Today:

• Fully Autonomous Robot

• Microsoft's $2.2B Boost for Malaysia's Tech

• MongoDB & Amazon Bedrock AI Integration

• Oracle & Microsoft Cloud Alliance

• Apex Launches AI Security

• IBM-AWS Collaboration 

China's FULLY AUTONOMOUS ROBOT Triggers Fears | Is it REAL?

The robot showcased in a demo video, impresses with its ability to cook, clean, and serve drinks without human control. Developed within a year by a company founded in December 2022, the robot integrates advanced technologies like object recognition and a large language model. 

It operates autonomously, showcasing skills from meal preparation to email sorting, emphasizing the rapid progress in integrating vision, language, and action models in robotics. This development could see the robot in homes by 2024, highlighting a significant advancement in robotic capabilities and autonomous household assistance.

WATCH THE VIDEO ON YOUTUBE

Microsoft announces US$2.2 billion investment to fuel Malaysia’s cloud and AI transformation

Microsoft has announced a substantial $2.2 billion investment to accelerate Malaysia's digital growth over the next four years, marking the largest investment in the region in its 32-year history. This initiative includes establishing new cloud and AI infrastructure, enhancing cybersecurity, and providing AI training for 200,000 Malaysians. 

The investment aims to make Malaysia a central hub for cloud computing and advanced technologies, boosting the nation’s economic growth, competitiveness, and digital capabilities. Satya Nadella, Microsoft’s CEO, emphasized the commitment to ensuring that these advancements benefit all Malaysians, supporting a broad and inclusive digital transformation.

MICROSOFT

MongoDB’s Atlas Vector Search Knowledge Bases is now generally available for Amazon Bedrock

MongoDB has publicly launched its Atlas Vector Search integration with Amazon Bedrock, enhancing the ability for developers to utilize foundation models and AI agents with MongoDB's data storage. This partnership, first revealed at Amazon Re:Invent, enables more precise and personalized AI responses through Retrieval Augmented Generation (RAG). It aims to address common concerns about data accuracy and security in AI applications by allowing users to integrate their own data securely. 

The integration also facilitates the creation of generative AI applications without manual data handling, promoting efficiency and tailored customer interactions. This move builds on MongoDB’s ongoing collaborations with AWS and its commitment to expanding AI capabilities for enterprise clients.

VENTUREBEAT

Oracle Looks to AI and Microsoft Partnership to Lift Cloud Business

Oracle is enhancing its flagship database software by integrating artificial intelligence (AI) features to bolster its cloud computing sales. This update, part of Oracle's Database 23ai service, allows businesses to query their data using plain English. The enhancement is designed to increase Oracle's cloud usage, particularly as it competes with larger cloud services like Amazon's and Microsoft's. 

Additionally, Oracle is expanding its collaboration with Microsoft, enabling faster access to Oracle databases through Microsoft Azure. This strategic move underscores Oracle's push to make its cloud platform more user-friendly and affordable compared to its competitors. The company reports that its cloud revenue now exceeds its traditional software licensing revenue, signaling a significant shift in its business operations.

THE WALL STREET JOURNAL

Sam Altman-backed Apex emerges from stealth with $7M to help secure AI adoption

The Israeli company Apex has launched a security platform designed to provide comprehensive oversight of AI activities within organizations. The platform allows companies to establish and enforce AI usage policies, detect and respond to AI threats, and prevent data breaches and other security incidents. Apex emerged from stealth with a $7 million seed funding round led by Sequoia Capital and Index Ventures, with significant backing from notable figures including OpenAI's CEO, Sam Altman. 

This platform is particularly relevant given the growing integration of AI in business, which increases exposure to potential cyberattacks and security challenges. Apex aims to enable secure and rapid AI adoption in major organizations, addressing the hesitation caused by security concerns.

CTECH

IBM brings its AI, data and automation software to AWS Marketplace in 92 countries

IBM has significantly expanded its software offerings on the Amazon Web Services (AWS) Marketplace, now available in 92 countries, up from only five previously. This move is part of a strategic collaboration between IBM and AWS aimed at boosting the adoption of hybrid cloud and AI technologies globally. Nick Otto, IBM's Head of Global Strategic Partnerships, highlighted the hybrid approach to generative AI and the integration of IBM’s AI and data products with AWS capabilities. The expanded availability includes 44 IBM products focusing on AI and data technologies like Watson x.data and Watson x.ai. 

This partnership simplifies the purchasing process for customers, allowing them to use their existing AWS committed spend to procure IBM’s solutions. This collaboration not only facilitates easier consumption of hybrid cloud and AI capabilities but also enhances the competitiveness of both IBM and AWS in the global market.

VENTUREBEAT

🧠RESEARCH

A Careful Examination of Large Language Model Performance on Grade School Arithmetic

Large language models (LLMs) perform well in math benchmarks but may be influenced by prior data exposure rather than actual reasoning. A new test, GSM1k, showed up to 13% accuracy drops in some models, suggesting possible data memorization issues, although top models displayed minimal overfitting.

Is Bigger Edit Batch Size Always Better? -- An Empirical Study on Model Editing with Llama-3

The paper examines the effectiveness of editing techniques on the Llama-3 language model, finding that larger batches of edits reduce performance more than smaller, sequential batches. It suggests a blend of batch and sequential editing could enhance future model editing strategies.

Clover: Regressive Lightweight Speculative Decoding with Sequential Knowledge

The paper introduces Clover, a new speculative decoding algorithm for large language models (LLMs) aimed at improving efficiency. Unlike traditional methods that suffer from high memory transfer times on GPUs, Clover uses sequential knowledge and additional decoding heads to predict multiple tokens at once, significantly enhancing generation efficiency. Results show Clover outperforming existing methods by a wide margin on benchmark tests.

Spectrally Pruned Gaussian Fields with Neural Compensation

The paper introduces SUNDAE, a memory-efficient approach for 3D Gaussian Splatting that enhances rendering quality without excessive memory use. By using spectral pruning to manage Gaussian primitives and a neural network to maintain quality, SUNDAE significantly reduces memory requirements while improving rendering performance, achieving higher quality and speed on benchmark tests compared to traditional methods.

STT: Stateful Tracking with Transformers for Autonomous Driving

The paper introduces STT, a transformer-based model for object tracking in autonomous driving. It effectively tracks objects and predicts their states, like velocity, by processing appearance, geometry, and motion data. STT not only handles data association but also state estimation, improving on existing methods by introducing new metrics, S-MOTA and MOTPS, to better evaluate performance. It shows strong real-time results on the Waymo Open Dataset.

🛠️TOP TOOLS

Film - find the perfect image to express your ideas

Threado AI - Instant, accurate AI assistance for customer support teams

Creatie - Turn ideas into stunning designs in a breeze

RenderNet - Create AI images with character consistency

AutoRepurpose AI - Turn your YouTube videos into text content for Twitter thread, LinkedIn posts, Newsletter, and etc

📲SOCIAL MEDIA

🗞️MORE NEWS

Microsoft made the biggest renewable energy agreement ever to fuel its AI ambitions

Microsoft has entered into the largest corporate renewable energy deal to date, aiming to develop 10.5 gigawatts of new renewable capacity globally to support its AI expansion and climate goals. This $11.5 billion investment will help the tech giant meet its carbon neutrality targets by 2030. THE VERGE

The first music video generated with OpenAI’s unreleased Sora model is here

OpenAI's unreleased Sora model, designed to create high-resolution videos up to 60 seconds, was used by filmmaker Paul Trillo to produce the first official music video for Washed Out's single "The Hardest Part." This marks a significant advancement in AI-driven media, showing Sora's capability to generate 55 unique clips stitched together for a seamless visual experience. VENTUREBEAT

Khosla, VCs Plow $100 Million Into Blood Test Startup Using AI

Karius Inc., a blood-testing startup, has successfully raised $100 million in a new funding round led by Khosla Ventures, 5AM Ventures, and Gilde Healthcare. The company plans to use this investment to expand the adoption of its AI-driven technology, which is designed to detect hard-to-identify infections in patients' blood, across more hospitals. BLOOMBERG

Dropbox, Figma CEOs back Lamini, a startup building a generative AI platform for enterprises

Lamini, a Palo Alto startup, is creating a generative AI platform tailored for enterprises, emphasizing high accuracy and scalability. Backed by $25 million from prominent investors including Andrew Ng, the company offers unique "memory tuning" for precise data recall, aiming to solve common corporate AI implementation challenges. TECHCRUNCH

Abu Dhabi AI firm Presight buys majority stake in technology joint venture AIQ

Abu Dhabi's AI firm Presight acquired a 51% stake in AIQ, a tech venture involving ADNOC and G42, valuing AIQ at $1.4 billion. ADNOC retains 49% and gains a 4% stake in Presight. AIQ uses AI to enhance oil and gas industry processes. This move aligns with Abu Dhabi's strategy to integrate more AI into its operations and expand its global tech influence. CNBC

Over 500 AI models are now optimised for Core Ultra processors, says Intel

Intel announced over 500 AI models are now optimized for Core Ultra processors. These models, sourced from industry leaders like OpenVINO Model Zoo, Hugging Face, and PyTorch, enhance AI capabilities in various applications such as object detection and image classification. This development marks a significant advancement in AI performance on client devices. BUSINESS STANDARD

Mercedes-Benz Invests in AI-Connected Tech Startup Blaize

Blaize, an AI computing company, has raised $106 million to advance AI and ML technologies across multiple sectors. The funding will support Blaize's expansion, enhance product offerings, and speed up the deployment of AI solutions. Focusing on high performance and energy efficiency, Blaize targets industries like automotive and healthcare with its AI platforms, aiming to boost operational efficiency and innovation. The investment round included contributions from existing investors like Mercedes-Benz and new participants like Rizvi Traverse. PYMNTS

Versus launches Vivi, an AI platform transforming passive content into interactive experiences

Versus has introduced Vivi, an AI platform designed to convert passive content into interactive experiences, like games and polls, in real time. This innovation enhances user engagement and retention, providing benefits for various industries. John Vitti, CEO of Versus, emphasized Vivi's quick, content-aware capabilities, which can generate interactive elements in less than a second. Early feedback from partners such as Disney and Microsoft shows high engagement rates, significantly extending session times. VENTUREBEAT