Search AI Products and News

Skywork AI's research team recently unveiled SkyReels-A2, a groundbreaking video generation framework pushing the boundaries of controllable video synthesis. This Element-to-Video (E2V) framework generates natural videos from text prompts by composing diverse visual elements (characters, objects, backgrounds), maintaining high fidelity to reference images. SkyReels-A2's core lies in its sophisticated data processing pipeline. The research team designed a comprehensive data construction pipeline.

Last week, around 100 authors protested outside Meta's London headquarters, accusing the American tech giant of using their works without permission to train its AI models. Protesters chanted "Meta, Meta, book thieves," with some holding signs such as "I wanted to write a slogan, but you'll steal it" and "Give us back our books, Zuck," a clear jab at Meta CEO Mark Zuckerberg. Earlier reports suggested Zuckerberg approved the company's use of a dataset called

Amidst the growing prevalence of artificial intelligence, the collaboration between DeepSeek and Tsinghua University has garnered significant industry attention. DeepSeek, a Chinese startup, is renowned for its breakthroughs in low-cost inference models. This collaboration aims to further reduce the training costs of AI models, thereby enhancing operational efficiency. DeepSeek recently launched a new low-cost inference model that has generated considerable market excitement. To further optimize this model, DeepSeek's research team...

Amidst the rapid advancement of Artificial Intelligence (AI), Microsoft is actively promoting AI literacy with its 50-day AI Skills Festival. This event is open to everyone, from beginners to professionals, offering free registration and access to a wealth of AI learning resources. The initiative aims not only to enhance public AI capabilities but also to break a Guinness World Record, making it a fun and practical event. AI is transforming the way various industries operate, particularly in daily office work. Microsoft hopes to...

Microsoft recently unveiled a remarkable research project – WHAMM (World and Human Action MaskGIT Model). This innovative AI model can generate and run the classic game Quake II entirely within the AI model, producing a playable version in real-time. This research, part of Microsoft's Copilot Labs, explores the potential and boundaries of generative AI in interactive media. A revolutionary breakthrough.

A recent joint study by Google, Carnegie Mellon University, and MultiOn explores the application of synthetic data in training large language models. According to Epoch AI, a research institution focused on AI development, currently available high-quality text training data totals around 300 trillion tokens. However, with the rapid advancement of large models like ChatGPT, the demand for training data is growing exponentially, projected to exhaust existing resources by 2026. Therefore, synthetic data is becoming increasingly crucial.

The Wikimedia Foundation has warned of increasing bandwidth strain on its projects caused by AI-powered web crawlers. Representatives noted a 50% increase in bandwidth consumption for multimedia files since January 2024, largely attributed to automated programs harvesting content from Wikimedia's openly licensed image library for AI model training. Wikimedia Foundation staff members Birgit Mueller, Chris Danis, and...

A new study has raised concerns that OpenAI may have used copyrighted O'Reilly Media books to train its latest AI models without permission. The study was released by the AI Disclosures Project, a non-profit organization founded in 2024 by media mogul Tim O'Reilly and economist Ilan Strauss. AI models can be viewed as sophisticated prediction engines that...

Microsoft's Chief Technology Officer, Kevin Scott, highlighted the importance of product managers in training AI agents. According to him, product managers are not only central to product design and development, but also play a crucial role in creating 'feedback loops'. These feedback loops help AI agents continuously learn and improve their ability to perform tasks, better meeting user needs. Kevin Scott points out that the effectiveness of AI systems is heavily reliant on human feedback. Product managers are integral to this process.

A group of prominent authors, including Richard Osman, Kazuo Ishiguro, Kate Mosse, and Val McDermid, have signed an open letter calling on the UK government to hold Meta accountable for using copyrighted books in its AI training. The letter requests that the Secretary of State for Culture, Media and Sport, Lisa Nandy, summon Meta representatives to discuss the matter.

Hubei province in China has released the nation's first price list for brain-computer interface medical services, starting from 6552 yuan.

Last week, The Atlantic launched a new tool to search LibGen, a database allegedly used by Meta to train its AI models. This has raised significant concerns due to the inclusion of numerous unpublished works. Writer Maris Kreizman, in an article for Literary Hub, revealed that her forthcoming essay collection was found in this database. Kreizman stated that her collection is...

Researchers in China have unveiled the world's first three-dimensional atlas of a brain-computer interface at the micrometer scale. This groundbreaking achievement significantly improves the safety and precision of brain surgery involving BCIs.

Bloomberg, citing sources, reports that Ant Group has made a significant breakthrough in AI, reducing AI model training costs by 20% using chips from Alibaba and Huawei. This injects new momentum into China's drive for technological self-reliance. The report further indicates that Ant Group's internal tests show these Chinese-made AI chips rival Nvidia's offerings in performance. Widespread validation and adoption of this could significantly alter the global AI landscape.

Researchers at the University of California, Berkeley have recently released their latest research achievement – the TULIP (Towards Unified Language-Image Pretraining) model. This model aims to enhance the performance of vision-language pre-training, particularly in visually-centric tasks requiring high-fidelity understanding, overcoming the limitations of existing contrastive learning models (such as CLIP). TULIP integrates innovative techniques such as generative data augmentation, enhanced contrastive learning, and reconstruction regularization.