This week’s list of top data news highlights covers May 17, 2025 to May 23, 2025 and includes articles on using AI to simulate urban planning designs and robots that can navigate spaces humans cannot reach.
1. Accelerating Short-Form Animation
Invisible Universe, a U.S.-based animation company, has launched Invisible Studio, an AI-powered content creation platform that streamlines how animated content is made. The system combines Orbi—an AI assistant that analyzes social media engagement to fine-tune scripts, visuals, and pacing in real time—with Image2Model, a tool that can generate consistent, reusable versions of a character from just a single image. The platform is trained on five years of the company’s performance data and has cut average production timelines from ten days to six hours, and reduced costs by 95 percent.
Researchers at the Technical University of Munich in Germany have developed AI Pro, a computing chip that performs AI tasks directly on devices by mimicking how the brain processes information. Unlike traditional chips that separate memory and processing, AI Pro integrates both, enabling local data analysis . This design reduces latency by avoiding round trips to external servers, lowers security risks by keeping sensitive information on the device, and significantly improves energy efficiency by reducing the power needed for data transfer and computation.
3. Advancing Arabic Language AI
The Advanced Technology Research Council in Abu Dhabi, a government agency, has developed Falcon Arabic, a large language model trained on high-quality native Arabic datasets to reflect linguistic diversity across Arab dialects and regions. The system analyzes contextual patterns in Arabic grammar and syntax to generate culturally appropriate responses, making advanced learning models more accessible to Arabic-speaking users. Using advanced tokenization techniques optimized for Arabic script, the model matches the performance of models ten times its size while using far less computational power.
4. Preserving Holocaust Testimonies
The Illinois Holocaust Museum and Education Center has partnered with the Orleans Public Library in New York to offer immersive virtual reality experiences guided by Holocaust survivors. Using VR headsets, visitors can accompany survivors through historical sites like Auschwitz, hearing firsthand accounts enhanced by spatial visuals and voice narration. Each session lasts 12 to 15 minutes and is designed to foster emotional engagement, preserve testimony as eyewitnesses pass away, and provide a level of engagement that goes beyond traditional screens.
5. Teaming Up for Tiny Inspections
Researchers at the University of Colorado Boulder and Purdue University have developed a robotic inspection system for confined spaces that combines two nature-inspired machines: mCLARI, a four-legged microrobot the size of an insect, and a vine-like inflatable robot. Weighing less than a gram, mCLARI can crawl through cracks smaller than a penny and collect visual data in cluttered or hard-to-reach environments. But because its small size limits its ability to carry components like batteries or communication gear, it relies on the vine robot for transport. The vine robot extends through complex spaces using live camera feedback, delivers mCLARI to hard-to-reach areas, and retrieves it after deployment. Together, they form a closed-loop system that speeds up inspections and lowers operational costs.
Researchers from the University of Stuttgart and the University of Freiburg in Germany have developed FlectoLine, a responsive building façade system that changes its shape to respond to changing weather conditions. Installed on a greenhouse at the Freiburg Botanical Garden, the structure includes more than 100 fiber-reinforced plastic flaps that open and close using compressed air, guided by real-time environmental data. Machine learning algorithms continuously adjust the system to strike an optimal balance of sunlight exposure and shading, optimizing both energy efficiency and indoor comfort.
China has launched the first 12 satellites in what is planned to be a 2,800-unit network designed to run AI systems directly in orbit. The initial cluster was developed by ADA Space, a private aerospace company, and Zhejiang Lab, a government-backed research institute focused on AI and big data. These satellites are built to process information and sensor readings in space, rather than sending it down to Earth for analysis. This reduces delays, lowers the risk of data loss, and uses the vacuum of space for natural cooling. Powered by solar panels and connected by laser links, the system marks the first operational AI computing array deployed in orbit.
Strategis Groep, a Dutch technology firm, has launched an upgraded version of its browser-based urban planning tool that uses AI and geospatial data to help cities design smarter, more sustainable environments. The latest version features AI-powered tools that automatically generate layouts for buildings, roads, green spaces, and infrastructure, helping planners quickly explore different design options based on zoning rules, land use, and sustainability goals.
9. Flagging Child Welfare Cases for Enhanced Oversight
The New York City Administration for Children’s Services (ACS) has introduced an AI-based risk assessment system to flag child welfare cases for additional review. Developed by ACS’s Office of Research Analytics, the tool analyzes more than 200 variables, including prior agency contact, caregiver health history, number of siblings, neighborhood, and caretaker age, to estimate the risk of severe harm. When a case exceeds a certain threshold, it is escalated for further screening, which may involve in-depth interviews and enhanced oversight.
10. Predicting Potato Yields with Microbial AI
Researchers at Utrecht University and Delft University of Technology in the Netherlands have developed an AI-powered tool that predicts potato yield by analyzing the microbial DNA found on seed potatoes. Using data from over 200 test fields, the system combines microbial profiles with drone imagery of crop growth to identify bacteria and fungi linked to strong or weak performance. The model can pinpoint species such as Streptomyces as beneficial for yield, allowing farmers to select high-performing seed stock early.
