This Week's Top AI & Tech News and Trends

Through April 28, 2024

Jeita Grotto, Lebanon, by Adam’s Creations

Hello AI Citizens,

This week in The AI Citizen, we explore a variety of intriguing advancements:

  • Synthesia's New Expressive AI Avatars Blur the Lines Between Real and Synthetic

  • China’s New Astribot S1 Humanoid Robot Shocks the Industry

  • Reid Hoffman Unveils REID AI: A Digital Twin Revolutionizing Interaction

  • Meta's Llama 3 Challenges OpenAI's Dominance in AI

  • The Rise of Real-Time Deepfakes: A New Challenge for Online Trust

  • Ray-Ban and Meta Elevate Smart Glasses with AI and AR

  • Moderna Transforms Operations with OpenAI's ChatGPT Enterprise Integration

  • OpenAI vs. Do-It-Yourself Large Language Models

  • Groundbreaking Personal mRNA Vaccines Show Promise in Melanoma Trials

  • AI Begins Exploring the Vast Complexities of String Theory

  • and more.

Let's dive in!

Synthesia’s New Expressive AI Avatars. Source: Synthesia’s Official YouTube Channel

Synthesia's New Expressive AI Avatars Blur the Lines Between Real and Synthetic

Synthesia's latest expressive AI avatars bring us closer than ever to digital doubles with unparalleled realism and expressiveness, launched from their studio in East London. This technological advancement not only enhances digital content creation but also stirs ethical and societal concerns. As these avatars make it difficult to distinguish between real and synthetic media, they pose significant challenges to our trust in digital content. The rise of such convincing synthetic media underscores the importance of rigorous content moderation and ethical practices to maintain the integrity of digital interactions. Source: Technology Review

China's New Astribot S1 Humanoid Robot.

China’s New Astribot S1 Humanoid Robot Shocks the Industry

The newly unveiled 'Astribot S1' by a Chinese research company showcases the future potential of general-purpose robot helpers, with "the most robust operational performance" among its peers and proven versatility in both home and workplace environments. Handling tasks from folding clothes to stir-frying, the S1, enhanced with LLMs, is set to revolutionize everyday life and is on schedule for commercialization by 2024.

Reid Hoffman meets his AI twin.

Reid Hoffman Unveils REID AI: A Digital Twin Revolutionizing Interaction

REID AI, developed by Hour One, closely mimics Reid Hoffman with highly accurate facial expressions and similar looks, offering the ability to endlessly customize his speeches for different audiences or provide personalized advice in a Zoom-like interaction. However, as with all AI technologies, it raises concerns about misuse and ethical implications, prompting discussions about stringent "rules of the road" to protect identity and privacy, and philosophical questions about AI's role in preserving personas posthumously. Source: Reid Hoffman Youtube Channel.

Meta’s Llama 3 instruct model performance. Source: Meta.

Meta's Llama 3 Challenges OpenAI's Dominance in AI

Meta's new open-source model, Llama 3, is rapidly gaining traction, posing a significant challenge to OpenAI's market leadership, which was established with the revolutionary ChatGPT. OpenAI's introduction of ChatGPT spurred a massive surge in AI investment and attracted over 2 million developers to its APIs. However, as models like Llama 3 demonstrate competitive performance, developers and entrepreneurs might shift away from costly proprietary models like those offered by OpenAI and Google, opting instead for open-source alternatives that continue to grow in capability and influence. Source: Wired

DeepFacelive can perform real-time face-swapping during live broadcasts and video calls

The Rise of Real-Time Deepfakes: A New Challenge for Online Trust

The development of tools like DeepFacelive and Microsoft's VASA-1 is transforming online communication. These technologies allow users to alter appearances in real-time during live broadcasts and video calls, creating highly realistic virtual personas from just a static image and audio. This surge in deepfake capabilities heightens the need for stringent authenticity checks to prevent misuse and maintain trust online. As digital interactions become more sophisticated, the emphasis on real-world interactions and secure verification methods becomes more critical, promising a future where virtual meetings are both innovative and secure. Here is an example of DeepFaceLive.

Ray-Ban and Meta are updating their smart glasses with new styles, video calling on WhatsApp and Messenger, and hands-free AI vision assistance. Source: Meta.

Ray-Ban and Meta Elevate Smart Glasses with AI and AR

Ray-Ban and Meta have upgraded their smart glasses, blending AI and augmented reality (AR) to enhance user interaction with their environment. This advanced integration processes and displays information in real time, offering features like real-time translation, video streaming, and social media connectivity through platforms like WhatsApp and Messenger. Users can also take photos, play music, and ask questions about what they see, all hands-free. This makes Ray-Ban and Meta’s glasses a practical choice, potentially more so than competitors like Apple's Vision Pro, emphasizing their everyday utility and seamless functionality. Source: Meta

Moderna Transforms Operations with OpenAI's ChatGPT Enterprise Integration

Moderna has fully integrated OpenAI's ChatGPT Enterprise, revolutionizing its operations across all departments to enhance efficiency and innovation. This strategic AI deployment supports Moderna's mission to deliver impactful mRNA medicines, with plans to launch up to 15 new products in the next five years. Since the integration, Moderna has created 750 unique GPTs and achieved an 80% adoption rate among employees, significantly advancing everything from optimizing clinical trials with tools like Dose ID to streamlining legal and corporate communications. This demonstrates the transformative power of AI in enhancing biopharmaceutical efficiencies and capabilities. Source: OpenAI

 OpenAI vs. Do-It-Yourself Large Language Models

Integrating LLMs like OpenAI into services offers scalability, but hitting high usage limits can prompt businesses to consider self-hosting for greater control and potential savings. However, self-hosting has significant costs and complexities. For example, deploying the LLaMA 2 model can incur around $67/day for fine-tuning and approximately $2,700/month for hosting on AWS, amounting to nearly $5,400/month when accounting for redundancy. To justify these costs over using OpenAI's services, a company would need to process an enormous volume—about 22.2 million words per day. Therefore, despite the appeal of independence, the financial and operational demands of self-hosting LLMs make it impractical for most businesses unless they operate on a very large scale. Source: VentureBeat

A nurse prepares to give Steve Young, one of the first patients in the trial, his first jab at UCLH in London. Photograph: Jordan Pettitt/PA

Groundbreaking Personal mRNA Vaccines Show Promise in Melanoma Trials

There is new hope in the fight against cancer as personalized mRNA vaccines undergo testing, showing promising results in preventing melanoma recurrence. These vaccines, tailored to individual patients, instruct the body to target and eliminate cancer cells. A phase 2 trial demonstrated a significant reduction in the risk of melanoma returning, and now, a critical phase 3 trial is underway at University College London Hospitals NHS Foundation Trust (UCLH). Led by Dr. Heather Shaw, this trial not only focuses on melanoma but also explores the vaccine's efficacy against other cancers like lung, bladder, and kidney, potentially heralding a new era in cancer treatment. Source: The Guardian.

Apple Introduces OpenELM: Compact AI Models for Smartphones

Apple has launched a new set of AI language models, known as OpenELM, that are compact enough to operate directly on smartphones. Introduced as primarily proof-of-concept research models, these small language models are gaining traction due to their ability to function on local devices without the need for cloud-based, data center-grade computing power. Although currently in the early stages, OpenELM represents a significant step towards future on-device AI applications, potentially transforming how users interact with their devices by enhancing privacy and processing speed. Source: ARSTechnica

Introducing Idefics2: A Leap Forward in Vision-Language AI Models

Idefics2 is a new 8B parameter vision-language model capable of processing both text and images for tasks such as describing visuals, creating stories, and extracting document information. It improves upon its predecessor with enhanced OCR capabilities and advanced image handling, maintaining native resolutions without resizing. Released under an open Apache 2.0 license and integrated with the 🤗 Transformers library, Idefics2 is designed for easy accessibility and fine-tuning across various multimodal applications, marking a significant advancement in the community-driven AI development space. Source: HuggingFace

Kouzou Sakai for Quanta Magazine

AI Begins Exploring the Vast Complexities of String Theory

Machine learning is redefining string theory by mapping complex string configurations to elementary particles using neural networks. This approach is helping physicists explore the vast possibilities of a 10-dimensional universe compacted into minuscule shapes, offering new insights into how these dimensions could influence our physical world. While these models have not yet matched our universe’s exact properties, they represent significant progress in understanding whether string theory can describe all fundamental forces and particles. Source: Quantamagazine

Do you find these insights valuable? Are there other topics you think we should cover? Share your thoughts with us. Email us at [email protected].

This Week's AI Tool Highlights 

  • Bland AI: A platform that enables companies to develop, test, and deploy AI-powered phone agents.

  • Kaiden: Accelerate lesson plan creation by 10x.

  • Recraft - Create and edit images with a consistent brand style

  • Layla: Your AI-powered personal travel planner.

  • Otter: An AI transcription service that converts spoken language into searchable text.

AI's Impact on Education: Transforming Teaching and Learning

As digital innovation reshapes education, artificial intelligence (AI) emerges as a transformative force. AI tools, from adaptive learning platforms to tutoring systems, streamline tasks and personalize learning experiences. Educators transition from instructors to facilitators, focusing on critical thinking and creativity. However, it's crucial to view AI as an enhancement, not a replacement, for the human connection in education. Through responsible integration, educators can harness AI's potential to shape the future of teaching and learning. Explore our AI for Educators program.

Free Course

Curious about AI? Jump into our free 60-minute AI Basics course! It's perfect for beginners and a great way to start chatting about AI like a pro. Join us, learn quickly, and enjoy the conversation—sign up now and explore AI with friends!

About The AI Citizen Hub - by World AI University (WAIU)

The AI Citizen Weekly News stands as the premier source for domain-specific AI tools, articles, trends, and news, meticulously curated for thousands of professionals spanning top companies and government organizations globally, including the Canadian Government, Apple, Microsoft, Nvidia, Facebook, Adidas, and many more. Regardless of your industry – whether it's medicine, law, education, finance, engineering, consultancy, or beyond – The AI Citizen is your essential gateway to staying informed and exploring the latest advancements in AI, emerging technologies, and the cutting-edge frontiers of Web 3.0. Join the ranks of informed professionals from leading sectors around the world who trust The AI Citizen Weekly for their updates on the transformative world of artificial intelligence.

For advertising inquiries, feedback, or suggestions, please reach out to us at [email protected].

Join the conversation

or to participate.