AI research

Summary Meta has released a new version of data2vec: data2vec 2.0. Update of December 29, 2022: Meta’s improved learning algorithm for different modalities is significantly faster than its predecessor. Nearly eleven months after the release of data2vec, Meta’s AI division presents an improved version of its multimodal learning algorithm. With data2vec, it is much easier …

Meta shows an extremely fast learning algorithm for speech, image and text Read More »

Summary AI content is currently in the Wild West phase – you generate what works. There are no transparency rules. China and OpenAI are working on countermeasures. Besides the issue of copyright on training content for AI models, another fundamental question plagues the AI ​​industry: what about all the AI ​​content that is barely recognizable …

Does AI content need transparency to protect society? Read More »

Summary Photos and AI images are still easy to tell apart – but for how much longer? A new tool further blurs the lines between real and generated images. Sharif Shameem, who created the vast AI image database Lexica Art, unveiled his new project: Lexica Aperture is an image generator that produces realistic images from …

New Image AI “Lexica Aperture” Generates Photorealistic Images Read More »

Summary Low-emission mobility, climate-friendly energy supply, clean industrial production or optimized circular economy – artificial intelligence (AI) can contribute to climate protection in many ways. At the same time, the development and deployment of AI systems is often itself accompanied by a high consumption of resources. Where the technology can best unleash its potential and …

Can artificial intelligence help stop climate change? Read More »

Summary Deepmind shows a language model that learns from different human opinions and suggests consensus statements. A major challenge in deploying large language models is ensuring that the generated text is useful and consistent with human values. Companies like OpenAI rely on Reinforcement Learning with Human Feedback (RLHF) for this. In this process, a language …

Deepmind’s new language model makes compromises Read More »

Summary With Stable Diffusion Version 2.0, Stability AI strives to be more legally compliant and future-proof. Two important changes have been made. First, Stability AI removed naked images from training datasets to limit their generation. According to Emad Mostaque, founder of Stability AI, this is not about censoring nude images per se, but about protecting …

Stable Diffusion v2 removes nude images and causes protests Read More »

Summary Version 2.0 of Stable Diffusion brings many advances. The most important new feature is the improved OpenCLIP text-to-image conversion model. In August 2022, AI startup Stability AI, in collaboration with RunwayML, LMU Munich, EleutherAI and LAION, released Stable Diffusion, an open source image AI that was immediately well received by the community. Stable Diffusion …

Stable Diffusion 2.0 is a completely improved version Read More »

Summary Generative AI can create text, audio, images, and 3D assets. The games industry combines these media – and could therefore benefit greatly. At least that’s what venture capitalist Andreessen Horowitz (a16z) assumes, making him the second major Silicon Valley investor after Sequoia Capital to embrace the potential of generative AI. In a blog post, …

a16z thinks a creative AI revolution is coming to gaming Read More »

Summary Currently, Stable Diffusion is just a powerful AI image generator. Long-term plans go way beyond that. Generative AI is clearly on the rise lately: from text to image, from text to HD video or from text to 3D, AI systems can create more and more media formats, some of which are fully automated. New …

Stable Diffusion lead plans an open-source Holodeck Read More »

Summary Generative AI models such as Stable Diffusion can generate images, but struggle to edit them. Google shows a new method that allows more control. With OpenAI’s DALL-E 2, Midjourney, or Stable Diffusion, interested parties have a variety of generative text-to-image conversion models to choose from. All models produce believable images and can be checked …

Stable streaming becomes easier to control with a new Google method Read More »