• "Towards AGI"
  • Posts
  • Apple and Nvidia Eye Investments in OpenAI’s Upcoming Billion-Dollar Funding Round

Apple and Nvidia Eye Investments in OpenAI’s Upcoming Billion-Dollar Funding Round

Author

Shen Pandi
August 30, 2024

In partnership with

Welcome to Towards AGI, your premier newsletter dedicated to the world of Artificial Intelligence. Our mission is to guide you through the evolving realm of AI with a specific focus on Generative AI. Each issue is designed to enrich your understanding and spark your curiosity about the advancements and challenges shaping the future of AI.

Whether you're deeply embedded in the AI industry or just beginning to explore its vast potential, "Towards AGI" is crafted to provide you with comprehensive insights and discussions on the most pertinent topics. From groundbreaking research to ethical considerations, our newsletter is here to keep you at the forefront of AI innovation. Join our community of AI professionals, hobbyists, and academics as we pursue the ambitious path toward Artificial General Intelligence. Let’s embark on this journey together, exploring the rich landscape of AI through expert analysis, exclusive content, and engaging discussions.

Apple and Nvidia Eye Investments in OpenAI’s Upcoming Billion-Dollar Funding Round

OpenAI, the artificial intelligence company, is preparing for a major funding round that could value the company at over $100 billion, according to a report by The Wall Street Journal. The report indicates that two of the world's largest tech companies, Apple and Nvidia, may participate in this investment round. Microsoft is already one of OpenAI's largest investors.

Apple's rumored interest isn't entirely unexpected. In June 2024, during its annual developer conference, Apple introduced several AI features under the banner of Apple Intelligence, some of which will be powered by OpenAI's ChatGPT. Earlier this year, Phil Schiller, a senior Apple executive, was slated to join OpenAI's board as a non-voting observer, but he ultimately did not take the position due to antitrust concerns.

Nvidia's potential investment is also unsurprising, as the company is a major supplier of the chips that power OpenAI's foundational models.

The report further notes that this funding round is being led by Thrive Capital and could value OpenAI at more than $100 billion.

From Ban to Build: Why Companies Like JPMorgan and Walmart Are Embracing Internal AI Assistants

Eighteen months after limiting employee use of generative AI tools like ChatGPT, JPMorgan Chase CEO Jamie Dimon introduced an AI assistant developed in-house. Ironically, this tool, named LLM Suite, is based on the technology behind ChatGPT, created by OpenAI. The bank has already made this service available to 60,000 employees for tasks such as writing reports and drafting emails.

This shift from restricting generative AI to developing an internal, secure solution is a trend seen across companies of all sizes. According to Cisco’s latest Data Privacy Benchmark report, over a quarter (27%) of organizations have temporarily banned public generative AI applications, while most have imposed limitations on what solutions employees can use and how they can use them.

Despite these restrictions, many companies report that employees continue to use unauthorized applications, as revealed by a survey from cybersecurity firm Extrahop. This has made it increasingly important for organizations to provide alternative solutions with robust safeguards.

The primary concerns surrounding employee use of generative AI include the risk of leaking sensitive information, generating inaccurate outputs (hallucinations), and ensuring industry compliance, according to a report by Veritas, an enterprise data management company.

AI platforms do not generate outputs in isolation; they often store user inputs to improve their capabilities, which can expose sensitive company or customer information. This risk has led many organizations to ban or restrict generative AI use until they can manage the technology internally.

JPMorgan isn't alone in this approach. Walmart also transitioned from restricting generative AI to rolling out its own tool, My Assistant, initially to 50,000 employees, with further expansion to 25,000 more in 11 countries. David Glick, Walmart's Senior Vice President of Enterprise Business Services, emphasized that their AI principles are shaped by employee needs, such as summarizing large volumes of information and improving efficiency in navigating internal systems.

Glick also highlighted that generative AI might best serve the company through numerous small, incremental improvements that enhance daily operations.

Ensono, an IT managed service provider, also reconsidered its stance on generative AI. CTO Tim Beerman initially restricted employee access to tools like ChatGPT to protect sensitive data but later introduced an internal AI assistant. This assistant, based on GPT-4, offers flexibility to incorporate different language models depending on the data type. Ensono is gradually deploying specialized language models for specific use cases across departments, with flexibility being crucial for future changes.

Similarly, Jason Hishmeh, CTO of startup software development company Varyence, who had previously prohibited using generative AI for confidential data, now oversees an internal system with built-in safeguards to ensure data security while allowing employees to experiment with AI.

As more companies move from banning generative AI to implementing their internal solutions, the role of this technology in the workplace is likely to evolve. However, given the rapid pace of AI innovation, companies must stay adaptable in their AI policies, with safety and security remaining a top priority.

Anthropic's Claude Powers Amazon's Upcoming 'Remarkable Alexa' Update

While there's been considerable discussion around Amazon's generative AI-powered upgrade for its voice assistant, Alexa, and the anticipated subscription fees, the e-commerce giant has remained quiet about a specific launch date. This week, reports surfaced indicating that the subscription-based update is expected to roll out in October, though it won't be powered by Amazon's proprietary AI models. Instead, the new and improved Alexa, featuring generative AI, is expected to launch in just over a month, reportedly using Anthropic's Claude AI models. Sources familiar with the situation told Reuters that Amazon's in-house software fell short of creating the "remarkable" generative AI virtual assistant the company had envisioned, with slow response times and unmet performance expectations.

The October launch for the updated Alexa marks a delay from earlier reports suggesting a September release. This new version will be a paid alternative to the "classic Alexa," as Amazon refers to the currently available free version in internal documents obtained by the Washington Post. Several reports suggest that a subscription to the new virtual assistant, called "Remarkable Alexa," could range between $5 and $10 per month. Alexa, Amazon's voice assistant available on Echo devices, traditionally uses AI to understand commands, respond accordingly, set timers, answer questions, and control smart home devices. By integrating generative AI features, Alexa would gain the ability to create content rather than merely repeat pre-programmed responses. For instance, instead of searching for a recipe, Alexa could generate one tailored to dietary restrictions or available ingredients.

Additionally, Alexa could create text such as stories, emails, poems, or songs. Children could ask Alexa to create a story featuring their family dog, offering a fresh bedtime story centered around Fido. This feature could be particularly appealing if, like mine, your kids have already exhausted Alexa's existing story library.

Along with generative AI capabilities, the new Alexa is expected to engage in more natural conversations, answer follow-up questions, and use a more human-like voice. According to the report obtained by the Washington Post, Amazon stated that "AI features that help customers curate, summarize, and explore current events" were among the top customer requests.

Amazon first announced that Alexa would gain generative AI capabilities in September 2023 at the company's Devices and Services event, where it showcased the new voice assistant's features. The announcement followed a year of rapid growth in AI technologies, highlighted by the release of ChatGPT in late 2022. Since OpenAI's chatbot became widely available, both Microsoft and Google have launched their own AI chatbots, Copilot and Gemini, respectively. This has increased pressure to improve existing virtual assistants like Alexa, Siri, and Google Assistant, which have struggled to compete with advanced AI tools like ChatGPT.

Anthropic, in which Amazon holds a minority stake, also released its own generative AI chatbot named Claude. Claude has undergone several upgrades to compete with ChatGPT, though it hasn't achieved the same level of popularity as OpenAI's bot. Anthropic's AI models are also featured in Amazon Bedrock, a service that offers developers and businesses foundational models to build upon.

OpenAI to Introduce ‘Strawberry’ Model Focused on Superior Logical Reasoning

Sam Altman’s OpenAI is reportedly close to unveiling a new artificial intelligence (AI) model that promises significant advancements in the mathematical and reasoning capabilities of its popular chatbot, ChatGPT.

According to a report from The Information, this project, codenamed "Strawberry," has been in development for several months, with OpenAI keeping much of its progress under wraps. The existence of this large language model (LLM) was first revealed in July, and Strawberry is being positioned as the potential successor to the well-regarded GPT-4.

The report also suggests that the Strawberry AI model is expected to launch this fall. OpenAI is reportedly planning to introduce this new AI model during the upcoming fall season, though details about the rollout are still unclear. One option being considered is releasing Strawberry as a standalone chatbot, while another possibility is integrating its advanced capabilities into the existing ChatGPT platform. The excitement around the Strawberry model comes from its rumored ability to handle complex mathematical and reasoning tasks. Unlike current AI models that struggle with unfamiliar problems, Strawberry is said to be capable of solving equations and logical puzzles it has never encountered before.

This development is particularly significant because most AI models are built as deep learning systems that recognize and process patterns. When these patterns become symbolic or ambiguous, traditional AI models often stumble. As a result, current chatbots face notable limitations when dealing with intricate mathematical problems, logical reasoning, or context-heavy queries.

However, if the recent reports are accurate, Strawberry may have overcome these challenges, marking a significant leap forward in AI technology. Although specific details about the model's architecture or parameters remain limited, sources indicate that Strawberry was previously known as Q* (pronounced Q-star).

Meta’s Llama Models Reach 350 Million Downloads, Leading the Open-Source AI Charge

Meta’s Llama series of AI models has quickly become the fastest-growing open-source model family, with 350 million downloads globally on Hugging Face. Of these, 20 million downloads occurred just last month, according to the company.

In addition to Hugging Face, Llama is also being accessed through Meta’s cloud partners, including Amazon Web Services (AWS), Microsoft’s Azure, Databricks, Dell, Google Cloud, Groq, NVIDIA, IBM watsonx, Scale AI, and Snowflake, among others.

"Usage of hosted Llama by token volume across our major cloud service provider partners more than doubled from May through July 2024, following the release of Llama 3.1," Meta stated in a blog post on Thursday.

For some of the largest cloud service providers, Llama's monthly usage (measured by token volume) grew tenfold from January to July 2024.

Meta added, "We've heard from several companies interested in becoming future LEAP and integration partners for Llama, including Wipro, Cerebras, and Lambda."

Industry experts noted that the increasing adoption of Llama indicates that Meta is narrowing the performance gap between open-source models and top proprietary models like OpenAI's GPT and Anthropic's Claude.

"Since its launch, thousands of Databricks customers have adopted Llama 3.1, making it our fastest-adopted and best-selling open-source model ever," said Ali Ghodsi, CEO & Co-Founder of Databricks, as quoted in the same blog post. "This generation of Llama models finally bridges the gap between open-source software (OSS) and commercial models in terms of quality. Llama 3.1 is a breakthrough for customers who want to build high-quality AI applications while maintaining full control, customizability, and portability over their base large language model (LLM)."

"Open-source wins. Meta is laying the foundation for an open ecosystem that rivals top closed models, and at Groq, we’re putting them directly into the hands of developers. We can't add capacity fast enough for Llama. If we increased the deployed capacity tenfold, it would be consumed in under 36 hours," said Jonathan Ross, Founder & CEO of Groq, Elon Musk’s AI project.

Meta also highlighted that with more than 60,000 derivative models on Hugging Face, there’s a thriving community of developers fine-tuning Llama for their specific use cases. The company shared several success stories of large enterprises, including AT&T, DoorDash, Goldman Sachs, Shopify, and Zoom, that are using Llama.

According to a survey by Artificial Analysis, an independent AI benchmarking site, Llama was ranked as the number two most considered model and the industry leader in open source.

How CogVideoX Could Forever Transform Video Creation with Open-Source AI

Researchers from Tsinghua University and Zhipu AI have introduced CogVideoX, an open-source text-to-video model poised to challenge the AI landscape currently dominated by startups like Runway, Luma AI, and Pika Labs. Detailed in a recent arXiv paper, this groundbreaking model empowers developers globally with advanced video generation capabilities.

CogVideoX can generate high-quality, coherent videos up to six seconds long from text prompts, outperforming well-known competitors like VideoCrafter-2.0 and OpenSora across various metrics, according to the researchers’ benchmarks.

The flagship model, CogVideoX-5B, features 5 billion parameters and produces videos with a resolution of 720×480 at 8 frames per second. While these specifications may not match the cutting-edge proprietary systems, the true innovation of CogVideoX lies in its open-source nature.

By making their code and model weights publicly available, the Tsinghua team has democratized a technology that was once reserved for well-funded tech companies. This move has the potential to accelerate advancements in AI-generated video by tapping into the collective expertise of the global developer community.

The researchers achieved CogVideoX’s impressive results through several technical innovations, including the implementation of a 3D Variational Autoencoder (VAE) for efficient video compression and the development of an "expert transformer" to enhance text-video alignment. According to the paper, “To improve the alignment between videos and texts, we propose an expert Transformer with expert adaptive LayerNorm to facilitate the fusion between the two modalities.” This advancement allows for more nuanced interpretation of text prompts and more accurate video generation.

The release of CogVideoX marks a significant shift in the AI landscape, providing smaller companies and individual developers with access to capabilities that were previously beyond their reach due to resource limitations. This democratization of technology could ignite a wave of innovation across industries such as advertising, entertainment, education, and scientific visualization.

OpenAI and Anthropic Ink Deals with U.S. Government for AI Research and Testing

AI startups OpenAI and Anthropic have entered into agreements with the U.S. government for the research, testing, and evaluation of their artificial intelligence models, according to an announcement by the U.S. Artificial Intelligence Safety Institute on Thursday. These agreements, the first of their kind, come at a time when both companies are under regulatory scrutiny for the safe and ethical use of AI technologies.

In California, legislators are expected to vote soon on a bill that would broadly regulate the development and deployment of AI in the state.

Jack Clark, Co-Founder and Head of Policy at Anthropic, which is backed by Amazon and Alphabet, emphasized the importance of safe, trustworthy AI. "Our collaboration with the U.S. AI Safety Institute draws on their extensive expertise to rigorously test our models before they are widely deployed," he said.

As part of these agreements, the U.S. AI Safety Institute will have access to major new models from both OpenAI and Anthropic before and after they are publicly released. The collaboration will also focus on joint research to assess the capabilities of these AI models and the risks associated with them.

Jason Kwon, Chief Strategy Officer at OpenAI, the maker of ChatGPT, highlighted the institute's role in establishing U.S. leadership in the responsible development of AI. "We hope that our work together provides a framework that can be adopted globally," he said.

Elizabeth Kelly, Director of the U.S. AI Safety Institute, called these agreements an important milestone in the effort to responsibly guide the future of AI.

The institute, which is part of the U.S. Commerce Department's National Institute of Standards and Technology (NIST), will also collaborate with the U.K. AI Safety Institute and offer feedback to the companies on potential safety improvements.

The U.S. AI Safety Institute was established last year under an executive order from President Joe Biden's administration to evaluate the known and emerging risks of AI models.

OpenAI Reports ChatGPT User Base Doubles to 200 Million Weekly Active Users

ChatGPT, launched by OpenAI in 2022, generates human-like responses based on user prompts and has quickly become one of the most popular tools for generating such responses globally. Initially, OpenAI's CEO mentioned that the chatbot had 100 million weekly active users in November.

However, the AI startup has now confirmed that ChatGPT has surpassed 200 million weekly active users worldwide. Additionally, the company revealed that 92% of Fortune 500 companies are utilizing its AI-powered products, including its automated Application Programming Interface (API). The use of this API, which enables software programs to communicate with each other, has doubled since the introduction of GPT-4o mini in July.

GPT-4o mini is a smaller, more cost-effective AI model designed to make OpenAI’s conversational technology more affordable and less energy-intensive. This has allowed a broader audience to access OpenAI’s tools, boosting their popularity over the past two months.

Taya Christianson, an OpenAI spokesperson, confirmed these developments to The Verge as the AI startup nears its second anniversary this fall. The rising popularity of ChatGPT has also increased interest in other AI tools like Copilot and Google Gemini, contributing to the meteoric rise in OpenAI’s valuation.

Unlock the future of problem solving with Generative AI!

If you're a professional looking to elevate your strategic insights, enhance decision-making, and redefine problem-solving with cutting-edge technologies, the Consulting in the age of Gen AI course is your gateway. Perfect for those ready to integrate Generative AI into your work and stay ahead of the curve.

In a world where AI is rapidly transforming industries, businesses need professionals and consultants who can navigate this evolving landscape. This learning experience arms you with the essential skills to leverage Generative AI for improving problem-solving, decision-making, or advising clients.

Join us and gain firsthand experience in how state-of-the-art technology can elevate your problem solving skills using GenAI to new heights. This isn’t just learning; it’s your competitive edge in an AI-driven world.

🦾 Master AI & ChatGPT for FREE in just 3 hours 🤯

1 Million+ people have attended, and are RAVING about this AI Workshop.
Don’t believe us? Attend it for free and see it for yourself.

Highly Recommended: 🚀

Join this 3-hour Power-Packed Masterclass worth $399 for absolutely free and learn 20+ AI tools to become 10x better & faster at what you do

🗓️ Tomorrow | ⏱️ 10 AM EST

In this Masterclass, you’ll learn how to:

🚀 Do quick excel analysis & make AI-powered PPTs 
🚀 Build your own personal AI assistant to save 10+ hours
🚀 Become an expert at prompting & learn 20+ AI tools
🚀 Research faster & make your life a lot simpler & more…