Agile: AI Updates - Week of [09/30/24]

The latest AI newsletter updates every week.

Overview - This week’s AI developments showcased significant progress in various fields, from new models and open-source releases to AI’s growing role in industries like weather prediction and protein design. Nvidia, Meta, and Mistral launched cutting-edge models, while Google DeepMind pushed

advancements in biotechnology. OpenAI’s DevDay introduced features to further democratize AI for devs.

0:00

1. Nvidia Unveils NVLM 1.0

• An open-source AI model to rival GPT-4, designed for multimodal tasks like vision and language. This marks a significant step toward Nvidia's ambition to dominate conversational AI and NLP applications.

2. OpenAI DevDay 2024 Highlights

• Four major feature updates from OpenAI, including vision fine-tuning for GPT-4o and a realtime API for speech-to-speech applications. These tools aim to make AI more accessible and efficient across industries​.

  
  

3. Meta’s Llama 3.2 Release

• A new suite of vision and text models with up to 90 billion parameters. Designed for edge and mobile devices, it supports advanced tasks like summarization and image understanding​.

4. Mistral Introduces Pixtral 12B

• Its first multimodal model capable of processing both text and images for tasks like image captioning. This release follows the company’s $645 million funding round​.

5. Google DeepMind announced AlphaProteo

• An AI system designed to generate protein binders for biological research. It promises to revolutionize drug development by significantly improving binding success rates​.

6. Microsoft’s Copilot Gets an AI Upgrade

• Microsoft upgraded its AI-powered Copilot, embedding it across Microsoft 365 applications to assist users with tasks like writing, data analysis, and image generation. The tool enhances productivity by responding to natural language commands​.

7. Qwen2.5-Math Sets New Standards

• An open-source mathematical AI model, achieved breakthroughs in solving complex math problems. It now outperforms models like GPT-4 on challenging benchmarks, setting a new standard for mathematical reasoning​.

8. John Snow Labs Automates Responsible AI Testing

• John Snow Labs introduced a no-code tool for evaluating the safety and efficacy of custom AI models. This tool enables non-technical users to ensure their AI models meet safety standards.
  

9. Salesforce’s ‘Agent Force’ Enhances Productivity

• Salesforce launched ‘Agent Force’ to automate repetitive tasks, allowing employees to focus on higher-value work. The tool aims to boost productivity and improve job satisfaction across industries.

10. Anthropic’s Claude Enterprise Expands AI Access

• Anthropic launched its Claude Enterprise plan, providing businesses with enhanced AI solutions for handling proprietary data. This positions Anthropic to compete more aggressively with OpenAI in the enterprise space​.