Google AI Mode 🔍, Apple Robot Lamp 💡, DeepSeek VL2 ✨

In partnership with

Your Breakfast Is Holding You Back

Coffee alone isn’t breakfast.

For lasting energy, peak performance, and real health benefits, your body needs more than just empty calories. That’s where Huel Black Edition steps in. Packed with 40g of protein and 27 essential vitamins, it’s the ultimate high-protein meal, ready in just 30 seconds.

Start fueling your day the right way. Use code BEHUEL15 for 15% off your first order, plus a FREE t-shirt and shaker.

Upgrade Your Breakfast Today

💬 OpenAI Developments

OpenAI’s ex-CTO, Mira Murati, has recruited OpenAI co-founder John Schulman

Former OpenAI CTO Mira Murati has recruited OpenAI co-founder John Schulman to join her new AI venture. Schulman, who left OpenAI in August 2024 to join Anthropic, has now departed from Anthropic after just five months to collaborate with Murati.

Researchers created an open rival to OpenAI’s o1 ‘reasoning’ model for under $50

Researchers from Stanford University and the University of Washington have developed an open-source AI model named "s1" that rivals OpenAI's o1 reasoning model. Remarkably, they trained s1 in just 26 minutes at a cost of under $50, utilizing distillation techniques with answers from Google's Gemini 2.0, highlighting the potential for cost-effective development of advanced AI systems.

🌐 DeepSeek Developments

Four Chinese AI startups to watch beyond DeepSeek

DeepSeek, a Chinese AI startup founded in 2023, has developed the DeepSeek-R1 model, which rivals leading AI systems like OpenAI's ChatGPT but at a fraction of the cost, utilizing only about 2,000 specialized chips compared to the 16,000 typically required by Western models. This breakthrough has led to significant market disruptions, including a $600 billion drop in Nvidia's stock value, and has prompted other Chinese companies, such as Alibaba Cloud, Zhipu, Moonshot AI, ByteDance, and Tencent, to accelerate their AI advancements, releasing models that challenge U.S. counterparts in both performance and cost.

DeepSeek’s R1 and OpenAI’s Deep Research just redefined AI — RAG, distillation, and custom models will never be the same

DeepSeek's R1 model delivers industry-leading reasoning capabilities at approximately 30 times lower operational costs compared to OpenAI's o1, offering full transparency in its reasoning processes. This advancement enables developers to build highly customized AI models cost-effectively through techniques like distillation, fine-tuning, or retrieval-augmented generation (RAG), facilitating the creation of smaller, task-specific models that inherit R1's superior reasoning capabilities.

DeepSeek's AI models give fresh boost to China's smart glasses sector to take on Meta

DeepSeek's open-source AI models are providing significant momentum to Chinese smart glasses manufacturers, enhancing their capabilities in areas such as real-time language translation and augmented reality applications. This development is expected to accelerate the integration of advanced AI features into wearable technology, potentially increasing consumer adoption and driving innovation within the industry.

🚀 Tech Industry Moves

Google is testing a new "AI Mode" in Search, powered by a custom version of Gemini 2.0, designed to handle open-ended and exploratory queries that traditional search struggles with, such as advice, comparisons, and follow-up questions. The feature offers AI-generated, organized summaries with links for deeper exploration, presented in a chatbot-like interface on both desktop and mobile. Currently being tested by Google employees in the U.S., AI Mode includes voice input and may launch publicly in 2025, a year CEO Sundar Pichai predicts will be pivotal for Search innovation.

Microsoft's AI division has successfully recruited two top scientists from Google who were behind an advanced AI tool that converts standard text into engaging podcasts. This move underscores Microsoft's ongoing strategy to strengthen its AI capabilities in content creation, positioning itself competitively against Google and OpenAI in the race for AI-driven media generation.

Meta has launched a new program in partnership with UNESCO to collect speech recordings and transcriptions, aiming to enhance its speech and translation AI capabilities. This initiative seeks to improve the accuracy and inclusivity of Meta's AI systems by diversifying the linguistic data used in training.

Nvidia has unveiled its latest AI-driven platform, Drive Thor, designed to significantly enhance the capabilities of autonomous vehicles. By integrating advanced AI algorithms and neural network frameworks, Drive Thor processes vast amounts of data in real-time, enabling self-driving cars to adapt swiftly to various driving conditions and improve decision-making processes. Additionally, the platform incorporates augmented reality features to provide passengers with real-time contextual information, aiming to create a more intuitive and safer user experience.

💡 Emerging AI Technologies and Companies

Figure AI says it can produce 100,000 humanoid robots in 4 years. They will have the company's own AI, not OpenAI models.

Figure AI has announced a partnership with an undisclosed major American company, aiming to deliver 100,000 humanoid robots over the next four years. Notably, Figure AI has decided to develop its own AI models tailored specifically for humanoid robotics, moving away from its previous collaboration with OpenAI due to integration challenges.

Kenyan-Chinese Robotics Partnership Breaking AI and STEM Barriers for Schoolchildren

Afribot Robotics, a Kenyan organization, has partnered with Chinese institutions to establish a robotics lab in Mombasa, Kenya, aiming to enhance AI and STEM education among schoolchildren. This collaboration seeks to break barriers in technology education by providing students with hands-on experience in robotics and artificial intelligence.

🎨 AI in Creative Industries

Apple has unveiled a prototype of an expressive robot lamp, offering a whimsical glimpse into its robotics ambitions. Unlike typical functional robots, this lamp—reminiscent of Pixar’s Luxo—focuses on conveying emotions and intentions through gestures, posture, and gaze, aiming for more natural human-robot interactions. Detailed in Apple’s ELEGNT research paper, the lamp demonstrates human-like behaviors rather than just performing tasks. This playful prototype aligns with reports that Apple is developing mobile home robots, hinting at a future where everyday objects are infused with personality and advanced robotics.

If you want more updates related to AI, subscribe to our Newsletter