• The Next
  • Posts
  • Google’s next-gen AI model Gemini outperforms GPT-4

Google’s next-gen AI model Gemini outperforms GPT-4

Friday News

#1. Google’s next-gen AI model Gemini outperforms GPT-4

Google has introduced Gemini, a versatile AI model capable of understanding text, code, audio, image, and video. Available in three optimized versions: Ultra, Pro, and Nano, Gemini surpasses human experts in language understanding and multimodal benchmarks.

Gemini, a Google model, is renowned for its native multimodality, a method that eliminates the need for separate components for different modalities. Its sophisticated multimodal reasoning allows for precise data extraction and high-quality code generation.

Google is introducing Gemini, an AI model that undergoes rigorous safety evaluations to ensure ethical deployment. The model is now available across various Google products, including the Bard chatbot, and can be accessed via the Gemini API in Google AI Studio or Google Cloud Vertex AI.

#2. AI multi-speaker lip-sync has arrived

Rask AI, an AI-powered video and audio localisation tool, has announced the launch of its new Multi-Speaker Lip-Sync feature. With AI-powered lip-sync, 750,000 users can translate their content into 130+ languages to sound as fluent as a native speaker.  

Lip movements in dubbed content have been lacking, making it unpopular in English-speaking countries. Lip reading helps perceive phonemic contrasts and is essential for learning to speak. Rask's new feature allows for more natural dubbed videos by automatically reshaping the lower face based on references, making the final result more realistic.

How it works:

  1. Upload a video with one or more people in the frame.

  2. Translate the video into another language.

  3. Press the ‘Lip Sync Check’ button and the algorithm will evaluate the video for lip sync compatibility.

  4. If the video passes the check, press ‘Lip Sync’ and wait for the result.

  5. Download the video.

Rask AI's founder and CEO, Maria Chmir, has introduced a new feature that visually adjusts lip movements to make characters appear fluent in a language. The technology uses generative adversarial network (GAN) learning and is available for all Rask subscription customers.

#3. Meta publicly launches AI image generator trained on your Facebook, Instagram photos

Meta Platforms, the parent company of Facebook, Instagram, WhatsApp, and Quest VR headsets, has launched a standalone text-to-image AI generator service, "Imagine," outside its messaging platforms. The service can be accessed at imagine.meta.com, requiring users to log in with their Meta or Facebook/Instagram account.

Early reactions are mixed

Already, AI artists around the web are experimenting with Meta Imagine to produce high-quality imagery quickly and consistently, with some comparing it to other popular AI image generators such as Midjourney, Stable Diffusion, and OpenAI’s DALL-E 3.

VentureBeat’s brief, unscientific tests showed that it only sporadically produced realistic human figures and structures — often our imagery included strange glitches like “melted” body parts and scenery.

Meta is launching Imagine, a minimalist interface with four generated images for users to download. The images are not customizable beyond a 1:1 aspect ratio square and include a watermark. Meta plans to add an invisible watermark in the coming weeks to increase transparency and traceability. Imagine aims to offer a functional, free competitor to existing AI art generators, which often require paid subscriptions.

Built atop Emu, trained on user-generated Facebook and Instagram images

Meta's Imagine service uses its own AI model, Emu, trained on 1.1 billion Facebook and Instagram user photos. The company excluded private messages and images not shared publicly. This decision is seen as prudent, but critics argue that users may not have intended for the photos to be used in this way.

Meta's researchers developed Emu based on quality metrics, revealing that a few thousand high-quality images and text can significantly impact the aesthetics of generated images without compromising the model's generality. Despite Meta's support for open source AI, neither Emu nor the Imagine by Meta AI service are open source. Meta is updating its apps with AI-enabled features, including a "reimagine" feature.

#4. San Francisco startup MaintainX raises $50 million to bring A.I. to industrial operations

MaintainX, a San Francisco-based startup, has raised $50 million in a Series C funding round led by Bain Capital Ventures, valued at $1 billion. The investment will help the company expand research, enhance artificial intelligence, and grow its customer base.

The platform also collects and analyzes data from various sources, such as sensors, equipment usage, and parts inventory, to provide insights and recommendations for improving operational efficiency and reducing downtime.

MaintainX, a software company, has raised $1.5m in a venture capital round. The company, which uses artificial intelligence and real-time data to identify and prevent potential breakdowns, is aiming to create a "zero-downtime future" in industrial operations. The company's chief executive, Chris Turlica, emphasized the importance of R&D and the vast amount of data it collects daily.

A new generation of software users

MaintainX, a software company, has been acquired by Bain Capital Ventures. The company aims to cater to the needs of a new generation of frontline professionals and purchasing managers, who value user-friendly software. MaintainX's growth, product quality, and customer satisfaction have impressed partners like Merritt Hummer, who believes the company is one of the best emerging growth-stage companies. Bain Capital's portfolio of over 400 companies, including many industrial and manufacturing businesses, could help MaintainX scale up and reach new customers.

A growing market opportunity

MaintainX, a company focusing on artificial intelligence and big data, has seen its revenue grow 13 times since raising $39 million in a Series B round in 2021. The industrial maintenance sector, valued at over $49 billion globally, has seen increased demand for software solutions to optimize workflows, reduce costs, and ensure compliance. MaintainX's success in securing funding and participation from high-profile business leaders suggests it has the potential to disrupt the sector.