New text-to-image models have advanced, enabling revolutionary applications like creating images from text. However, existing approaches struggle to consistently produce content across zoom levels. A study by the University of Washington, Google, and UC Berkeley introduces a text-conditioned multi-scale image production method, allowing users to control content at different zoom levels through text prompts. The…

Neural Radiance Fields (NeRF) use neural networks to render detailed 3D scenes without explicit 3D model storage. However, they are limited in dynamic scenes. Shanghai Tech University proposes VideoRF, a real-time streaming solution for dynamic radiance fields on mobile devices. It leverages novel neural modeling and deferred rendering to enable seamless viewing experiences. The approach…

In late November 2023, following Sam Altman’s dismissal from OpenAI, Microsoft’s proposal to employ the entire OpenAI team was met with little enthusiasm. Employees cited concerns about corporate culture, financial losses, and the bureaucratic nature of Microsoft. They saw Microsoft as a less dynamic company, preferring to seek opportunities with other AI startups.

NeurIPS, the world’s largest AI conference, will occur in New Orleans from December 10-16, 2023. Google DeepMind teams will present over 150 papers.

Gemini AI, an advanced NLP model, is designed to exceed current benchmarks due to its multimodal capabilities, scalability, and potential for integration with Google’s ecosystem, marking a substantial advancement in AI technology.

Meta is rolling out over 20 generative AI updates to its platforms, introducing features like AI-enhanced search, invisible watermarking, and improvements to Meta AI. This update boosts user experience in areas such as messaging, social media interaction, and content creation, with further advancements expected in the upcoming year.

The “KnowNo” model teaches robots to ask for clarification on ambiguous commands to ensure they act correctly and minimize unnecessary human interaction. It combines language models with confidence scores to determine if intervention is needed. Tested on robots, it achieved consistent success and reduced the need for human aid.

Neosync is an open-source platform helping software development teams anonymize and generate synthetic data for testing while maintaining data privacy. It connects to production databases to facilitate data synchronization across environments and offers features like automatic data generation, schema-based synthetic data, and database subsetting. With its GitOps approach, asynchronous pipeline, and support for various databases…

MIT researchers developed an automated onboarding system that improves human-AI collaboration accuracy by training users when to trust AI assistance. Their method uses natural language to teach rules based on the user’s past interactions with AI, leading to a 5% improvement in image prediction tasks.

Generative AI in academia spurs debate without clear answers on its role, plagiarism, and permissible use. A study shows students and educators divided, seeking policy clarity. Concerns include detection of AI use, the risk of mental enfeeblement, equitable access, and the potential for false positives in AI-written work detection.