Google's AI Video Generator BEATS ALL, OpenAI Announces New Models, AI for Art Therapy, & More
Welcome to the AI Search newsletter. Here are the top updates in AI this week.
Google’s New Video Generator That Beats Competition
Google Research introduces Lumiere, a groundbreaking text-to-video diffusion model that tackles the challenge of synthesizing realistic and diverse motion in videos. Unlike existing video models, Lumiere's Space-Time U-Net architecture generates the entire temporal duration of a video in a single pass, achieving global temporal consistency effortlessly. By utilizing spatial and temporal down- and up-sampling techniques and leveraging a pre-trained text-to-image diffusion model, Lumiere can directly generate high-quality, low-resolution videos at full-frame rate. This impressive model showcases state-of-the-art text-to-video generation results
Too long to read?
We’ve condensed this week’s top updates in AI into this short video. Check it out:
OpenAI New Models & API Updates
OpenAI is excited to announce the launch of their new generation of embedding models, including the high-performance text-embedding-3-small model and the powerful text-embedding-3-large model. Additionally, OpenAI is releasing the updated GPT-4 Turbo preview model, gpt-4-0125-preview, which offers enhanced capabilities for code generation. Developers can now assign permissions to API keys and have access to a usage dashboard to track usage metrics on a per feature, team, or project level.
New Research Combats Deepfake Audio
In a world where deepfakes and doctored audio are becoming increasingly common, researchers have developed techniques to distinguish fake from real audio. Their research involved analyzing perceptual features like pauses and variation in volume, extracting spectral features using an audio wave analysis package, and training a deep-learning model to generate multi-dimensional representations for accurate detection. This research aims to address the rising concerns surrounding deepfake audio's potential misuse, such as biometric verification bypasses and scam calls.
Flexispot
Elevate your home office with the finest, ergonomic standing desks and chairs, designed to promote good posture and reduce the risk of back and neck pain.
Revolutionizing Art Therapy with AI
Enter DeepThInk, an AI-assisted digital art tool aimed at enhancing the expressive abilities of art therapy patients. In collaboration with art therapists, the program combines traditional drawing and painting tools with an AI brush that generates complex images based on user suggestions. The team focused on creating an easy-to-use and accessible tool, especially for those lacking art supplies or digital art skills. DeepThInk aims to empower users by augmenting their artistic abilities and enhancing the therapeutic process. The researchers hope to release the tool as a free, open-source tablet app in the future.
💡Prompt Tip of the Week💡
Specify Output Format: You can specify how ChatGPT formats the reply. For example, "What are the longest highways in the United States? List only the top four in the form of a bullet list.”
The Incredible Potential of 3D Printed Electronic Skin
In a groundbreaking development, researchers have created a 3D-printed electronic skin (E-skin) that mimics human skin, providing an array of sensory experiences. This utilizes nanoengineered hydrogels with electronic and thermal biosensing capabilities, allowing the E-skin to flex, stretch, and sense like real skin. With applications in industries such as robotics, prosthetics, wearable technology, sports, and entertainment devices, this innovation has the potential to transform human-machine interaction and improve the quality of life for individuals with disabilities. Furthermore, the E-skin can be used for wearable health devices that continuously monitor vital signs, enhancing motor skills, coordination, and overall well-being.
Muscle-Powered Two-Legged Robot Takes Inspired Steps
Researchers from the University of Tokyo have crafted a biohybrid robot that combines muscle tissues and artificial materials to create a two-legged robot with human-like movements. The robot mimics human gait and can walk, pivot, and make sharp turns. By attaching lab-grown muscle tissues to a silicone rubber skeleton, the researchers were able to achieve efficient and silent movements. The robot is powered by zapping the muscle tissue with electricity, allowing it to walk at a speed of 5.4 mm/min.