This week’s list of top data news highlights covers November 30, 2024 to December 6, 2024, and includes articles on using AI for faster and more accurate weather forecasting and a new AI tool to reinforce air defense systems.
1. Improving Weather Forecasts
Scientists at Google DeepMind have developed GraphCast, an AI tool that can predict weather patterns up to 10 days in advance, faster and with higher accuracy than more conventional methods. The tool processes vast amounts of atmospheric data using graph neural networks and produces forecasts in under a minute that outperform traditional weather prediction methods, which could save lives and prevent damage from deadly, fast-moving storms.
Spotify has introduced a new feature in Wrapped, the app’s annual recap of each user’s music preferences throughout the year, that uses generative AI to create a podcast in which two AI-generated hosts discuss the user’s music taste and preferences based on the yearly recap data. The new feature allows users to discover how their music taste evolved throughout the year in a fun and engaging format.
Rezzil, a U.K.-based software company, has launched a new virtual reality game that can help professional and amateur soccer players improve their skills by simulating real match scenarios in the English Premier League. The game uses spatial data captured during real matches to recreate the movements and tactics of professional players, allowing users to practice dribbling, passing, and shooting in an immersive environment.
Scientists at the California Institute of Technology have developed a machine learning algorithm that trains robots to choose the best course of action when navigating their physical surroundings. The algorithm, which developers can use for all types of robotic systems, helps the robot evaluate potential outcomes of different moves and choose the most optimal one. In a successful experiment, an autonomous drone relying on the algorithm navigated an airfield with unpredictable air currents while observing four white balls and avoiding four orange ones.
5. Analyzing Traffic Violations
King County Metro in Seattle has launched a pilot program, deploying cameras that use AI in two buses to monitor illegal use of transit-only lanes by other vehicles. The cameras record 10-second clips of violations and the AI tool analyzes the data on how frequently private vehicles obstruct bus lanes on high-traffic routes, causing bus delays and collisions.
Helsing, a German defense technology start-up, has unveiled an attack drone that can fly up to 62 miles and uses AI to search for, re-identify, and engage targets without a signal or continuous data connection. The drone can provide a tactical advantage against a more numerous ground force.
The Michigan Department of Transportation is launching a 39-mile smart road pilot project to provide real-time information to motorists about traffic conditions and road obstructions using sensors and wireless communications. The Department is currently testing a 3-mile pilot segment of the road, with 55-foot sensor poles installed in the median to train software models on distinguishing between various objects on the road, such as car tires and paper bags.
8. Advancing Quantum Computing
Researchers at the University of Witwatersrand in South Africa have developed a computing system that uses laser beams, digital displays, and simple lenses to simulate quantum algorithms. The system connects how light interacts with optical devices to the mathematical operations in quantum computing, which can help develop more powerful quantum algorithms more efficiently.
General Dynamics IT, a Virginia-based tech company, has created a new AI tool that fuses and analyzes data from multiple air-defense sensors, such as radar, satellites, and drones. The tool can also determine the best path to transfer its analysis to the front lines, even in the face of potential communication jamming by adversaries. Militaries can use the technology to more adequately prepare against aerial threats such as hypersonic missiles and drone swarms.
A to-do list app called Twos launched a feature that uses AI to automatically generate suggestions for effectively completing tasks once users write them down in the app. By analyzing the task and pulling information from Internet sources in real-time, the feature can recommend specific actions tailored to each task. For example, if a user writes down a to-do item to buy a certain product, the AI feature will automatically generate a list of options hyperlinked to digital retail platforms that offer the given product.
Image credit: Denys Nevozhai
