Ai & Tech

👍Nvidia has introduced an advanced AI model for generating 3D models. Edify 3D AI: — Generate in 4K quality; — There are grids with detailed geometry inside; — The project will turn out with high-quality textures and clear albedo colors. You can try it here. The first 50 attempts are free. @MediumAi

💥Hedra's main video console has been powerfully updated. • Background noise can be removed — the sound quality will be perfect. • You can clone someone else's voice completely or create a completely new one. • Generate videos up to 12 minutes — previously, you could only make a two-minute video. https://www.hedra.com/ @MediumAi

🎧Washington University students have assembled headphones for introverts — they block all unnecessary sounds at a distance of 2 meters. It is possible to create a "bubble of silence" using a neural network and 6 microphones that detect sounds that interfere with communication in noisy places like bars and parties. @MediumAi

🧺A washing machine for people appeared in Japan — Mirai Ningen Sentakuki will wash a person from all sides. The whole process takes 15 minutes. The device will even determine how dirty you are and select the method of washing, pressure and temperature of the water. In addition to washing, there is a drying function and a YouTube screen so as not to get bored. @MediumAi

🧠Alibaba's Qwen2.5 Turbo Reads Ten Novels in One Minute The new Qwen2.5 Turbo model has a context window of 1 million tokens. This allows her to process ten novels, 150 hours of transcripts or 30,000 lines of code at a time.❗️In several tests for understanding long texts, the Qwen2.5-Turbo is superior to the GPT-4, while the model's performance with short texts is comparable to the GPT-4o-mini. Also, the developers from Alibaba have reduced the time it takes for the model to process large text from 5 minutes to 68 seconds. Qwen2.5-Turbo is available via the API at a price of 4 cents per 1 million tokens. Demo: https://huggingface.co/spaces/Qwen/Qwen2.5-Turbo-1M-Demo @MediumAi

💥Pokemon Go Players trained Artificial Intelligence Niantic has announced the development of a "large geospatial model" (LGM) that will help robots and computers better understand the world. The model uses data from Pokemon Go players' smartphones and other Niantic products. Currently, the database contains more than 10 million scanned 🗺 locations from all over the world. The company calls this data unique because it is obtained from a pedestrian's point of view and includes places inaccessible to cars. @MediumAi

📚Google has pleased students by a cool new neural network — LearnLM 1.5 Pro. The base is the same neural network that underlies "Learn About", but now you can chat with it. AI greatly simplifies learning — it selects an approach for each student. Enjoy it! @MediumAi

✨This is an AI photoshop that works right in the browser. The toolkit is simple: select the desired area with a suitable tool, and then change the color, delete the object or generate new ones. https://huggingface.co/spaces/AI4Editing/MagicQuill @MediumAi

❗️Eleven Labs has launched the ability to create voice chatbots Chatbots work on the basis of third-party LLM Gemini, GPT, Claude or the creator's own model. You can set the tone of responses for the bot, adjust the voice and upload your own knowledge base. In addition, the AI will be able to collect data from customers who communicate with it, such as name and email. @MediumAi

👾 Kia Cerato streamer now broadcasts with the 1X robot He bought himself an android assistant robot EVE from the Norwegian company 1X Tech for $ 24,000 (~2.4 million rubles) to brush his teeth, stream with him and perform a variety of tasks around the house. Watch live how the humanoid copes with the tasks. @MediumAi

⚡️Scientists have digitized the brain of a fly, making it IMMORTAL! For the first time, American scientists from Princeton University have managed to make a three-dimensional map of such a complex organism as the brain of an adult fruit fly. 🧠It has 139.5 thousand neurons and 54.5 million synaptic connections between them. Hundreds of scientists have been working since 2016. It took science 33 years to complete the markup! The digital image of the brain is publicly available here. This is a giant step towards understanding the human brain and treating its diseases. @MediumAi

🖨 Nike printed sneakers on a 3D printer: the company introduced the Air Max 1000 The only unprinted part is the air cushion in the heel. All other elements of the shoes are made of a single flexible material. The price and release date are not announced. @MediumAi

💊A robot surgeon was able to successfully perform an operation, having learned all the skills of a surgeon from uploaded videos. Researchers from Johns Hopkins University (USA) conducted an experiment in which they decided not to train the 🤝 robot in each individual movement, but simply showed the entire process of a particular operation in the format of short videos. ⏺Thanks to the use of AI in the course of training, scientists were able to copy every action with 100% accuracy - while in non—standard situations, the robot took over decision-making. @MediumAi

🫂 MIT students showed a talking teapot with which they can spend their evenings. The device pours tea by itself, maintains the temperature and conducts conversations in parallel. It has a ChatGPT column inside, so if your friend has the same kettle, they can be connected, and the device will tell you how his day went. 🟢There is also a screen that generates real-time images related to the story. Some kind of black mirror. @MediumAi

🚁Xpeng company presented an invention — a car with an integrated helicopter. The whole "transformation" takes 5 minutes. There are 2 people on board, and one charge is enough for 5-6 flights of 10 minutes each. The price starts from 280 thousand dollars @MediumAi

💥SpaceX founder Elon Musk has proposed replacing airplanes with spaceships. From Moscow to Bangkok in 30 minutes, from London to New York in 29 minutes, from New York to Shanghai in 40 minutes — and all this with maximum comfort. The spacecraft, which accelerates to 27,000 km/h, promises to replace airplanes and open a new era of ultrasonic flights. Musk is confident that the future of intercontinental travel lies with his Starship. @MediumAi

🍿Organize a pleasant movie evening for yourself: a new AI clearly selects films to suit your mood. All you need to do is fill out a 17-point questionnaire, and the FeelFlix neural network will choose a movie based on your emotions. @MediumAi

⚡️A Tesla robot taxi was spotted on the territory of a gigafactory in Texas Tesla has begun testing a prototype Cybercab robotaxi on private roads near a Texas factory, as the company does not yet have permits to test driverless cars on public highways. A car without steering wheel and pedals is due to be released by Tesla in 2026. @MediumAi

❗️😮Skynet is coming: at an exhibition in China, a baby robot escaped to persuade adult robots to rest. He asked them about their work, and when he found out that they worked seven days a week and at home, he invited them to leave with him. People tensed up. @MediumAi

💦Tunisian startup Kumulus has developed equipment for obtaining drinking water from the air. Powered by solar panels, it connects to an application to monitor water quality and replace filters. The facilities are located in remote trees and villages where there is a shortage of drinking water. @MediumAi