• Posts
  • Google Enhances Work with AI Agents

Google Enhances Work with AI Agents

PLUS: Google, MIT's Free AI Course for Teachers, Apple Unveils AI-Driven M4 Processor Line and more.

Sponsored by

Artificial Intelligence online short course from MIT

Study artificial intelligence and gain the knowledge to support its integration into your organization. If you're looking to gain a competitive edge in today's business world, then this artificial intelligence online course may be the perfect option for you.

  • Key AI management and leadership insights to support informed, strategic decision making.

  • A practical grounding in AI and its business applications, helping you to transform your organization into a future-forward business.

  • A road map for the strategic implementation of AI technologies in a business context.


  • Google Enhances Work with AI Agents

  • Meta AI's OpenEQA Boosts Real-World AI

  • Andrew Ng Joins Amazon Board 

  • Apple Unveils AI-Driven M4 Processor Line

  • Google, MIT's Free AI Course for Teachers

  • Adobe Pays for Videos to Train AI

Google Announces STUNNING AI Agents | Google Cloud Keynote AI Agents

In the latest Google Cloud Next keynote, Google announced its heavy investment in AI agents designed to revolutionize various sectors including customer service, marketing, driving, content creation, cybersecurity, coding, and data analysis. 

The agents are tailored for various tasks, from simplifying customer interactions to assisting employees with mundane tasks, improving software development efficiency, and enhancing data analysis capabilities. Additionally, Google introduced security agents to safeguard against the increasing threat of cyberattacks, showcasing their commitment to advancing AI technology across the board.

Meta AI releases OpenEQA to spur ’embodied intelligence’ in artificial agents

Meta AI has launched OpenEQA, an open-source dataset to evaluate AI's ability for "embodied question answering," understanding real-world environments to respond to natural language queries. It features 1,600+ questions covering 180+ settings like homes and offices, testing AI in areas such as object recognition and spatial reasoning. Even top models like GPT-4V struggle with OpenEQA. 

The project merges robotics, computer vision, and language AI, aiming for AI assistants in AR glasses or mobile robots aiding in daily tasks. Questions were generated from real-world videos, assessing perception and reasoning. OpenEQA aims to push AI's real-world understanding.

Amazon adds AI heavyweight Andrew Ng to board of directors as Jassy says it may be biggest tech breakthrough ‘perhaps since the internet’

Amazon appoints AI expert Andrew Ng to its board of directors, amidst fierce AI competition. Ng, known for leading AI teams at Baidu and Google, replaces Judy McGrath. He founded the AI Fund, investing in AI startups. 

Amazon CEO Andy Jassy sees generative AI as pivotal, comparing its impact to cloud computing and the internet in his shareholder letter. This move reflects Amazon's strategic focus on AI's transformative potential.

Apple Plans to Overhaul Entire Mac Line With AI-Focused M4 Chips


Apple plans to revamp its Mac lineup with M4 processors focusing on AI capabilities, reports Bloomberg's Mark Gurman. The new chips are set to debut later this year, aiming to counter rival AI PCs and boost Mac sales. The M4 chips will span three tiers: Donan for entry-level devices like MacBook Air, Brava for higher-specced models, and Hidra for the Mac Pro. 

Apple may reveal more AI features at its Worldwide Developers Conference in June. This strategic move aligns with Apple's commitment to innovation and staying competitive in the tech market.

Google and MIT launch a free generative AI course for teachers

Google and MIT introduce a free course, "Google Generative AI Educators," aiming to empower middle and high school teachers in utilizing generative AI tools. The course, developed by MIT RAISE, offers insights on integrating AI to streamline tasks like email writing, lesson planning, and creating assessments. 

It consists of five modules accessible online, with endorsements from various school districts like Anaheim Union High School District and Chicago Public Schools. The initiative reflects the growing importance of AI in education, enhancing both teachers' workflow and students' learning experiences.

Adobe Is Buying Videos for $3 Per Minute to Build AI Model

Adobe cloud icon in inflated and 3d form

Adobe is investing in video procurement at $3 per minute to develop its AI text-to-video generator, following OpenAI's lead. They seek footage of everyday actions and emotions from their network of photographers and artists to train their AI models. 

This move reflects Adobe's efforts to stay competitive in the AI space and harness the potential of AI-generated content.


New method, Infini-attention, scales Transformer-based Large Language Models (LLMs) for infinite inputs with limited memory and computation. It combines compressive memory, masked local attention, and long-term linear attention in a single block. Tested on long-context tasks, it enables fast streaming inference with minimal memory parameters.

The RULER benchmark assesses long-context language models beyond simple retrieval tasks like the needle-in-a-haystack test. It offers flexible configurations for sequence length and task complexity, including multi-hop tracing and aggregation. Despite claims of handling large context sizes, most models show performance drops with longer contexts. Notably, only four models maintain good performance at 32K tokens.

RealmDreamer generates 3D scenes from text descriptions using a novel approach. It optimizes a 3D Gaussian Splatting representation to match text prompts, incorporating inpainting and depth diffusion models. Unlike existing methods, it doesn't need video or multi-view data, producing high-quality scenes with various styles and objects from single images.

This study explores adapting LLaMA decoder architecture, originally for language models, to vision tasks. To overcome challenges like attention collapse, they reposition the class token and introduce a soft mask strategy. The resulting model, iLLaMA, achieves competitive performance on ImageNet with fewer parameters and exhibits reliability across various tasks.

DreamScene360 introduces a text-to-3D 360-degree scene generation system for creating immersive virtual environments rapidly. It leverages a 2D diffusion model and self-refinement to generate a coherent panoramic image, then transforms it into 3D Gaussians for real-time exploration. Semantic and geometric constraints enhance scene consistency and completeness, improving the immersive experience.


Wiser - Pinterest for knowledge.

HeyGen - flawlessly track, translate, and lip-sync your video with any input text.

Jumprun - AI-powered research as stunning, interactive canvases.

AI Chatbot by Zapier - Create an AI chatbot that answers questions, resolves issues, and nurtures leads with the power of automation.

Tone - AI-enabled pendant that acts as your second brain, distilling your lived experiences into a personal AI to give you superpowers.



Snowflake Copilot, a Mistral Large-powered AI assistant, launches in public preview

Snowflake launches Copilot, an AI assistant for SQL queries, in public preview. It uses Mistral Large to generate SQL queries from natural language inputs, simplifying data analysis. Snowflake aims to gather user feedback to refine the assistant for broader use. Similar tools exist, but Snowflake's approach could streamline decision-making with faster insights. VENTUREBEAT

Learning agile soccer skills for a bipedal robot with deep reinforcement learning

A team of researchers explored teaching a low-cost robot soccer skills using deep reinforcement learning. They trained a robot to play one-on-one soccer, showcasing dynamic moves like fast recovery from falls, agile walking, and tactical gameplay like blocking shots. The trained robot outperformed scripted controllers, hinting at potential for complex robot interactions. SCIENCE ROBOTICS

TikTok may add AI avatars that can make ads

TikTok is planning to introduce AI-generated avatars to promote and sell products on its platform, potentially competing with human influencers' sponsored content. The feature allows virtual influencers to read scripts from advertisers or TikTok Shop sellers. While still in development, TikTok aims to integrate AI creators alongside human ones, though details on revenue sharing and user reception remain unclear. THE VERGE

Simbian brings AI to existing security tools

Simbian, founded by Ambuj Kumar and Alankrit Chona, integrates AI into cybersecurity, streamlining operations for stressed security teams. Using a chatbot interface, it orchestrates various security tools, reducing alert fatigue. Despite privacy concerns, Simbian secured $10 million investment, aiming to expand and enhance its AI-driven platform for better cybersecurity management. TECHCRUNCH

AI model has potential to detect risk of childbirth-related post-traumatic stress disorder

Researchers have developed an AI model to detect signs of childbirth-related post-traumatic stress disorder (CB-PTSD) by analyzing short narrative statements from postpartum patients. With further refinements, it could efficiently identify those at risk, potentially aiding rapid diagnosis and treatment. This NIH-funded study aims to address disparities in CB-PTSD diagnosis. NATIONAL INSTITUTE OF HEALTH

Alphabet Heads Toward $2 Trillion With Investors Cheering AI Progress

Alphabet Inc. inches closer to a $2 trillion market value as investors gain confidence in its artificial intelligence (AI) strategy. Recent glitches had sparked concerns, but a successful cloud event showcasing enterprise-ready AI models has revived optimism. Alphabet's path to AI monetization remains unclear, but its stock's relatively low valuation keeps it attractive. YAHOO!

OpenAI’s Altman Pitches Global AI Coalition on Trip to Middle East

OpenAI's CEO, Sam Altman, seeks to rally global support for AI infrastructure development, focusing on chips, energy, and data centers. Meetings in UAE and Washington aim to engage private and government sectors. Concerns over chip supply prompt Altman's outreach, aiming for multibillion-dollar investments while navigating geopolitical scrutiny. BLOOMBERG

The Weather Channel’s parent company has a new AI tool to make hyperlocal weather videos

The Weather Company introduces ReelSphere, an AI tool for creating hyperlocal weather videos. It automatically adds captions, graphics, and local weather info, with an option for AI-generated voiceovers. Aimed at simplifying weather updates for broadcasters, it integrates with Max, the company's weather system. Human editing is available for corrections. THE VERGE

What'd you think of today's edition?

Login or Subscribe to participate in polls.

What are MOST interested in learning about AI?

What stories or resources will be most interesting for you to hear about?

Login or Subscribe to participate in polls.

Join the conversation

or to participate.