Your daily AI dose

5 Reasons to join Mindstream

• We’re the only AI newsletter you need

• We’re so good HubSpot bought us (like they bought The Hustle)

• 150,000+ strong community staying ahead of the curve

• We’re actually fun to read

• Written by an awesome team of real people, not AI tools

Click here to subscribe

P.S - you get a load of free stuff when you subscribe

Today:

• Best AI Learning Platforms 2025

• Elon Musk Targets OpenAI Shift

• Japan Invests $9.9 Billion in AI

• Alibaba Unveils Marco-o1 AI

Best Resources and Courses For AI and Machine Learning In 2025

Artificial Intelligence and machine learning are becoming essential in all jobs and industries. To enhance your skills, consider these resources:

1. GitHub: A free website where people share computer code. You can find AI projects to learn from and contribute to, gaining practical experience.

2. Simplilearn Courses: They offer classes starting from basic programming in Python to advanced topics like building your own AI models. Courses include real-world projects and are partnered with top universities and companies.

These resources cater to beginners and those seeking structured learning, helping you advance in the growing field of AI.

WATCH THE VIDEO ON YOUTUBE

Elon Musk Targets OpenAI Shift

Elon Musk is suing OpenAI to stop its change into a profit-driven company. Musk’s lawyers argue that if OpenAI becomes for-profit, it might not have enough money to pay any damages Musk could win in court. They also claim OpenAI and Microsoft are breaking US laws by preventing investments in competing companies and using sensitive information improperly. 

OpenAI says these accusations are false and have no basis. The lawsuit follows OpenAI’s recent moves to become a for-profit business and discussions with regulators about this change. Musk has made several similar legal attempts before.

Japan Invests $9.9 Billion in AI

Japan is investing an additional ¥1.5 trillion (about $9.9 billion) this year to boost its chip and artificial intelligence sectors. The government’s extra budget until March allocates ¥1.05 trillion for developing advanced chips and quantum computers, which use the principles of quantum physics to perform complex calculations. 

Another ¥471.4 billion will support the production of high-tech chips within Japan. Part of these funds may go to Rapidus Corp., a new company focused on chip manufacturing, although the exact amount is yet undecided. This investment aims to strengthen Japan’s technology industry and enhance its position in the global market.

Alibaba Unveils Marco-o1 AI

Alibaba has launched Marco-o1, a new language model designed to think more deeply and solve complex problems better than previous models. Building on their existing Qwen2-7B-Instruct model, Marco-o1 uses methods like chain-of-thought and Monte Carlo Tree Search (MCTS) to explore different ways to answer questions. 

It also has a feature that lets it check and correct its own mistakes. Trained with various datasets, Marco-o1 performs exceptionally well in math problems and translating informal language. This advancement is part of a broader movement to create smarter AI systems capable of handling tasks without clear answers, despite challenges in expanding model sizes.

🧠RESEARCH

Critic-V: VLM Critics Help Catch VLM Errors in Multimodal Reasoning

Critic-V, a framework enhancing vision-language models' reasoning by separating response generation (Reasoner) from feedback critique (Critic). Using reinforcement learning and text-based critiques, it improves reasoning accuracy and efficiency. Critic-V surpasses existing models on benchmarks, offering a robust solution for complex tasks like autonomous driving and intelligent systems.

TryOffDiff: Virtual-Try-Off via High-Fidelity Garment Reconstruction using Diffusion Models

VTOFF, a task to extract detailed garment images from photos, differing from virtual try-on by focusing on high-fidelity garment reconstruction. Using TryOffDiff, a diffusion-based model, it achieves superior results with minimal processing. This approach enhances e-commerce imagery, generative model evaluation, and advances reconstruction quality assessment using the DISTS metric.

ChatGen: Automatic Text-to-Image Generation From FreeStyle Chatting

ChatGen, a framework for automating text-to-image generation through freestyle chatting, eliminating tedious prompt crafting and model selection. It includes ChatGenBench for benchmarking and ChatGen-Evo, a strategy enhancing automation skills. The approach improves image generation quality and usability, advancing accessibility and efficiency in creative tasks.

Free^2Guide: Gradient-Free Path Integral Control for Enhancing Text-to-Video Generation with Large Vision-Language Models

Free²Guide, a framework for improving text-to-video generation alignment without extra model training. It uses gradient-free methods and path integral control to integrate large vision-language models as reward systems. This approach enhances text alignment and video quality while supporting flexible reward model combinations with minimal computational cost.

AfriMed-QA: A Pan-African, Multi-Specialty, Medical Question-Answering Benchmark Dataset

AfriMed-QA, a Pan-African medical question-answering dataset with 15,000 questions spanning 32 specialties across 16 countries. It evaluates 30 large language models (LLMs), revealing underperformance in biomedical-specific tasks and geographic variability. While LLMs lag behind USMLE standards, user preferences lean toward LLM-generated answers over clinician responses, highlighting their potential in low-resource healthcare settings.

🛠️TOP TOOLS

Secureframe - Ai-powered platform that automates compliance and improves security.

Magicroll - Your AI-Powered Creative Studio

WiziShop - Automate Your ECommerce Store with AI

OfferGenie - Your AI-Powered Interview Copilot

Sana AI - Your AI assistant for work that analyzes, drafts, finds, and automates across your apps.

📲SOCIAL MEDIA

🗞️MORE NEWS

• Canadian news companies, including Toronto Star and CBC, sued OpenAI, claiming ChatGPT unlawfully used their content without consent for profit. They seek damages and demand OpenAI stop using their materials.

• Microsoft research reveals AI agents using large language models (LLMs) can now control software interfaces via natural language, simplifying tasks. While promising, challenges like privacy, security, and reliability hinder widespread enterprise adoption.

• Viz.ai partners with Microsoft to integrate AI-powered diagnostic imaging and care coordination into clinical workflows via Microsoft's Precision Imaging Network. This collaboration enhances radiology efficiency, accelerates treatment, and improves patient care across diverse medical specialties.

• Honda introduced its AI-powered social robot, Haru, at a Spanish hospital to improve children’s well-being. Haru supports emotional care, rehabilitation, education, and companionship, demonstrating AI's potential to enhance healthcare experiences.

• Simulated AI pets are gaining popularity in China, driven by advancements in AI technology and demand for emotional support. Offering low maintenance and interactive features, these robotic pets appeal to consumers but lack real companionship.