Moderna Creates Over 750 GPT's for Medical Advancements

Microsoft Expands Anti-Deepfake Tools to Global Political and Media Sectors

Microsoft is broadening its initiative against digital disinformation by extending its "content credential" technology to EU political entities and global news organizations. This enhancement helps verify the authenticity of media content, signaling whether it's AI-generated or has been altered post-creation. By providing tools previously available only to U.S. political campaigns, Microsoft aims to bolster trust in media shared by political groups and journalists, ensuring the public can recognize and rely on genuine content amidst increasing concerns over deepfakes and manipulated media.

Why it matters: This expansion is critical for maintaining the integrity of information during pivotal global elections, offering a clearer, verified information landscape for voters worldwide.

Source

Tech Giants Unite for Child Safety in AI Development

In a significant move, leading technology firms including Amazon, Google, Meta, Microsoft, and OpenAI have pledged adherence to 'Safety by Design' principles, engineered by child safety group Thorn and ethics-focused nonprofit All Tech is Human. These principles are designed to embed child safety measures into the lifecycle of generative AI technologies—from development to deployment.

Why it matters: By committing to ensure that training data is free of illicit content, eliminating harmful material, and rigorously evaluating AI models for child safety, these companies aim to prevent AI from being used to create or spread child exploitation material.

Source

Adobe Launches Groundbreaking ‘Firefly Image 3 Tool’

Adobe has unveiled Firefly Image 3, a transformative upgrade to its image generation software, now integrated into Photoshop (beta) and available on the Adobe Firefly web app. This latest version addresses previous shortcomings by enhancing its ability to interpret complex text prompts, significantly improving the realism of generated images. The tool also boasts advancements in lighting effects, text generation, and the accurate depiction of typography, iconography, and detailed features, marking it as Adobe’s most sophisticated image generation utility to date.

Why it matters: Adobe's release of Firefly Image 3 sets a new standard in image generation technology, offering creators advanced tools for producing highly realistic and detailed digital art.

Source

Microsoft Unveils ‘Phi-3 Mini’: A Compact AI Powerhouse

Microsoft has launched the Phi-3 Mini, the smallest iteration in its innovative Phi-3 series of AI models, designed to deliver robust AI capabilities on devices as compact as smartphones. The Phi-3 Mini, which boasts 3.8 billion parameters, not only competes with but also exceeds larger models like Meta's Llama 3 in key performance benchmarks. Tailored for optimal performance with less computational demand, this model offers a sustainable and cost-effective alternative to the power-intensive larger models currently dominating the market. Available across platforms like Azure, Hugging Face, and Ollama, Phi-3 Mini is set to revolutionize how AI is integrated into daily technology use, making advanced AI more accessible to a broader audience.

Why it matters: Microsoft's Phi-3 Mini brings high-level AI performance to everyday devices, drastically reducing the cost and environmental impact of deploying advanced AI technologies.

Source

Generative AI Powers New CRISPR Gene Editing Tool, ‘OpenCRISPR-1’

ProfluentBio, a biotech startup, has unveiled OpenCRISPR-1, a revolutionary gene editing tool developed using generative AI. CEO Ali Madani announced the successful DNA editing in human cells and revealed that OpenCRISPR-1 will be open-sourced, promoting widespread access to this advanced technology. This new tool enhances the precision and efficiency of traditional CRISPR methods, creating proteins that could lead to effective, once-and-for-all treatments for genetic disorders.

Why it matters: By making these AI-designed molecules freely available, ProfluentBio aims to democratize the future of gene therapy, potentially transforming the landscape of medical treatment by allowing for more personalized and effective interventions.

Source

Moderna Integrates ChatGPT Enterprise to Revolutionize mRNA Medicine Development

Moderna has teamed up with OpenAI, integrating ChatGPT Enterprise across its operations to boost the productivity of its workforce and speed up the development of new mRNA-based treatments. With OpenAI's help, Moderna has created over 750 specialized GPTs, including tools like Dose ID for reviewing clinical data and Contract Companion GPT for parsing legal documents. This partnership has enabled 100% of Moderna’s legal team to utilize AI tools that do not rely on user data for model training, showcasing a commitment to both innovation and privacy.

Why it matters: Moderna's adoption of AI technologies could significantly enhance the speed and efficiency of developing new medical treatments, potentially setting a new standard for the pharmaceutical industry in leveraging artificial intelligence.

Source

Snowflake Launches ‘Arctic’, a High-Efficiency Generative AI Model for Enterprise

Snowflake has debuted Arctic, a cutting-edge large language model tailored for complex enterprise tasks, including SQL and code generation, as well as instruction following. Utilizing a 'mixture-of-experts' architecture, Arctic divides its 480 billion parameters into 128 expert subgroups, significantly enhancing efficiency by actively using only 17 billion parameters at a time. This allows Arctic to perform with less computing power while still matching or exceeding competitors on benchmarks such as SQL generation and coding tasks.

Why it matters: Arctic represents a significant advancement in AI for business applications, offering enterprises a powerful yet cost-effective tool for automating complex tasks and optimizing operations.

Source

Apple Launches ‘OpenELM’: Small, On-Device AI Models for Enhanced Privacy

Apple has introduced OpenELM, an open-source project offering a range of small AI language models, from 270 million to 3 billion parameters, designed to operate directly on devices such as smartphones and laptops. This move marks a significant shift from Apple's typically closed system approach, prioritizing user privacy by eliminating the need for cloud processing. The models, available on Hugging Face, are optimized for efficient performance and are pre-trained on a diverse dataset to enhance text generation capabilities across various applications.

Why it matters: Apple's OpenELM project not only extends AI capabilities to on-device applications, promoting user privacy and data security, but also democratizes AI research, potentially spurring further innovation in the field of lightweight, efficient AI models.

Source

Hugging Face Unveils ‘Jack of All Trades’: A Versatile, Single-Model AI for Multiple Tasks

Hugging Face has launched JAT (Jack of All Trades), an innovative AI project designed as a multi-functional neural network capable of performing a variety of complex tasks, from gaming and robotics to language understanding. Unlike traditional AI systems that require different models for different applications, JAT combines these capabilities into one robust framework. Developed using the extensive JAT dataset, this model can seamlessly switch between tasks, utilizing its broad training to optimize its approach. Although JAT's advanced requirements limit direct access to specialized users, its open-source nature allows developers and researchers to contribute to and expand its applications.

Why it matters: JAT represents a significant advancement in AI technology, demonstrating potential for a unified AI system that can adapt and respond across a range of complex scenarios, simplifying technology integration and enhancing efficiency across numerous industries.

Source

Adobe Develops ‘VideoGigaGAN’: A New AI Upscaling Model to Enhance Video Quality

Adobe has unveiled VideoGigaGAN, a cutting-edge AI model designed to significantly upscale blurry video clips to HD quality, achieving up to 8 times the original resolution. This model leverages the power of Generative Adversarial Networks (GANs) to enhance image detail while minimizing common issues like flickering. Early demonstrations showcase remarkably natural enhancements, adding realistic textures and details to videos. Currently a research project, it's uncertain whether VideoGigaGAN will be integrated into consumer-facing Adobe products like Premiere Pro

Why it matters: VideoGigaGAN's development highlights a significant leap in video processing technology, offering content creators revolutionary tools to restore and enhance video quality, which could transform the production and consumption of digital media.

Source

Synthesia Enhances AI Video Avatars with Emotionally Expressive Features

Synthesia, an AI startup specializing in video avatar technology, has launched an updated version of its platform featuring "Expressive Avatars." These avatars, built on advanced pre-trained models, exhibit more natural speech and facial expressions, making them less robotic and more lifelike than previous iterations. While the technology is still evolving, Synthesia's CEO anticipates significant improvements, especially in the realism of avatar hands, by the end of 2024. Focused primarily on business applications such as training and marketing, Synthesia aims to offer "safe" AI solutions amidst concerns over potential misuse for disinformation.

Why it matters: Synthesia's advancements in expressive AI video avatars could revolutionize business communications and training, offering more engaging and realistic interactions through digital mediums.

Source

If you are looking for help with leveraging AI in your business, check out my new website for more information!