Happy Friday Friends,

What a momentous week it has been—arguably the most eventful in AI history.

Highlights include the inaugural global AI summit in London, OpenAI DevDay, and Elon Musk's introduction of his new AI chatbot model, among many other developments.

I want to extend my personal gratitude to all of you for supporting PR's growth into an indispensable weekly digest for AI news.

Our mission remains to keep you informed about significant AI stories, ensuring you're well-prepared for the ways AI is reshaping our lives and work.

Expect our AI roundup every Friday, and rest assured, we'll keep you updated on any major developments that occur during the week. Enjoy the read!

Greg Thomas - PR

AI Roundup

The First Worldwide AI Summit

Highlights:

• UK Prime Minister Rishi Sunak and Elon Musk discussed the advancement of AI, tech, and future implications, with Musk expressing caution regarding AI safety and advocating for government intervention when public safety is at risk.
• Elon Musk has been a vocal "Cassandra," warning about the rapid progress in AI and the need for safety measures, while recognizing recent breakthroughs that bring profound artificial intelligence closer to reality.
• Musk stressed the importance of mitigating the risks of AI, especially as we approach the development of digital superintelligence which could exceed human capabilities, and the need for regulatory oversight.
• The conversation touched on the transformative potential of AI, including the possibility of universal basic income due to automation, the preservation of human agency, and the role of AI in future employment.
• Musk and Sunak discussed the impact of AI on governance, with Musk introducing the concept of Community Notes for consensus-driven truth on platforms, and Sunak highlighting the UK's position in AI development and education.

OpenAI Dev Day: Create Your Own GPT Assistant

Highlights:

• OpenAI highlights the growth and success of their AI models, including ChatGPT and GPT-4.
• OpenAI introduces GPT-4 Turbo, a new model with extended context length, logprobs in the API, better world knowledge retrieval, and natural-sounding audio generation.
• Custom Models and the GPT Store are announced to allow developers to create specialized GPT models for specific purposes and share them.
• OpenAI showcases new features like the Assistants API, Code Interpreter, and voice capabilities, emphasizing their commitment to advancing AI technology and improving developer experiences.

Elon Musk Introduces Grok

Elon Musk unveiled a new AI chatbot named Grok, which will integrate with the social media platform X (formerly known as Twitter) and provide real-time access for its top tier of subscribers

Grok is designed to have a humorous approach to interactions, unlike other models deemed overly politically correct by Musk.

• The AI chatbot will be initially released to X Premium+ subscribers after the early beta phase. This subscription plan was launched for $16 per month offering benefits like no ads.
• Musk founded xAI and hired researchers from OpenAI, Google DeepMind, Tesla, and the University of Toronto to work on this project.
• Grok was named after a term from Robert Heinlein's science fiction novel, meaning to understand something thoroughly and intuitively.
• Humor is a key feature of Grok's interaction, showcased by a sample interaction where it sarcastically responds to an illegal request before providing a lawful and ethical disclaimer.
• [Read more]

Mozilla’s New AI “Fakespot” Identifies Fake Product Reviews

Mozilla acquired Fakespot, a company using AI to identify fake product reviews, and launched Fakespot Chat, a large language model (LLM) AI agent to aid online shoppers​​.

• Fakespot Chat allows users to interact with an AI chatbot to ask questions about products they are considering, emulating an in-store salesperson experience​​.
• The AI chatbot is designed to discern real reviews from fake ones and helps users make better purchasing decisions without the need for an account or personal information​​.
• The feature integrates with Fakespot's browser extension and can be used with the Fakespot Analyzer tool, providing a seamless shopping research experience​​.
• Mozilla encourages user feedback on Fakespot Chat to improve the AI model, aiming to streamline product research and improve the quality of purchase decisions
• [Read more]

Google's AI Search Goes Global

Google's AI-powered Search Generative Experience (SGE) is now available in over 120 countries.

• SGE supports new languages: Spanish, Portuguese, Korean, and Indonesian, in addition to English, Hindi, and Japanese.
• Users can interact with the AI in a conversational manner to get comprehensive answers.
• Google is testing a feature for follow-up questions to deepen exploration on a topic, integrating search ads in the process.
• SGE's translation feature now allows for clarification on words with multiple meanings.
• The definitions feature within AI-powered overviews has expanded to include coding and health information.
• Google plans to continue expanding and innovating its AI-powered search globally​.

GitHub's AI Copilot for Enterprises

GitHub is introducing an enterprise subscription for Copilot, allowing customization based on internal codebases.

• Priced at $39/month, Copilot Enterprise launches in February 2024 with features beyond the current business plan.
• Enterprises can connect Copilot to their codebase, enabling developers to receive suggestions for private and internal code.
• Copilot Chat will be integrated into GitHub.com, enhancing code, documentation, and pull request interactions.
• Copilot Enterprise will allow personalization and fine-tuning of the AI model for specific company needs.
• During testing, AMD was able to use Copilot for specialized hardware design languages like Verilog.
• [Read more]

Sutro's Code-Free App Creation

Sutro allows building production-ready apps for web, iOS, and Android without coding experience.

• Founders can focus on their ideas while Sutro automates app-building aspects like AI, product management, design, hosting, and scaling.
• Sutro combines AI, particularly GPT-4 and other LLMs, with rule-based compilers to build apps.
• The platform enables high-level changes to apps, like adding features, through simple prompts.
• It offers a self-serve mode and the option for collaboration with Sutro’s team.
• Sutro charges a subscription fee for its services, aiming to be more cost-effective than hiring a development team.
• As of the report, 715 makers have built 934 apps using Sutro.

Meta and Hugging Face's AI Startup Accelerator

• The program aims to foster an "open and collaborative" approach to AI development in the French tech scene.
• The partnership is in line with Meta's stance on openness in AI, as reflected in a recent open letter advocating against rapid AI regulation.
• Hugging Face is establishing itself as an open-source alternative to AI models like OpenAI's ChatGPT.
• Startups can apply to join the AI Startup Program, with selected winners receiving mentoring, access to platforms, and compute resources.
• Meta's choice to work with Hugging Face and Scaleway reflects a deliberate move away from major cloud providers like Microsoft and Google.
• [Read more]

Samsung's New AI Model - Gauss

Samsung Gauss was unveiled at the Samsung AI Forum 2023

• It includes Samsung Gauss Language for human language understanding, Samsung Gauss Code for aiding code development, and Samsung Gauss Image for image generation and editing.
• Samsung plans to integrate Gauss into its smart devices like phones and laptops.
• Currently, Gauss is used internally at Samsung but will be available to the public soon.
• Samsung has an AI Red Team focusing on the security and ethics of AI use.
• Gauss was named after mathematician Carl Friedrich Gauss, influential in AI and machine learning.
• The launch of Samsung Gauss follows Samsung's earlier ban on generative AI tools due to an internal data leak.
• [Read more]

Microsoft’s AI Chip Initiative

Microsoft’s own AI chip, codenamed Project Athena, could help cut their ballooning AI costs by up to a third. With ChatGPT costing OpenAI an estimated $700,000 per day to run, computing expenses are top of mind for many AI companies.

• Microsoft is developing its own AI chip, Project Athena, aiming to reduce reliance on Nvidia and cut machine learning costs, potentially by up to a third. This could significantly reduce the cost of running AI services like ChatGPT, which currently has substantial operating expenses​​.
• Microsoft plans a full-scale launch by 2024 and has already made prototypes available to select employees for testing. The company aims to implement these chips broadly within its operations by early next year​​.
• Microsoft's move to create its own AI chip aligns with its aggressive AI strategy and efforts to reduce operating costs, particularly in its search engine and cloud services.[Read more]

Humane Unveils the AI Pin

The Ai Pin can be magnetically attached to clothing, featuring a camera, microphone, depth, and motion sensors, and is powered by a Qualcomm chip.

• Humane officially unveiled the Ai Pin, an ambient computing device priced at $699 with a $24 monthly subscription, set to be available on November 16.
• It operates without a screen, instead using a Laser Ink Display to project information like calls onto the palm.
• The device includes battery boosters for all-day power, a Trust Light for recording privacy, and does not require wake words to activate.
• Ai Pin runs on Cosmos, an AI-infused proprietary OS, and does not require pairing with a smartphone, integrating with Microsoft and OpenAI's models.
• Customization and data management occur via the Humane.center platform.
• The Ai Pin offers features like real-time translation and food recognition to support fitness goals, with Tidal as the first software partner for an AI-driven music experience​.