Tencent enters AI video space, Spotify integrates Google's NotebookLM

⚡️ Quick Hits

🤖 Artificial Intelligence

Amazon introduces Nova, a new generation of foundation AI models - The suite includes understanding models for text, image, and video processing, as well as creative content generation models for image and video creation, all available through Amazon Bedrock [AWS News Blog]

ElevenLabs launches conversational AI platform - The new platform allows users to create and deploy AI agents with customizable voices, personalities, and knowledge bases [ElevenLabs]

Exa AI Labs announces the release of Exa-1 - The new “slow web search” is another name added to a growing list of AI search competitors [Twitter]

Apple faces challenges in adapting Baidu's AI models for Chinese users - The tech giant is struggling to integrate Baidu's AI technology into its products for the Chinese market, potentially delaying the release of AI features for iPhone users in China [The Information]

🎨 Creative & Cultural

HailuoAI announces new AI model with impressive performance - Optimized for stability and subtle expression, Hailuo I2V-01-Live supports a wide range of artistic styles[Twitter]

Cameo shifts focus to Creators - The celebrity video messaging platform is pivoting to social media creators with new platform called CameoX [Yahoo Finance]

TikTok Shop generates $100 million in Black Friday sales - The platform hosted over 30,000 livestream shopping sessions, with fashion, beauty, and home products being the most popular categories[1][2] [Business Insider]

₿ Crypto & Blockchain

Trump nominates Paul Atkins as SEC chair - The former president has put forward the name of a former SEC commissioner for the top regulatory position [The Hill]. Crypto liquidations hit $588 million amid market volatility - A significant amount of leveraged positions were forcibly closed as cryptocurrency prices experienced sharp fluctuations [Yahoo Finance]

⚖ Legal & Regulation

EU court adviser backs $2.8bn Google antitrust fine - The advocate general of the European Court of Justice has recommended upholding the substantial penalty imposed on Google for alleged anti-competitive practices in online shopping [Financial Times]

EU Commission requests additional information from TikTok under Digital Services Act - The European Commission is seeking details on TikTok's risk management practices related to information manipulation, particularly in the context of ongoing Romanian elections

🧪 Research

MIT researchers develop photonic chip for ultrafast AI computations - The new chip performs all key computations of a deep neural network optically, potentially enabling faster and more energy-efficient deep learning for demanding applications

🎱 Random

Meta says less than 1% of misinformation during election was AI ****- Meta seems keen on endearing itself to the incoming administration [Meta Newsroom]

🔌 Plug In To These Details

Tencent's HunyuanVideo: Another Groundbreaking AI Text-to-Video Generator

Tencent, the Chinese video gaming and social media giant, has recently launched HunyuanVideo, a powerful new text-to-video generation tool based on its Hunyuan artificial intelligence foundation model. This launch marks Tencent's entry into the competitive AI video generation arena, challenging both domestic rivals and international players like OpenAI's Sora.

• The model generates high-quality videos up to 720p resolution in both Chinese and English.

• HunyuanVideo is free for enterprise and individual users, accessible via various platforms.

• It requires significant GPU VRAM (45-60GB) and can take up to 15 minutes to generate a 5-second video.

• Tencent's tool competes with offerings from Kuaishou, Alibaba, and international players like OpenAI's Sora.

🚀 By open-sourcing HunyuanVideo, Tencent aims to accelerate innovation in AI video generation technology and present as a legitimate player in a now crowded field.

Spotify’s 2024 “Wrapped” Feature Uses Google's NotebookLM

Spotify has launched its highly anticipated 2024 Wrapped feature with an exciting new addition: a personalized AI podcast powered by Google's NotebookLM. This innovative collaboration brings a fresh dimension to the annual listening recap experience.

• Spotify's 2024 Wrapped features a personalized AI podcast using Google's NotebookLM technology.

• The "Your Spotify Wrapped AI Podcast" uses two AI hosts to guide users through their listening recap.

• Available in English for eligible users in select countries, including the U.S., U.K., and Australia.

• New features include "Your Music Evolution" data story and AI-generated playlists for Premium users.

• Audiobooks are included in the Wrapped recap for the first time.

🎵 This collaboration between Spotify and Google marks a significant step in personalizing user experiences, blending AI technology with music streaming to create a unique, engaging way for listeners to reflect on their year in audio.

Challenges Facing Perplexity's AI Shopping Agent

Perplexity's AI shopping agent, called "Buy with Pro," has been experiencing several issues since its debut in mid-November 2024. These problems highlight the challenges of implementing AI-powered shopping assistants:

• Slow processing: Purchases can take hours to complete, with some confirmations arriving up to 8 hours later.

• Inventory inaccuracies: The AI often fails to reflect real-time stock, leading to attempted purchases of unavailable items.

• Unconventional payment process: Users pay Perplexity directly, not the retailer, causing potential discrepancies between displayed and actual availability. • Privacy concerns: Human checkers are involved in the process, potentially accessing users' purchase information and addresses.

• Retailer relationship issues: The AI appears to scrape retailer websites without authorization, which could lead to complications with purchases or returns.

🤖 Despite these challenges, Perplexity's AI shopping agent represents an ambitious step towards AI-assisted e-commerce. However, significant improvements in speed, accuracy, and retailer relationships are needed before it can become a reliable tool for consumers.

📸 Creator Corner

Runway’s New Video Keyframing Prototype

Today, Runway AI announced an exciting new video keyframing prototype that promises to revolutionize the creative process for artists and content creators. This innovative tool allows users to map out their ideas and transform them into videos in a unique and intuitive way.

Key Features

Nonlinear Discovery

The prototype enables users to explore their creative ideas in a nonlinear fashion, allowing for serendipitous discoveries and unexpected connections.

Image-to-Video Conversion

Users can take individual images and connect them like points on a map, creating smooth transitions between them to form a cohesive video.

Latent Space Exploration

The tool utilizes a concept called "latent space," which is a vast, invisible world of creative possibilities. Users can navigate this space to discover and connect new ideas.

Graph-Based Interface

The prototype employs a graph-like interface, where each point represents an image, and the connecting lines determine how these images transform into a video.

Flexible Editing

Users can modify aspects of an image while maintaining its overall narrative, add surprises without drastically altering the original image, and change image styles using text prompts.

Benefits for Creators

1. Enhanced Exploration: The tool allows users to visualize their video in the latent space before generation, helping to refine the storytelling process.

2. Non-Linear Thinking: It caters to the non-linear thought processes of many creatives, allowing them to branch out and explore multiple directions.

3. Customizable Workspace: Users can organize their ideas freely, grouping similar concepts or spreading them out for experimentation.

4. Style Experimentation: The prototype enables users to preview their videos in various styles and paths, facilitating extensive exploration and saving preferred options.

Runway AI plans to gradually roll out access to this new tool over the coming months, giving users the opportunity to integrate it into their creative workflows and expand their artistic visions.

🤔 Final Thoughts

Well, as the AI video field grows ever more crowded, Runway again articulates itself as a leader in the space. It makes a ton of sense to develop novel forms of video assembly as well as generation. As getting great footage looks to be getting much easier, the pain points of creating quality content are shifting and it’s a great time for experimentation.

~ JL