There’s a reason 400,000 professionals read this daily.

Join The AI Report, trusted by 400,000+ professionals at Google, Microsoft, and OpenAI. Get daily insights, tools, and strategies to master practical AI skills that drive results.

Sign up now for free and work smarter, not harder.

Today:

• OpenAI, Andrew Ng Launch Reasoning Course

• Microsoft Unveils Task-Performing AI

• KoBold Secures $2.96bn Valuation

• Nvidia Invests $1B in AI Startups

• Microsoft Bets Big on AI Productivity

OpenAI, Andrew Ng Launch Reasoning Course

DeepLearning.AI launched a free course called "Reasoning with o1" in collaboration with OpenAI. The course teaches how to maximize the o1 AI model, which focuses on thoughtful and accurate responses by generating multiple reasoning steps. OpenAI also announced the upcoming release of the improved o3 model today. Andrew Ng, founder of DeepLearning.AI, emphasized the course’s value in enhancing skills for advanced AI tasks like coding and problem-solving.

Microsoft Unveils Task-Performing AI

Microsoft has developed a "Large Action Model" (LAM), an AI that can autonomously operate Windows programs like Microsoft Office. Unlike traditional language models that only generate text, LAMs execute tasks by understanding user inputs and performing actions in real-time. In tests, a LAM based on Mistral-7B succeeded 71% of the time and was faster than GPT-4o. While promising for advancing AI capabilities, challenges like security and scalability remain.

KoBold Secures $2.96bn Valuation

KoBold Metals raises $537mn in funding round, backed by Gates and Bezos, aiming to secure critical minerals for energy transition. Utilizing AI for mineral exploration, KoBold discovers significant copper deposit in Zambia, plans to expand to Finland and Botswana. Company intends to go public in 3-5 years, with strong bipartisan support in the US for diversifying critical mineral supply chains.

Nvidia Invests $1B in AI Startups

Nvidia invested $1 billion in AI startups in 2024, becoming a key supporter in the AI boom powered by its GPUs. This marks an increase from previous years, focusing on core AI companies that also use Nvidia’s chips. Despite major customers like Microsoft and Google developing their own chips, Nvidia aims to foster competition by backing new players. However, its aggressive investments have raised antitrust concerns. Nvidia denies linking funding to technology use, emphasizing ecosystem growth.

Microsoft Bets Big on AI Productivity

In 2024, Microsoft aimed to demonstrate that its AI tools could justify their high costs by boosting business productivity. However, as the year ends, this goal remains unmet and is postponed to the New Year. A recent demonstration of Microsoft’s Copilot AI tool at a Sydney business conference showcased ongoing efforts to integrate AI into workplaces. Microsoft continues to invest in AI to prove its value and enhance business operations.

🧠RESEARCH

Explanatory Instructions: Towards Unified Vision Tasks Understanding and Zero-shot Generalization

This paper introduces Explanatory Instructions to improve zero-shot task generalization in computer vision (CV), inspired by successes in natural language processing. By using intuitive linguistic definitions for CV tasks, it enables a vision-language model trained on 12 million examples to understand and generalize to unseen tasks. Code and data will be open-sourced.

On the Compositional Generalization of Multimodal LLMs for Medical Imaging

This study examines compositional generalization (CG) in multimodal large language models (MLLMs) for medical imaging. By using 106 datasets in the Med-MAT benchmark, it shows how CG enables understanding of unseen images, supports limited data, and improves multi-task training. The findings highlight CG's potential for advancing medical imaging AI. Data is available on GitHub.

Bringing Objects to Life: 4D generation from 3D objects

This paper introduces a method to animate user-provided 3D objects into 4D moving content guided by text prompts. By converting 3D meshes into static 4D Neural Radiance Fields and using a text-driven diffusion model, it preserves object identity while enabling realistic animations. The approach outperforms baselines in motion realism, identity retention, and visual quality.

🛠️TOP TOOLS

DomoAI - AI-powered platform that enables users to transform images and videos into creative formats, including anime, 3D cartoons, and other artistic styles.

GPT Engineer - AI-powered tool designed to streamline software development by generating entire codebases from simple text prompts.

Hotpot AI - AI-powered platform designed to simplify and enhance the creative process by transforming text prompts or images into unique artworks.

Fontjoy - AI-powered font pairing tool that simplifies typography selection for designers and non-designers alike. 

Tripo AI - AI-powered platform that revolutionizes 3D content creation by enabling users to generate high-quality 3D models from text descriptions or images in seconds. 

📲SOCIAL MEDIA