AI News

Vay has launched a teledriving car service in Las Vegas, where remote drivers control vehicles via an app, enhancing privacy and safety compared to traditional ridesharing. Users rent electric Kia Niro vehicles for affordable rates, while the company prioritizes technology and regulatory support for urban mobility.

Tokyo-based Sakana AI, co-founded by ex-Google engineers, raised $30 million in seed funding to develop smaller, more efficient AI models inspired by natural phenomena. Supported by top investors, Sakana aims to redefine AI development and enhance Japan's role in the tech landscape.

Synopsys plans to acquire Ansys for $35 billion, combining EDA expertise with simulation capabilities to enhance innovation in silicon-to-systems design, particularly for AI applications. The deal is expected to increase Synopsys’ market reach significantly and is pending regulatory approval, with a target closure in mid-2025.

Since 2023, generative AI, especially ChatGPT, has transformed industries like agriculture and manufacturing, potentially boosting global GDP by 7%. Over 80% of businesses plan to adopt AI-generated content, with investment soaring to $23.2 billion, leading to over 1,500 startups and innovative tools reshaping productivity.

Baotou Goose Intelligent Technology raised nearly $10 million for its AI agent, BetterYeah AI, founded by ex-Alibaba member Zhang Yi. The company addresses enterprise needs, improving efficiency and sales through integrated solutions. With tools for quick deployment, it positions itself as a leader in the AI agent market.

NeurIPS 2023 in New Orleans featured 12,343 submissions, with a 26.1% acceptance rate. Ant Group excelled, submitting 20 papers on AI topics like generative AI, forecasting, and 3D image generation. Their insights push boundaries in AI, reflecting the global industry's competitiveness.

OpenAI’s upcoming GPT-4.5 is expected to enhance GPT-4's capabilities, possibly featuring more parameters, better memory retention, and improved functionalities like web searches and video analysis. Its release timeline is uncertain, with speculation for late 2023. GPT-5 is not currently in development.

A new CNN architecture, UniRepLKNet, developed by Tencent AI Lab and the Chinese University of Hong Kong, surpasses Transformers in image recognition and processing speed. Its large-kernel design adapts easily to various data modalities, showcasing impressive performance across tasks, including time series predictions.

In October, Apple and Columbia University launched Ferret, an AI model and framework focusing on academic applications. Since then, Apple has advanced AI significantly, introducing tools for running large models on devices. Ferret excels in multi-modal tasks, showing impressive spatial reasoning and fewer hallucinations compared to competitors. The AI community is now recognizing this innovation, with anticipation for future integrated services on iOS.

EMNLP 2023 will take place in Singapore from December 6-10, featuring 1,047 accepted papers from 4,909 submissions. The WeChat AI team achieved eight accepted papers and won the Best Long Paper Award for a study on in-context learning mechanisms, marking a historic first for China. Their work emphasizes collaboration and practical application in AI research.

Google DeepMind's FunSearch, a novel tool using large language models, has addressed a major unsolved mathematical problem, the "hat set" problem, by generating new solutions not found in existing data. This innovation showcases the potential for large language models to contribute to pure mathematics, expanding their role beyond mere content generation.

Google's AI music tool, MusicFX, launched on December 14, enables users to create original compositions with text prompts while addressing copyright issues through watermarking technology. Available via AI Test Kitchen, it aims to democratize music creation and provoke discussions on originality and ownership in AI-generated content.

A recent report from Getty Images urges businesses to carefully evaluate the risks associated with using AI-generated images in their marketing strategies to avoid potential consumer backlash. The "Building Trust in the Age of AI" report surveyed 30,000 adults across 25 countries through four extensive studies conducted between July 2022 and September 2023.

Google has enhanced its Shopping service with AI features like Gemini, which offers personalized product recommendations and summaries based on searches. Users can apply filters for size and availability, use virtual try-ons, view a new "Deals" page for discounts, and manage privacy settings. Rollout starts in the U.S. today.

The startup "Zero One Everything," founded by Kai-Fu Lee, has launched its official website, aiming to create a global AI 2.0 platform to enhance productivity and generate economic value. They foresee significant opportunities surpassing the mobile internet era and plan innovative AI-first applications.