Trending AI news stories + papers

Today's Top AI news stories and papers

AI news stories

1. Google debuts powerful Gemini generative AI model

   Google has introduced Gemini, its new generative AI model, positioning it as a formidable competitor to OpenAI's GPT-4, backed by Microsoft. Gemini is a multimodal model capable of processing text, audio, video, images, and code, designed from the start to handle these varied inputs integrally. It demonstrates superior multimodal understanding and can aid in diverse tasks like identifying objects, analyzing hand-drawn images, and assisting with homework. Google has released three versions of Gemini: Ultra, Pro, and Nano, with varying capabilities and applications, including integration into Google products like Search, Chrome, and Ads. Gemini marks Google's strategic move in the AI space against OpenAI and Microsoft.

2. AMD rolls out new rival to Nvidia’s AI chip

   AMD, the chip designer, has introduced its MI300X chip, aiming to challenge Nvidia's dominance in the artificial intelligence processor market, which is projected to reach $400 billion by 2027. AMD claims its MI300X is the "most advanced AI accelerator in the industry" and expects it to be its fastest product to reach $1 billion in sales, possibly by mid-2024. This move comes as demand for AI chips grows rapidly, with Nvidia's H100 chip leading the market. OpenAI, Microsoft, and other companies plan to incorporate AMD's chips into their AI workloads, making it a significant competitor to Nvidia.

3. McDonald’s will use Google AI

   McDonald's is partnering with Google to introduce generative AI technology in its operations, starting in 2024. This initiative will involve hardware and software upgrades in thousands of stores, along with improvements to ordering kiosks and the mobile app. The goal is to leverage generative AI to optimize operations, potentially resulting in hotter and fresher food for customers.The specific applications of AI are not detailed, but it will focus on reducing business disruptions and enhancing the overall customer experience.

4. Meta will let you ‘reimagine’ your friends’ AI-generated images

   Meta has unveiled updates to its AI tools, including a feature called "reimagine" that allows users to remix AI-generated images shared in Messenger or Instagram chats by adding text prompts. This tool, powered by Meta's Emu image foundation model, will be accessible on the web at imagine.meta.com for users in the US. Additionally, Meta plans to introduce "invisible watermarking" to images created with its AI tools in the future. Other improvements include enhanced responses from Meta's AI assistant on mobile devices, more accurate search result summaries, and the availability of Meta's AI characters in WhatsApp, Messenger, and Instagram in the US. These updates are part of Meta's efforts to compete with Google's recent Gemini news in the AI space.

5. AI could pose new risks to UK financial stability, Bank of England warns

   The Bank of England is closely examining the use of AI in the financial sector due to concerns that its widespread adoption could pose risks to UK financial stability, including increased cyber-threats and group behavior influencing financial decisions. Governor Andrew Bailey emphasizes the need to embrace AI cautiously while acknowledging its potential for economic growth. The Bank and financial regulators are set to release a consultation paper addressing risks associated with "critical third parties" providing data and AI models.

AI Repos

1. THUDM / ChatGLM-6B

   ChatGLM-6B is a language model optimized for Chinese QA and dialogue with 6.2 billion parameters, offering efficient deployment and customization options for various applications.

2. modelscope / modelscope-agent

   ModelScope-Agent, a general and customizable agent framework for real-world applications, based on open-source LLMs as controllers, offering comprehensive customization, diverse APIs, and practical utility for various tasks.

3. openai/evals

   Evals is a framework for evaluating LLMs (large language models) or systems built using LLMs as components. It also includes an open-source registry of challenging evals.