This week’s list of top data news highlights covers October 12, 2024 to October 18, 2024, and includes articles on Mistral’s new AI models for phones and laptops and how an algorithm can help scientists estimate the next big earthquake in the Pacific Northwest.
A San Francisco-based startup called Abel has launched a tool that uses generative AI to draft police reports from analysis of body camera footage and data from dispatch call centers. The tool helps police officers spend less time writing reports, especially for incidents such as car chases where they must note down every detail, allowing them to focus on other duties.
The Ohio Secretary of State has launched a new dashboard that provides nearly real-time data on early voting in the state. Voters and researchers can use the interactive tool to see statistics, broken down by county, party affiliation, and age, on absentee ballots submitted before Election Day. The tool can provide more transparency to the public on the electoral process and help researchers study voting patterns in Ohio.
3. Translating Product Listings
Chinese e-commerce platform Alibaba has launched a tool that uses generative AI to translate product listings into 15 languages, including Arabic, English, Russian, and Turkish while preserving contextual and cultural nuances in a given language to enhance the quality of the listing. The tool can help sellers create appealing product listings in various languages to boost their foreign sales.
UBTech, a Chinese robotics company, has launched a humanoid robot that can ease labor shortages in car factories and boost productivity by automating strenuous tasks like carrying parcels and conducting visual quality inspections using AI to analyze camera footage. BYD, the world’s largest electric vehicle maker, is using UBTech’s robot.
5. Enabling On-Device Computing
Paris-based startup Mistral AI has released two generative AI models specifically optimized to run on consumer laptops and smartphones. Developers can use these smaller generative models to run tasks directly on devices, process data locally, and fine-tune them for specific needs, more cost-effectively than other models.
6. Identifying Counterfeit Products
Cypheme, a Paris-based tech company, has created a tool called VraiAI that identifies counterfeit or fraudulent products by analyzing microscopic visual details of an item using computer vision models. Lacoste is using VraiAI to help identify and remove fraudulent products based on the brand’s crocodile logo.
7. Combating Cocaine Addiction
Researchers at the King’s College London are testing strategies to help people overcome their cocaine addiction using virtual reality. The researchers create various scenarios in which people with addiction usually experience the urge to consume cocaine, such as during parties. Users then attempt to break their addiction patterns by practicing how to refuse cocaine in a safe, immersive environment.
Adobe has launched several new features for its Photoshop app that use generative AI to help people edit their photos more efficiently. One of the new features uses AI to analyze an image, automatically find common unwanted objects in the background, and remove them with a single click.
9. Enabling Multi-Agent AI Development
OpenAI has released an open-source framework called Swarm that provides developers with a blueprint and example code for building interconnected multi-agent AI systems. Swarm’s framework helps developers create AI agents that can exchange data and collaborate to complete tasks autonomously. For example, a company could use Swarm to create AI agents in different departments, which could autonomously work together on administrative and research tasks.
Researchers at the University of Texas at Austin have built a new algorithm to more closely study layers of sedimentary deposits resulting from earthquakes, which allows scientists to estimate the intervals between past earthquakes. The researchers used the algorithm to analyze X-ray images of these layers to estimate when the next big earthquake in the Pacific Northwest will happen. The researchers concluded that the time until the next big earthquake may be longer than the widely accepted interval of 500 years.