The AI Rundown by Lightscape Partners – 01/29/24

Good morning and welcome back to another week of AI news.

Last week in artificial intelligence, George Carlin’s estate filed a lawsuit following the release of an AI generated comedy special using Carlin’s likeness, Google released a stunning new AI video generation tool, Biden’s AI project gained more support from both public and private tech giants, and much more.

Let’s get into it.

Top AI Developments of the Week

Lawsuit Against Dudesy for AI-Generated George Carlin Special. Link.

George Carlin's estate is suing Dudesy, a media company, for copyright infringement over an AI-generated comedy special mimicking Carlin's style and voice, titled "George Carlin: I’m Glad I’m Dead."

• The lawsuit, filed in Los Angeles federal court, alleges unauthorized use of Carlin's copyrighted materials and likeness.

• Dudesy's special has been online since January 9th, amassing about 500,000 views.

• Carlin’s estate demands the removal and destruction of all copies of the AI-generated special and seeks unspecified damages.

This legal action underscores the growing tension between AI-generated content and copyright laws, highlighting the need for clear guidelines and protections for artists’ legacies and intellectual property rights. The case could set a precedent for how AI-generated content involving deceased celebrities is handled legally.

Google Lumiere: AI Video Generation Tool. Link.

Google has introduced Lumiere, an advanced AI video generation model using a new diffusion model, Space-Time-U-Net (STUNet), enhancing the realism in AI-generated videos.

• Lumiere's STUNet framework efficiently approximates objects' movements in videos, creating seamless motion.

• It surpasses competitors like Runway in generating realistic videos, notably in the fluid movement of objects like turtles.

• The model can produce 80 frames, exceeding the 25 frames from Stable Video Diffusion.

Lumiere represents a significant advancement in AI video technology, showcasing Google's potential in the competitive AI video generation market. However, it also raises concerns about the technology's misuse and the need for responsible development and deployment.

Biden's AI Science Project Gains Support from Tech Giants

The National Artificial Intelligence Research Resource (NAIRR), part of President Joe Biden's initiative to democratize AI research, has gained pledges from major tech firms like Microsoft and Nvidia. This program, led by the National Science Foundation (NSF) and other agencies, aims to provide extensive AI resources to US-based researchers.

• Alongside NSF, agencies like DARPA, NASA, NIH, and others are involved.

• Major tech companies like Amazon, Google, OpenAI, and others are partnering to provide resources.

• Microsoft offers $20 million in Azure credits, while Nvidia provides access to DGX Cloud and AI Enterprise software.

This collaboration between government and private sector entities underscores the US's commitment to maintaining its lead in AI innovation. It also highlights the cooperative approach between the government and tech companies in the absence of explicit AI regulations. The initiative focuses on areas like healthcare and environmental sustainability, indicating a strategic approach to addressing key societal challenges through AI.

Venture

ElevenLabs Achieves Unicorn Status with $80M Funding, Launches Voice Marketplace. Link.

ElevenLabs, an AI voice startup founded by ex-Google and Palantir employees, has raised $80 million in Series B funding, reaching a valuation of $1.1 billion. This round was co-led by Andreessen Horowitz, Nat Friedman, and Daniel Gross, with participation from Sequoia Capital and SV Angel.

• ElevenLabs specializes in machine learning for voice cloning and synthesis in multiple languages.

• The startup announced a tool for movie dubbing and a marketplace for users to sell their cloned voices.

• ElevenLabs aims to make content universally accessible in any language using AI.

• A Jailed Pakistani Prime Minister used ElevenLabs to release propaganda from prison.

ElevenLabs' rapid growth and substantial funding signify its strong position in the AI voice generation market, which is projected to reach $5 billion by 2032. The company's focus on professional-grade tools and a voice marketplace differentiates it from competitors in the field.

Ola Founder's Krutrim Becomes India's First AI Unicorn. Link.

Krutrim, an AI startup founded by Bhavish Aggarwal of Ola, has achieved a valuation of $1 billion, becoming India's first AI unicorn and the fastest in the country to reach this status. The startup focuses on developing a large language model catering to local Indian languages and English.

• Led by Matrix Partners India, Krutrim secured a $50 million funding round.

• he startup plans to launch a conversational AI assistant capable of understanding and speaking multiple Indian languages.

• Krutrim aims to release a beta version of its chatbot, followed by APIs for developers, and intends to manufacture AI-optimized chips.

• This is Aggarwal's third venture, following successes with Ola and Ola Electric.

Krutrim's rise as an AI unicorn signifies a major milestone in India's growing presence in the global AI landscape. Its focus on local language models addresses a significant gap in the AI market, offering potential for widespread application in India's diverse linguistic environment. This development reflects the increasing global interest in AI as a transformative force across various industries.

Software + Hardware

OpenAI's New Generation of Embedding Models and GPT Updates. Link.

OpenAI has announced a new generation of embedding models and updates to its GPT-4 Turbo and moderation models. This includes the introduction of new API usage management tools and a price reduction for the GPT-3.5 Turbo model.

• The text-embedding-3-small and text-embedding-3-large models offer improved performance and lower pricing compared to previous models.

• Significant score increases in multi-language retrieval and English task benchmarks.

• Enhancements in GPT-4 Turbo and GPT-3.5 Turbo include improved instruction following, JSON mode, and a new 16k context version for longer processing.

• The updated text moderation model now supports more languages and provides prediction explanations.

• New features for managing API keys and monitoring usage.

• A 25% price reduction for the GPT-3.5 Turbo model.

These updates represent OpenAI's ongoing efforts to enhance the utility and accessibility of its models for developers, contributing to more innovative and cost-effective AI solutions in various applications.

Hugging Face and Google partner marking a big win for open-source AI development. Link.

Google has teamed up with Hugging Face, a significant player in open-source AI models, to provide developers with integrated access to Google Cloud services. This strategic collaboration aims to accelerate the development of open generative AI applications.

• Developers using Hugging Face's open-source models can train and serve them using Google Cloud’s infrastructure, including Vertex AI, TPUs, and GPUs.

• Known as the "GitHub for AI," Hugging Face offers a repository with over 500,000 AI models and 250,000 datasets, serving a vast user base including 50,000 organizations.

• Hugging Face users on Google Cloud can train, tune, and deploy models using Vertex AI and Google Kubernetes Engine, with the process streamlined for simplicity.

• Developers will have access to Google Cloud’s advanced hardware, including TPU v5e, A3 VMs with Nvidia H100 Tensor Core GPUs, and C3 VMs with Intel Sapphire Rapid CPUs.

• The collaboration will enable easy deployment of models for production on Google Cloud, with efficient application acceleration and management of usage and billing through Google Cloud accounts.

The new features are expected to be available to Hugging Face Hub users in the first half of 2024, enhancing the capabilities for developers in AI application development.

Nvidia's RTX GPUs Enhance SDR Content to HDR with AI. Link.

Nvidia has introduced RTX Video HDR, a feature for RTX GPU owners that uses AI to upgrade Standard Dynamic Range (SDR) videos to High Dynamic Range (HDR) quality. This feature is available with the latest 551.23 Game Ready driver release, coinciding with the launch of the RTX 4070 Ti Super.

• RTX Video HDR is compatible with HDR10 monitors and works in Microsoft Edge and Google Chrome.

• The feature adds subtle color details to non-HDR videos, with the effect varying based on monitor quality.

• Users can enable RTX Video HDR in the Nvidia Control Panel under Video settings.

• An RTX card is necessary due to the reliance on tensor cores for AI processing.

This innovation enhances the viewing experience for RTX GPU users, allowing them to enjoy improved video quality without the need for native HDR content. It further establishes Nvidia's commitment to AI-powered enhancements in multimedia, gaming, and other applications, leveraging their advanced GPU technology.

Ethics in AI

FTC Investigates Big Tech Investments in AI Companies. Link.

The Federal Trade Commission (FTC) is investigating the investment relationships between major tech companies like Microsoft, Amazon, and Google, and AI startups such as OpenAI and Anthropic.

• The FTC has sent letters to these companies asking for details on their investments and the impact on the AI market.

• Focus areas include influence on product releases, oversight rights, competition, and market share.

• Companies like Microsoft and Amazon have made significant investments in AI startups, influencing AI development and deployment.

This investigation highlights the increasing scrutiny of Big Tech's influence on emerging technologies and market competition. The outcome could shape future regulatory policies and business practices in the rapidly evolving AI industry.

Rising Energy Demand from AI and Crypto Mining in Data Centers. Link.

The International Energy Agency (IEA) reports that data centers' electricity consumption could double by 2026 due to the increasing demands of cryptocurrencies and AI technologies. These data centers are crucial for cloud storage, Bitcoin mining, and training AI tools like ChatGPT.

• Data centers, cryptocurrencies, and AI accounted for about 2% of global electricity demand in 2022, using 460TWh.

• By 2026, this consumption could rise to 1,050TWh, with crypto mining alone accounting for a significant portion.

• The US leads in data center numbers and Bitcoin mining, while Ireland's data centers could consume nearly a third of the country’s electricity by 2026.

The exponential growth in data center electricity usage poses challenges for global power grids and raises concerns about environmental impacts. Although renewable energy growth is forecasted, it may not fully address the surge in electricity demand. Improving energy efficiency in data centers becomes crucial to balance this technological advancement with environmental sustainability.

Other

BMW Integrates Humanoid Robots in Factory. Link.

BMW has partnered with FigureAI to introduce humanoid robots, dubbed "Figure 01," into its manufacturing process. These advanced robots are designed for versatility in performing various physical tasks.

• Bipedal with a height of 5'6" and weight of 130 pounds.

• Equipped with five-fingered hands, capable of constructing objects and performing complex tasks.

• Robots to be deployed at BMW's Spartanburg, South Carolina facility.

• Handling difficult, unsafe, or tedious tasks in the manufacturing environment.

• Despite concerns, FigureAI's CEO emphasizes that their goal is not to replace human jobs but to create safer and more efficient work environments.

This development marks a significant step in the integration of advanced robotics into traditional manufacturing processes, potentially reshaping the dynamics of industrial automation.

Thank you for reading the AI Rundown by Lightscape Partners. Stay tuned for the latest updates in the world of Artificial Intelligence!