OpenAI Hiring Developers at $10M/yr

Author

AI Breakfast
November 13, 2023

Good morning. It’s Monday, November 13th.

Did you know: On this day in 2006, Google completed its acquisition of YouTube for $1.65 billion.

In today’s email:

  • AI Industry Growth and Collaborations

  • AI in International Policy and Security

  • AI Innovations in Tech and Services

  • AI in Healthcare and Research

  • AI in Entertainment and Media

  • 5 New AI Tools

  • Latest AI Research Papers

You read. We listen. Let us know what you think of this edition by replying to this email, or DM us on Twitter.

Today’s trending AI news stories

AI Industry Growth and Collaborations

Google in talks to invest in AI startup Character.AI, an AI chatbot startup co-founded by ex-Google employees. This move could intensify Google AI’s capabilities in creating interactive AI companions. Character.ai is known for its AI chatbots mimicking celebrities and customizable virtual assistants.

OpenAI lures Google's top AI researchers with multimillion-dollar offers, aggressively recruiting Google’s top AI researchers with offers ranging between $5 to $10 million annually, primarily in stock options. This move is part of AI’s strategy as it approaches a significant increase in valuation to over $80 billion. The competition for AI talent is intensifying, with OpenAI focusing on those working on Google’s Gemini AI models. Meanwhile, Microsoft is developing its own AI chip, “Athena,” to rival Nvidia and support OpenAI’s hardware needs.

News Corp. seeks AI licensing deals for its content to keep generative AI up-to-date. The company wants to keep these AI systems updated with constant, relevant inputs. This follows News Corp's recent revenue report and mirrors the AP's deal with OpenAI. Amidst these developments, Meta has expressed concerns about the high costs of licensing. At the same time, Google and OpenAI argue for the fair use of copyrighted data in AI training as transformative, not violating copyright laws.

AI in International Policy and Security

Biden, Xi set to pledge ban on AI in autonomous weapons like drones, nuclear warhead control. This significant issue will be central to their bilateral talks, aiming to maintain human control in nuclear systems. This move comes amid growing concerns about AI’s unregulated role in conflicts.

AI Innovations in Tech and Services

RunwayML text-to-video AI service introduces a new motion feature “Motion Brush,” allowing users to animate specific areas of an image. This feature is part of an upgrade to the Gen-2 model, which now supports higher resolution video creation of up to 2,816 x 1,536 pixels, surpassing standard Full-HD resolution. This enhancement boosts the fidelity and consistency of videos generated from existing images.

Prompt Engineering plus AlphaZero: Microsoft's XOT improves LLM's ability to generalize Microsoft, in collaboration with Georgia Institute of Technology and East China Normal University, has developed “Everything of Thoughts” (XOT), a new method to enhance language models’ problem-solving abilities. Inspired by AlphaZero, XOT integrates external domain knowledge using Monte Carlo Tree Search and reinforcement learning, improving the models’ ability to generalize to new problems. This innovative approach has shown promising results in various challenging tasks, indicating advancement in AI problem-solving techniques.

Snap launches generative AI lenses that creates AR effects based on text Snap's latest Lens Studio 5.0 brings new generative AI lens development tools and a ChatGPT integration API. Lens Studio 5.0 uses AI to facilitate content creation and interactive experiences, employing partnerships with DALL-E and OpenAI for real-time, AI-generated face mapping and effects. It also integrates with Meshy for material and texture generation, simplifying AR experience development. A notable addition is the "remote API" allowing developers to embed ChatGPT into lenses, broadening AI's accessibility to a wider audience through popular social media platforms.

Microsoft Stock Notches Record High Ahead Of AI News as the tech giant gears up to reveal new AI advancements at an upcoming conference. The shares reached $370.09, surpassing the buy point on IBD MarketSmith charts, and closed the session with a 2.5% increase. The surge follows a series of gains attributed to Microsoft’s strong fiscal first-quarter results, driven by its flourishing cloud computing segment. The upcoming Ignite conference in Seattle, where Microsoft will present its AI initiatives, is eagerly anticipated as the next potential boost for the company’s shares.

AI in Healthcare and Research

AI that reads brain scans shows promise for finding Alzheimer’s genes with more than 90% accuracy. This groundbreaking approach uses brain imaging as a visual indicator to pinpoint genes linked to Alzheimer's. Led by neuroscientist Paul Thompson, the project merges genetic, cognitive, and imaging data, offering significant advancements in treatment discovery and disease risk assessment.

AI in Entertainment and Media

AI Could Make Grand Theft Auto NPCs 'Really Interesting and Fun', Says Take-Two CEO CEO Strauss Zelnick believes AI could make NPCs in the next Grand Theft Auto game more engaging by diversifying dialogue interactions. This technology, he suggests, could enhance the gaming experience by adding complexity and variety to non-playable characters, potentially re-attracting previous players. While Zelnick confirmed Rockstar is working on a new Grand Theft Auto iteration, details on AI integration remain undisclosed.

Cyborg pianist: Transforming classical music with AI and brain data Zubin Kanga, a pianist, composer, and technologist, has premiered his innovative work "Cyborg Pianist" at Kings Place, London. This unique performance blends classical piano with advanced technologies like MiMu motion detection gloves, brainwave sensors, and artificial intelligence. Kanga's ambition is to revolutionize classical piano, creating a new repertoire incorporating these cutting-edge tools, and heralding a new era of 'cyborg musicians' and modern music creation.

5 new AI-powered tools from around the web

​​Intelis elevates Shopify stores with AI-driven dynamic pricing, offering real-time market data and actionable insights for optimal pricing strategies. Users can adjust prices, maximizing profitability and staying ahead in the competitive e-commerce landscape.

Customers.ai revolutionizes lead generation by converting website traffic into email and contact info. It enriches B2C/B2B data for AI emails, restores ad retargeting capabilities, and offers email follow-ups, tackling the challenges of cookie/script deletion in digital marketing.

Vidiofy revolutionizes digital storytelling, turning text into dynamic vertical videos with AI. Perfect for brands and publishers, it offers custom templates, licensed footage, diverse music, multiple languages, and AI narration.

 Voc.AI offers a GPT-4 turbocharged AI chatbot that resolves 80% of customer queries without code. It learns from past interactions and website content, integrating with email, Zendesk, Shopify, live chat, and more, optimizing customer support efficiency.

GPT-4 Vision Chatbot Builder merges GPT-4’s advanced language capabilities with image processing, enabling chatbots to understand and respond to both text and images. This no-code tool enhances interactivity and broadens applications, revolutionizing AI chatbots’ functionality.

arXiv is a free online library where scientists share their research papers before they are published. Here are the top AI papers for today.

LOGIPT, a new language model, uniquely emulates logical solvers for deductive reasoning, sidestepping typical parsing errors in existing solver-augmented models. Fine-tuned on a custom dataset, it directly interprets and answers complex logical questions, showing superior performance over conventional Large Language Models like ChatGPT and GPT-4 on two public deductive reasoning datasets.

The study explores GPT-4V’s integration in autonomous driving, focusing on its advanced scene comprehension and causal reasoning capabilities. GPT-4V excels in processing diverse traffic scenarios, including corner cases and decision-making. It adeptly recognizes environmental conditions and traffic participants using image understanding. However, it struggles with precise direction discernment, accurate traffic light recognition, and complex spatial reasoning tasks. The research highlights GPT-4V’s potential to surpass existing systems in nuanced scenario understanding.

JARVIS-1, a multi-task agent for Minecraft, excels in long-horizon, open-world tasks using a memory-augmented multimodal language model. It processes visual and textual inputs to form plans, utilizing a multimodal memory for learned knowledge and in-game experiences. This innovative approach enables adaptive planning and self-improvement. Excelling in over 200 Minecraft tasks, JARVIS-1 surpasses existing systems in complex challenges, marking a breakthrough in generalist agent development.

Instant3D introduces a rapid method for generating high-quality, diverse, 3D assets from text prompts. Utilizing a two-stage process, it first creates four-structured views from text with a fine-tuned 2D text-to-image diffusion model, then reconstructs Neural Radiance Fields (NeRF) from these images using a transformer-based reconstructor. This approach, trained on 750K 3D objects, outpaces traditional optimization-based methods, producing 3D assets in about 20 seconds, a stark contrast to the hours taken by previous methods.

PE2 is a novel method for automatic prompt engineering of large language models (LLMs), focusing on the construction of meta-prompts that effectively guide LLMs in this process. It employs detailed instructions, context specification, and concepts analogous to optimization techniques like batch size and step size. PE2 consistently outperforms existing methods across various datasets, demonstrating strong performance in counterfactual reasoning and the ability to amend incomplete or erroneous prompts.

Thank you for reading today’s edition.

Your feedback is valuable.


Respond to this email and tell us how you think we could add more value to this newsletter.

Did you know we wrote a book about AI? Check it out here for $4.99