Search AI Products and News

In recent years, AI-generated music has flooded the YouTube platform, especially the popular 'lo-fi' music playlists. These playlists often use anime character thumbnails to create a relaxing and immersive atmosphere. However, with the rapid rise of AI music, many playlist creators have started labeling 'No AI' to ensure listeners enjoy truly human-created music. At the end of last year, TikTok music star Derrick Ji (Derric)

The Gemini API has launched an implicit caching feature, offering developers significant cost optimization experiences. This new functionality eliminates the need for developers to manually create caches; when requests share a common prefix with previous requests, the system automatically triggers cache hits, providing up to a 75% discount on tokens. This update now covers the Gemini2.5Pro and 2.5Flash models, further enhancing the cost-effectiveness of AI development. For more details, please refer to the official website: https://develo

The US Department of Labor is investigating Scale AI, a prominent data labeling company, over allegations concerning employee rights and compensation. The investigation follows reports of disputes between the company and its employees regarding fair wages and working conditions.

The National Internet Information Office recently released the 'Measures for Labeling AI-Generated Synthetic Content (Draft for Comments)', aimed at regulating the labeling of AI-generated synthetic content, safeguarding national security and public interests, and protecting the legitimate rights and interests of citizens, legal persons, and other organizations. This measure is formulated based on the 'Cybersecurity Law of the People's Republic of China', 'Regulations on the Administration of Algorithms for Internet Information Services', 'Regulations on the Administration of Deep Synthesis in Internet Information Services', and the 'Interim Measures for the Administration of Generative Artificial Intelligence Services', among other laws and regulations. The deadline for feedback is the 20th.

Recently, Meta announced that it will adjust the way it labels AI-edited content on Instagram, Facebook, and Threads. This change means that the 'AI Information' label indicating whether content has been edited by AI tools will be moved from below the username to the post menu. Previously, users could easily see this label below the username, but now they will need to click the menu to find it. However, Meta also stated that content generated entirely by AI will still be labeled.

Language models have shown widespread applications across various fields, from education to legal consulting, and even in predicting medical risks. However, as these models gain more weight in decision-making processes, they may unintentionally reflect the biases present in the human training data, exacerbating discrimination against minority groups. Research has found that language models exhibit implicit racism, particularly in their treatment of African American English (AAE), demonstrating harmful dialect discrimination that is more negative than any known stereotypes against African Americans. The 'masking disguise' method was used to compare AAE with Standard American English.

AutoAlign is an innovative knowledge graph alignment method designed to address the entity recognition challenges of different knowledge graphs. It constructs predicate proximity graphs using large language models to automatically identify similar predicates, achieving entity alignment between knowledge graphs. Without the need for manual seed alignment, it significantly enhances alignment efficiency and accuracy through predicate alignment, entity alignment, and a joint learning mechanism. In entity alignment tasks, AutoAlign outperforms existing methods and shows potential in applications such as knowledge graph completion.

Researchers from Nanjing University and Megvii Research Institute have collaborated to bring a revolution to visual large models. Their proposed unsupervised paradigm, SeVa, successfully addresses the preference alignment issue of visual language models, and the entire process does not require the involvement of humans or GPT-4, significantly reducing alignment costs.