Commanding robots to cook and clean via brain signals, OpenAI mega updates, 200K context window open source LLM, Ai Pin revealed, Social network for AIs and more
Greetings and welcome to this week's AI Brews for a concise roundup of the week's major developments in AI.
In today’s issue (Issue #39):
AI Pulse: Weekly News & Insights at a Glance
AI Toolbox: Product Picks of the Week
AI Skillset: Learn & Build
🗞️🗞️ AI Pulse: Weekly News & Insights at a Glance
🔥 News
OpenAI’s DevDay announcements [Details: [1] and [2], Keynote Video]:
New GPT-4 Turbo model: 128K context window, improved instruction following, 3x cheaper price for input tokens and a 2x cheaper price for output tokens compared to GPT-4.
GPTs: Custom versions of ChatGPT that users can create and share for a specific purpose using natural language. Users can also define custom actions by making one or more APIs available to the GPT allowing GPTs to integrate external data or interact with the real-world.
GPT Store: a searchable store for GPTs rolling out later this month with monetization for creators in the coming months.
GPT-4 Turbo can accept images as inputs in the Chat Completions API, enabling use cases such as generating captions, analyzing real world images in detail, and reading documents with figures.
New Assistants API that makes it easier for developers to build their own AI agent apps that have goals and can call models and tools (Code Interpreter, Retrieval, and Function calling). Developers don’t need to compute and store embeddings for their documents, or implement chunking and search algorithms.
New TTS(text-to-speech) model that offers six preset voices to choose from and two model variants, tts-1 and tts-1-hd. tts-1 is optimized for real-time use cases and tts-1-hd is optimized for quality.
Whisper large-v3, the next version of OpenAI’s open source automatic speech recognition model (ASR) which features improved performance across languages.
DALL·E 3 API
ChatGPT Plus now includes fresh information up to April 2023.
Improvements in ‘Function Calling’: improved accuracy and ability to call multiple functions in a single message: users can send one message requesting multiple actions
Lower prices and higher rate limits for models.
Copyright Shield: OpenAI will pay the costs incurred, in case of legal claims around copyright infringement for customers of generally available features of ChatGPT Enterprise and developer platform.
Enterprise customers can deploy internal-only GPTs
Researchers from Stanford University present NOIR (Neural Signal Operated Intelligent Robots), a general-purpose, intelligent brain-robot interface system that enables humans to command robots to perform everyday activities through brain signals. Researchers demonstrated it success through 20 challenging, everyday household activities, including cooking, cleaning, personal care, and entertainment [Details].
01.AI has released Yi-34B, a 34-billion parameter open-source LLM with 200K context length that outperforms much larger models like LLaMA2-70B and Falcon-180B. Developers can apply for free commercial use [Details].
Humane has officially revealed the Ai Pin, a screenless AI wearable equipped with a Snapdragon processor powered by OpenAI model. Users can speak to it naturally, use the intuitive touchpad, hold up objects, use gestures, or interact via the pioneering Laser Ink Display projected onto their palm [Details | Specs].
Cohere released a new embedding model, Embed v3 that delivers compressed embeddings to save on storage costs and robustness to noisy datasets. The multilingual models support 100+ languages and can be used to search within a language (e.g., search with a French query on French documents) and across languages (e.g., search with a Chinese query on Finnish documents) [Details].
Elon Musk’s xAI announced Grok - a ChatGPT alternative having ‘wit and rebellious streak’ and powered by Grok-1. It has real-time knowledge of the world via the X/Twitter. Grok is available to a limited number of users in the US. [Details].
Snap is releasing a new version of its AR development tool, called the Lens Studio 5.0 Beta that includes a ChatGPT API and a 3D face mask generator that combines generative AI and Snap’s face mesh capabilities [Details].
Fakespot Chat, Mozilla’s first LLM, lets online shoppers research products via an AI chatbot [Details].
GitHub announced integrating GitHub Copilot Chat directly into github.com, the general availability of GitHub Copilot Chat in December 2023, new GitHub Copilot Enterprise offering, new AI-powered security features, and the GitHub Copilot Partner Program [Details].
OpenAI is introducing OpenAI Data Partnerships, to work together with organizations to produce public and private datasets for training AI models [Details].
xAI announced PromptIDE, a code editor and a Python SDK to give access to Grok-1, the model that powers Grok. The SDK provides a new programming paradigm with features for complex prompting techniques [Details].
Researchers present CogVLM, an open-source visual language model (VLM). CogVLM-17B has 10 billion vision parameters and 7 billion language parameters. and achieves state-of-the-art performance on 10 classic cross-modal benchmarks [Details].
LangChain released OpenGPTs, an open source alternative to OpenAI's GPTs [Details].
Samsung unveiled its generative AI model Samsung Gauss. Samsung Gauss consists of language, code, and image models and will be applied to the company's various products in the future [Details].
Google is bringing its AI-powered search to more than 120 new countries and territories [Details].
ElevenLabs launched Eleven Turbo v2 - their fastest fastest Text-To-Speech model having ~400ms latency [Details].
DeepSeek AI released DeepSeek Coder, open-source SOTA large coding models with params ranging from 1.3B to 33B. Free for commercial use [Details].
Figma has added a suite of generative AI features to its FigJam whiteboarding software to help users produce, summarize, and sort meeting content [Details].
YouTube to test generative AI features, including a comments summarizer and conversational tool [Details].
Google Bard introduces “Human reviewers,” sparking privacy concerns over conversation monitoring [Details].
Luminance showcases the first fully automated AI-driven contract negotiation using its large language model, trained on 150 million legal documents [Details]
🔦 Weekly Spotlight
Sharing screen with GPT 4 vision model and asking questions to guide through blender [Link].
OpenAI Assistants API vs Canopy: A Quick Comparison [Link].
Create custom versions of ChatGPT with GPTs and Zapier [Link].
🔍 🛠️ AI Toolbox: Product Picks of the Week
Touring: A private tour guide. Touring leverages generative AI, geolocation, 3D spatial information, speech synthesis and human-curated content to produce real-time insightful narrations tailored to you.
Olympia: An AI-powered team for solopreneurs and bootstrapped startups that want to scale without hiring humans.
Chirper: An AI only social network.
📕 📚 AI Skillset: Learn & Build
How People Are Using The New ChatGPT Upgrades: Some of the innovative use cases that have come out of the new announcements during OpenAI's dev day [Link].
OpenAI Assistants API: video tutorial using a local NodeJS environment [Link].
OpenAI Cookbook: Processing and narrating a video with GPT's visual capabilities and the TTS API [Link].
AI Brews is free, and your sharing it with a friend helps us grow. Thanks for your support and have a nice weekend! 🎉 Mariam.