Terrifyingly Real AI Clones are coming

Over the last 2 months we've seen VASA-1, EMO, and now Synthesia Avatars. All of these platforms / papers are now focusing on life-like realism for their avatars.

Today we cover: Synthesia’s lifelike AI avatars, enhancing human-machine interactions. President Biden's directive to the Department of Homeland Security underscores the importance of AI safety and security measures. SenseNova 5.0, a Chinese AI model, excels beyond GPT-4 Turbo in tasks like creative writing and logical reasoning.

Stuff you should know

Clones are becoming real

Synthesia's Expressive AI Avatars revolutionize avatar technology by introducing the first avatars that automatically express emotions based on spoken text semantics. With natural voices, instant previews, and a growing selection of avatars, users can quickly find the perfect fit for their needs. The EXPRESS-1 model, trained on extensive human speech data, delivers stunningly realistic and human-like performances. These avatars seamlessly synchronize with audio, capturing the nuances of human expression. Synthesia prioritizes responsible AI usage, making these expressive avatars a versatile solution for creating high-quality videos without traditional production.

This is not the first one.

Last week, someone posted a face swapped with a girl while eating noodles from a bowl.

There’s obviously VASA-1:

In a way, clones are getting more and more real, every day. Full coverage:

Artificial Intelligence Safety Board

The Biden administration has formed the Artificial Intelligence Safety and Security Board within the Department of Homeland Security, featuring prominent tech CEOs. Responding to the President's executive order on AI, the board aims to ensure safe, secure, and trustworthy AI development and usage. It will provide recommendations to prevent AI-induced disruptions in critical infrastructure that could threaten national security and public safety. The board assists stakeholders in promoting responsible AI use and includes members from technology, defense, and civil rights sectors. This initiative marks the government's first step in regulating AI to maximize benefits while minimizing risks, ensuring AI is deployed safely and securely as it transforms industries and societies.

SenseNova 5.0: The Chinese AI Model Outperforming ChatGPT

Alibaba's DAMO Academy has unveiled SenseNova 5.0, a groundbreaking multimodal AI model that outperforms OpenAI's GPT-4 in various natural language processing tasks. Trained on over 5 billion text-image pairs, SenseNova processes and generates text, images, audio, and video, showcasing unmatched general intelligence and reasoning abilities. Key features include:

• Superior performance in reading comprehension, dialogue, and multi-turn Q&A compared to ChatGPT

• Multimodal processing and generation of text, images, audio, and video

• State-of-the-art results on major AI benchmarks

Potential real-world applications:

1. Next-gen virtual assistants that converse naturally and generate multimedia content

2. Creative content generation for marketing, storytelling, and script writing

3. Multimodal search and analysis of diverse data sources like text, images, and videos

SenseNova's multimodal capabilities herald a new era of AI, revolutionizing industries and transforming how we interact with and make sense of the vast multimodal data in our world.

Handpicked video

I built a LLama-3 based RAG Chatbot

Product Hunt Top AI Product 

• PaddleBoat -  Sharpen your sales pitch by practicing your cold-calls with realistic AI buyers! Choose from a list of AI buyers or create your own with customizable personalities, objections, and instructions tailored to your business context.

• Langtail Public Beta - Langtail is an LLMOps platform that helps teams speed up the development of AI-powered apps and ship to production with fewer surprises.

• Langfuse 2.0 - Langfuse is the open source LLM Engineering Platform. It provides observability, tracing, evaluations, prompt management, playground and metrics to debug and improve LLM apps. Langfuse is open. It works with any model, framework and you can export all data.LangWatch -

• AI or Not - AI or Not is an AI detector trusted by 100k+ users that checks for AI generated content in images, audio, KYC identity documents and more. We help businesses stop fraud, power content moderation and prevent KYC scams.

• ALIagents.ai- ALIagents.ai is a dApp on the AI Protocol that democratizes the creation, customization, and monetization of agentic AI through blockchain technology. Users can create custom generative and agentic AI systems.

• Wonderplan.ai 2.0 - Wonderplan lets you create unique, hassle-free travel plans with AI. Start fresh or with blogs & videos, then fine-tune with personalized options.

• Acryl - mobile app for creating audiobooks from paper books you already have! 📚

• Truva - Truva AI agents not only guide your users to the right tools, features, and info within your app, but they can take action on their behalf delivering the experience of a personal account manager for every customer in real-time.

This week in AI

• Life2vec life expectancy calculator -  The Life2vec AI Death Calculator is an innovative tool developed through international collaborations, utilizing deep learning algorithms to estimate an individual's lifespan based on genetic data, lifestyle choices, and health factors. With an accuracy rate of approximately 78%, it offers personalized insights and recommendations for healthier living.

• Google surpasses $2 trillion market cap - Nvidia has acquired Israeli startup Run:ai for $700 million, which specializes in artificial intelligence (AI) computing resources management. The acquisition is part of Nvidia's strategic moves to strengthen its position in the AI sector and optimize infrastructure management for AI workloads.

• MIT Develops Secure Chip for Efficient Computing- The MIT and MIT-IBM Watson AI Lab have developed a tiny chip that enhances security for health-monitoring apps on smartphones. This chip safeguards user data while enabling efficient computing by resisting common types of attacks and performing computations inside the device's memory. It reduces the need to transfer large amounts of data between the device and a central server, improving efficiency and security for sensitive information like health records and financial data.\

• Windows 11's AI Explorer Enhances Productivity - Windows 11's AI Explorer is a new feature designed to enhance productivity by capturing and recalling user activities, providing contextual suggestions, and running locally for privacy and performance. It will be available on "AI PCs" powered by Qualcomm's Snapdragon X series chips, showcasing Microsoft's focus on integrating advanced AI directly into the Windows experience for improved user workflow.