channels. The battle is over the customer relationship Retailers want visibility inside AI assistants, but many do not want the assistant to own the transaction. ...

  OpenAI says preliminary testing of its upcoming Astra model is strong enough that it cannot rule out a critical level of autonomous cybersecurity capability. Why Astra is getting extra scrutiny OpenAI has tightened internal controls around an upcoming model called Astra after preliminary cybersecurity evaluations produced results serious enough that the company says it cannot rule out a “critical” capability level. Reuters reported on August 7 that OpenAI paused internal activities involving Astra that did not meet newly strengthened security requirements and moved development into more isolated environments. Under OpenAI’s safety framework, the critical threshold is reserved for capabilities such as autonomously finding and exploiting severe real-world vulnerabilities or carrying out complex attacks against highly secured targets without meaningful human assistance. OpenAI has not said Astra definitively meets that bar. The important point is that early results were strong en...

  A new research framework tries to solve a major world-generation problem: making the inside of a building actually correspond to the city and exterior that contain it. The missing link in generated worlds AI systems can already generate outdoor scenes, building forms and interior layouts, but those outputs are often disconnected. A convincing street may contain buildings whose interiors were generated with no meaningful relationship to the façade, footprint or surrounding block. HoloWorld, a new research framework posted on arXiv on August 6, attempts to treat the city as one continuously updated context rather than a collection of separate generation tasks. From city plan to room The framework starts from a user description and progressively represents information at different scales. It generates urban exteriors while maintaining context across neighboring blocks, grounds those exteriors in 3D building instances and footprints, and then uses the building geometry and appea...

  AMD is adding Taalas and its specialized inference silicon to a growing stack of hardware and software bets aimed at the next phase of AI deployment. Inference is becoming the next battleground AMD said it will acquire Toronto-based chip startup Taalas for an undisclosed sum, bringing another specialized inference team into its expanding AI business. Taalas develops silicon intended to reduce compute and memory bottlenecks when trained models are actually run in production. That distinction matters. Training has dominated the AI hardware narrative for years, but deployment is increasingly about inference: serving huge numbers of prompts, agent actions, generated images, video frames and model calls at the lowest possible cost and latency. How Taalas fits into AMD’s strategy Reuters reports that AMD plans to integrate Taalas technology into its accelerator roadmap and pair it with AMD Instinct GPUs in broader system-level solutions. Taalas had raised about $219 million since ...

  The proposed Terafab would bring logic, memory, packaging and testing under one roof for chips destined for robots, autonomous vehicles and space-based computing. A fab at infrastructure scale SpaceX and Tesla plan to initially invest $16.8 billion in a semiconductor complex called Terafab in Grimes County, Texas, according to Reuters. The companies describe the project as a vertically integrated site for manufacturing, packaging and testing advanced logic and memory chips. The planned scale is unusual even by semiconductor standards. Reuters reports a footprint of roughly 100 million square feet, with at least 3,000 jobs expected in the initial buildout. Earlier filings suggested later phases could push total investment far higher. What the chips are for The project is tied directly to the compute ambitions of both companies. Tesla needs more AI silicon for systems such as Optimus robots and Cybercabs, while SpaceX has been pushing toward more tightly integrated AI and spac...

  The training-free framework aims to preserve motion and long-range consistency by changing how existing diffusion models sample video across time. AI video still has a time problem Modern video diffusion models can produce impressive short clips, but extending those clips is difficult. Long generation tends to expose temporal drift, discontinuities between windows and a loss of motion quality as the model moves further away from its original training regime. A new paper posted on August 6 proposes Diff-VF, a training-free and model-agnostic framework designed to convert existing short-video diffusion backbones into longer-form generators. How Diff-VF approaches continuity The method combines three sampling strategies. Hybrid Noise Initialization constrains global semantics, Weighted Window Sampling aims to reduce visible discontinuities between temporal windows, and Temporal Extended Sampling creates longer-range dependencies through timestep-dependent fusion. The authors a...

  Gaku Tada’s stylized Blender animation shows why the most interesting 3D work may be moving away from perfect realism and toward visible artistic authorship. When realism stops being the point One of the most persistent goals in computer graphics has been realism: better materials, better lighting, better simulation and fewer visual clues that an image was rendered. Gaku Tada’s latest Blender work moves in the opposite direction. Creative Bloq highlighted the artist’s new Deep Paint scene on August 5, describing a quiet woodland animation with painterly textures, soft color transitions and an illustration-like finish. The underlying scene still uses the logic of 3D — camera, lighting, geometry and motion — but the image is designed to feel hand-made rather than physically perfect. The 2D–3D boundary keeps dissolving Blender has become particularly interesting for artists who do not want to choose between 2D and 3D. Grease Pencil, non-photoreal rendering workflows and painter...

  Instead of forcing Berlin into one fixed wordmark, the museum’s new identity draws from the city’s signage, architecture and typographic contradictions. A city that refuses a single voice Stadtmuseum Berlin is becoming Berlin Museum, and the institution has paired the new name with an identity designed to behave more like the city than a conventional museum brand. The official museum site says its colors, logos and typefaces are intended to reflect the many stories and perspectives that make up Berlin. Brand studio Stan Hema built the system around typography pulled from the visual memory of the city: U-Bahn signage, historic buildings, streets, shops and everyday lettering. Instead of one immutable wordmark, the identity can change while staying inside a recognizable system. Why the logo count became the headline Creative Bloq notes that the identity combines two typefaces drawn from a pool of nine, producing up to 81 wordmark combinations. That sounds like a direct violati...