This is a YouTube news roundup (uploaded April 28, 2024 by LARRUZ) summarizing recent AI developments worldwide. Key items: the unveiling of Vidoo (a Chinese text-to-video model by Shengshu AI with Tsinghua University capable of generating 16-second 1080p clips using transformers + diffusion, with a cultural focus), renewed talks between Apple and AI companies (OpenAI, Google, Anthropic) about adding chatbot/AI features to a future iPhone OS release ahead of WWDC, Google's new image completion/trademarked RealFill (developed with Cornell researchers; can expand/repair images using up to five reference images) and broader Gemini app availability on older Android versions, Microsoft's planned AI Explorer for Windows 11 (reportedly demoed May 20 and possibly requiring Qualcomm Snapdragon X-series NPUs to run locally), and Italy's April 24, 2024 AI legislation creating two national AI authorities, a €1 billion support plan for SMEs, tax incentives, and criminal penalties (including 1–5 years jail for spreading fake AI images/videos), plus Italian-language Minerva models from Sapienza University.
Questions this video answers
April 24, 2024.
Shengshu AI in collaboration with Tsinghua University.
Approximately 16-second video clips at 1080p.
Up to five reference images.
The Digital Italy Agency and the National Cybersecurity Agency.
Chapters
Over the past few days, there's been a flurry of exciting AI developments, from unveiling the video AI model, challenging Sora to Apple exploring partnerships with OpenAI for new iPhone features, and Google launching an AI photo editing tool. Meanwhile, Microsoft is enhancing Windows with its new AI Explorer and Italy is stepping up with pioneering AI laws. So let's talk about this at this year's Zhonggguancun Forum in Beijing. The they introduced an incredible new AI model named Vidoo, which was developed by Shengshu AI in collaboration with Tsinghua University. What's cool about Vidoo is that it can generate 16 second video clips in crystal clear 1080p just by responding to a simple text prompt. It's set up to go head to head with OpenAI's Sora, marking a big stride for China in the competitive world of generative AI. They're really pushing to lead the pack in cutting edge technology. The tech at the heart of VDU combines the best of both worlds in AI technology, Transformers and diffusion models. This blend allows VDU to produce videos that are not just sharp, but also incredibly lifelike, featuring intricate facial expressions and dynamic lighting effects. But what makes VDU really stand out is its unique focus on Chinese cultural elements. It's tailored to include iconic symbols like pandas and the mythical Lung Dragon Dragon, which really resonates with Chinese content creators and audiences. This isn't just a technological breakthrough, it's a strategic move too. It underscores China's ambition to advance in AI while also staying true to its cultural roots. With its ability to create such realistic and culturally rich videos, Vidoo is redefining the standards for AI generated media and showcasing the innovation happening within China's AI scene. Alright now Apple has started talking to OpenAI again up about using the startup's technology and new iPhone feature, set to launch later this year. These conversations are picking up after a previous discussion earlier in the year, which didn't lead to much. Apple is also talking to Google about possibly using Google's Gemini Chatbot. It's not decided yet which tech partners Apple will go with, and it's still uncertain if any deal will happen at all. They might end up making deals with both OpenAI and Google, which would be very exciting for iPhone users. The upcoming version of the iPhone's operating system will have some new features that use Apple's own AI software, but Apple is also looking to include a Chatbot feature similar to OpenAI's ChatGPT and is searching for partners to help with this News about these talks first came out in March through a Bloomberg report, and another company, Anthropic, has also been part of these discussions. All this is happening as Apple gets ready for its Worldwide Developers Conference in about a month and a half, where the company plans to show off new AI software and services. They aim to make these new features work really well with Apple devices and to offer stronger privacy than other AI tools. Last year, Apple's CEO mentioned he uses OpenAI's ChatGPT himself, but he also noted there were several problems that needed to be fixed okay, now Google is making some major moves to improve AI for everyone. The tech giant has recently filed a trademark for a new AI photography feature named RealPhill. Developed in collaboration with researchers from Cornell University, RealPhill allows users to expand and repair existing images by leveraging up to five reference images. This tech wants to deliver more realistic and accurate outcomes compared to older image completion methods. The tool can add missing details or fix damaged sections in photographs, making it an especially useful addition for users of Google's Pixel smartphones or or those who use the Google Photos application. Now Google is making its newest AI features available on older phones too. They just announced that their Gemini app will work on Android devices that are still on older versions like Android 10 and 11. They said that they are doing this to make sure that even older devices stay up to date and their users can enjoy the newest tech no matter how old their phone is. Across the ocean in the uk, Google has been proactive in rolling out these AI features, following up on Samsung's own announcements. Samsung recently updated its Galaxy S23 range with new AI features and subsequently extended these updates to the Galaxy S22 series. Google's approach mirrors this by expanding access to the Gemini app, thus enhancing user experience across a broader range of devices. Alright, Next AI Explorer is a new feature coming to Windows 11 that will make everything on your computer searchable. It will help you summarize emails and do tasks like removing the background from images. Now Microsoft has been adding AI features to Windows lately, and AI Explorer is getting a lot of attention. It could be a big reason people buy AI powered PCs. There's a report that Microsoft will introduce AI Explorer at a special event for Windows on May 20, but it might only be available on computers with the new Snapdragon X series CPU from Qualcomm. This CPU has a neural processing unit that lets the computer handle AI tasks on its own without needing to send data to the cloud. Now AI Explorer is expected to make Windows 11 smarter by being aware of the context of what you're doing. It could make everything you do on your PC searchable, even across different apps. This means you could look up web pages you've visited, images you've opened, or even find parts of conversations. It could also summarize web pages, emails and conversations right on your computer. This would be quicker than using cloud based AI tools because it processes everything locally. You could choose to stop certain apps from being monitored by AI Explorer, or turn off the monitoring completely if you prefer. The feature will have a user interface that appears at the top of your screen. It will track what you do and make it searchable. For example, you could ask it to show you conversations about the Fallout TV show and it will find those chats for you. And it isn't just a search tool. It can also analyze what's on your screen and suggest actions like like removing an image background with a single click using Paint. A feature called Screen Understanding could help you write email replies or compare documents using Microsoft Copilot. Besides AI Explorer, Microsoft is working on other features too, like live captions and translation in real time, enhancing games with better visuals and smoother frame rates and new Windows Studio effects. Alright now. On April 24, 2024, Italy became the first country to introduce a set of laws specifically specifically for artificial intelligence. The Italian government approved new rules that focus on making sure AI is used safely and helps Italy stay competitive in this technology. The laws cover five main national strategy, oversight authorities, promotional efforts, copyright protection and criminal penalties. They also aim to make sure Italian rules line up with European Union standards, especially concerning how people learn about AI, job training and updating criminal laws for AI related crimes. Alessio Buti, Italy's innovation undersecretary, mentioned that these laws align with the EU's AI act and were discussed with other G7 countries. He explained that the laws clearly set out who will develop strategies, oversee, inform and enforce AI regulations. The legislation establishes two national authorities for AI within the Digital Italy Agency and the National Cybersecurity Agency. These bodies will oversee the application and implementation of both national and EU AI laws. The new rules also try to attract AI talent to Italy by offering tax breaks to Italians who return from abroad. They include a plan to invest 1 billion euros to support small Italian businesses in using AI. With the upcoming EU elections, Italy is also focusing on preventing harmful AI generated content. The Justice Minister Carlo Nordio stated that anyone spreading fake AI created images or videos without permission could face one to five years in jail. Furthermore, sneaky uses of AI will be considered a serious additional crime in cases like identity theft, price manipulation, fraud and money laundering. Meanwhile, researchers at Sapienza University are working on new Italian language AI models called Minerva. These models are being developed from scratch using a large amount of Italian and English texts and are meant to be open to the scientific community and soon the public. Alright, don't forget to hit that subscribe subscribe button for more updates. Thanks for tuning in and we'll catch you in the next one.
