AI revolutionizes everything! Last week’s hottest news revealed the secrets of artificial intelligence
In last week’s AI news, there were no major announcements like GPT-5 or Mid-Journey version 6, but there were many small advancements in the AI field. Meta released an open-source AI music generation model that anyone can use on Hugging Face. This model sounds better than other models like Google’s Riffusion or Music LM. It can be obtained on GitHub and can be installed locally if you have a graphics card with at least 16GB of memory.
Adobe has updated Adobe Express, including the addition of Adobe Firefly Generative AI, which allows users to use AI prompts for images and add text effects directly within Adobe Express. They have also added generative AI tools in Adobe Illustrator, including Generative Recolor, which allows users to upload black and white images and add colors, or upload designs and regenerate them with various new color variations. These updates are now available to Adobe Creative Cloud users.
AI companies such as OpenAI, DeepMind, and Google’s AI department have agreed to open their models to the British government for research and security purposes. However, they have not yet provided the source code to the public. Some have noticed that AI functionality is being integrated into Google products, such as the “help me write” button in Gmail and Google Docs. The article suggests that AI is changing the world and can be used to write articles and emails.
Google has announced the release of a new AI model that can take a small portion of an image and use its training data to fill in the rest of the image. This model is based on Yann LeCun’s vision for more human-like AI and is being open-sourced. The model works more like the human thought process, recognizing a small portion of an image and using its knowledge of the subject to fill in the blanks, unlike other AI generative models that attempt to recreate images at the pixel level.
AMD has developed an AI model that is trained like a human and can fill in the remaining context based on its dataset and understanding of small pieces of information it sees. This model is available on GitHub, and AMD has partnered with Hugging Face to provide them with computing power for their machine learning products. This is significant because AMD, not Nvidia, is providing the computing power. AMD has announced that they are building hardware specifically tailored for AI, with the goal of optimizing the combination of hardware and software. They are also trying to compete with Nvidia by creating chips that are more powerful than Nvidia’s. AMD’s new chip, Mi 300x, is designed specifically for generative AI and provides 2.4 times the memory and 1.6 times the memory bandwidth of its competitor. The article discusses the competition between Nvidia and AMD in creating better chips for large language models.
OpenAI has announced an update to their GPT language model, which includes a new 16,000-context version that is four times larger than the previous version. This new version allows for approximately 12,000 words between input and output for chat GPT. Developers using the GPT 3.5 API will see a decrease in costs, but it is unclear whether the company will pass on these savings to end consumers.
Microsoft and OpenAI have an “amazing partnership”, but tensions have arisen as they work together and compete on AI features. It is reported that OpenAI warned Microsoft to proceed with caution when integrating GPT 4 into the Bing search engine to avoid inaccurate and unpredictable responses, but Microsoft proceeded anyway. People are concerned that if they can use Bing with built-in GPT for free, why bother buying chat GPT. Microsoft and OpenAI have a unique partnership, which has resulted in some conflicts behind the scenes as both companies support and compete with each other. Both companies have chat models, GPT+ and Bing Chat, which have built-in GPT.
There are reports that Sir Paul McCartney says that artificial intelligence has allowed the completion of the last Beatles song. Google has also announced a new generative AI model for virtual try-ons, allowing users to create try-on images using AI. The model was created using Google’s shopping graph and diffusion model. Google’s shopping graph has already launched a virtual try-on feature for women’s tops for brands such as Anthropologie, Loft, H&M, and Everlane. This feature will become more precise over time and expand to more brands.
Google Shopping has partnered with Shopify, a commercial platform that allows anyone to set up their own store to sell physical products. Shopify simplifies the complexity of running a business, making it possible for anyone, anywhere to become an entrepreneur. The article discusses the convenience of using Shopify to set up an online store and how Shopify is democratizing technology for entrepreneurs, helping to build tomorrow’s economy.
A new trend has emerged on Twitter, using control grids and stable diffusion with QR codes (a type of two-dimensional barcode) to create stunning images. The article provides a tutorial link on how to set up this trend. Another trend on Twitter is creating 3D models that can be downloaded from Hugging Face and made at home. Rowan Chung created a tutorial with over 1.3 million views on how to make these models. While the resulting images look cool, generating them is difficult as some QR codes work while others do not.
A new technology called “re-render” can improve video quality and solve flickering and artificial artifact problems. The text discusses a new video model that doesn’t have the flickering effects seen in other models. The model is demonstrated in a video and compared to other models. The new model is described as clear and without continuity issues. The author is excited about the potential uses of this model in various tools and platforms. Whether the model will be available to the public is still uncertain.
The text discusses the recent Mid Journey office hours phone call, as well as what to expect from the upcoming 5.2 release. This version is expected to drop any day now and will bring new features like limited Discord compatibility and changes to scaling and aspect ratios. The prompt analyzer was also discussed, which is currently used by inputting “description” and uploading an image. Version 6 was originally expected to be ready in 6-8 weeks, but now it’s slated for release in July. The article also talks about developing a standalone web and mobile version to reduce word prompts in Discord. The team is working to improve reviews and reduce rejected prompts.
In addition, 11 Labs has introduced a speech classifier tool that can determine whether an audio was generated by AI. The goal of this tool is to combat deepfakes, which are becoming increasingly common in the world of AI. The author excitedly announced a new authentication tool that uses AI speech classification to identify whether an audio sample contains AI-generated audio. They discussed the importance of taking a proactive stance against the malicious use of AI, as well as the impact of deepfakes and misinformation.
