AI Robots are Around the Corner 🤯

Hey there 👋

Welcome to this drama-filled week in AI, with one side calling for regulations and the other building AI robots. Sounds fun, right? But wait, there's more!

EU took Uncle Sam's congressional hearing speech on AI safety to heart and made a few interns work around the clock to come up with an AI Amendment Act that will severely hamper the state of AI innovation. Now, as a result, all these bright-eyed European startup founders are lining up for U.S. visas faster than you can say the word 'artificial intelligence'.

Just when we thought we'd heard it all, Elon Musk pops up on CNBC, throwing around zingers like 'I am the reason that OpenAI exists' and 'Working from home is morally wrong'! That's enough puns for now; let's dive deep into this week's content:

This issue covers:

• Latest Developments 🌍

• News from the Industry 🧑‍🏫

• Tools of the Trade ⚒️

• Hot Takes 🔥

• AI Meme of the Week 🤡

Latest Developments 🌍

Our Pick 👌

Phoenix by Sanctuary AI: A humanoid general-purpose robot powered by Sanctuary AI’s Carbon AI control system designed to give Phoenix human-like intelligence and physical capabilities.

• Guidance by Microsoft: Enables controlling language models more effectively than traditional prompting or chaining through streamlined prompts and output structures.

• SoundStorm: Fast and consistent audio generation model capable of producing high-quality 30 seconds of audio in 0.5 seconds.

• Using AI to Detect Alzheimer’s: The paper explores using speech data and domain knowledge to detect Alzheimer's dementia, achieving 69% accuracy.

• TorToise: A multi-voice text-to-speech system that applies advances in image generation to improve speech synthesis through scaling.

• CodeT5+: A family of encoder-decoder LLMs for understanding and generating code that addresses the limitations of existing models.

• Make-An-Animation: Text-guided 3D human motion generation model that leverages large-scale image-text datasets to improve performance and diversity.

• Leveraging LLMs in Conversational Recommender Systems: By enabling real-time dialogues, controlling conversations, and providing explainable recommendations.

• ULIP-2: A scalable multimodal pre-training framework that utilizes LLMs to automatically generate language counterparts for 3D objects.

• GPT-Sentinel: Distinguishes human-written and ChatGPT-generated text with high accuracy using language models and showcases key differentiating features.

• ArtGPT-4: Artistic vision-language model with adapter-enhanced MiniGPT-4, excelling in generating and understanding artistic visuals and language.

• MEGABYTE: Multi-scale decoder architecture that enables efficient modeling of long sequences without tokenization.

• Towards Expert-Level Medical Question Answering with LLMs: Med-PaLM 2 achieved a new SOTA performance of 86.5% on the MedQA dataset and significant improvements compared to previous models.

• AR-Diffusion: Auto-regressive diffusion model for text generation outperforming existing models and incorporating token dependencies for better results.

• DarkBERT: Language model specifically trained on Dark Web data, provides valuable insights for research on the Dark Web domain.

• Dr. LLaMA: Enhances small language models in domain-specific QA through generative data augmentation, achieving better performance on medical QA tasks.

• TinyStories: Challenges the notion that LLMs are necessary for coherent text generation, demonstrating that small models can produce fluent stories.

News from the Industry 🧑‍🏫

Our Pick 👌

Exiii Inc. has developed a backpack with six robotic arms, Jizai Arms, that can be controlled, offering potential benefits for tasks and assisting disable individuals.

• OpenAI launched ChatGPT app for iOS that syncs your history across devices, and brings the newest model improvements.

• Sam Altman, OpenAI CEO, testified before a Senate and advocated for the regulation of AI, acknowledging the need to manage the potential risks and harms associated with the technology. Watch the full hearing here.

• The EU's amended AI Act aims to ban American companies from providing API access to generative AI models, targeting open-source software and potentially putting American small businesses at risk. Read the highlights here.

• 'I am the reason OpenAI exists' claims Elon Musk and criticizes the company for turning towards a for-profit model.

• Meta has created a special computer chip called MTIA v1 designed specifically for recommendation systems and is integrated into PyTorch.

• Google is introducing AI coding features powered by Codey, providing code completions, natural language to code generation, and a code-assisting chatbot

• Stability AI releases StableStudio, an open-source version of their DreamStudio text-to-image consumer application.

• Meta has developed CodeCompose, a generative AI tool for coding similar to GitHub Copilot, which provides code suggestions for Python and other languages.

• Perplexity AI has introduced Perplexity Copilot, an interactive AI search companion that guides your search and provides personalized answers.

• OpenAI adds a new button “Continue generating” to ChatGPT UI for generating long outputs.

• Amazon plans to incorporate a ChatGPT-like AI chatbot into its product search engine in order to improve user experience and compete with Microsoft and Google, as indicated by a recent job posting from Amazon.

• Google plans to use AI to create ads and provide video ideas for YouTubers, and also AI chatbots for customer support and Play Store listings.

• Apple has restricted its employees from using ChatGPT and GitHub Copilot due to concerns over data leaks and potential privacy violations.

• AI pilots could potentially replace human pilots in passenger planes, leading to single-pilot aircraft, according to Emirates Airline's President.

• Hippocratic AI has raised $50 million to develop a LLM powering voice and text chatbots in healthcare to address the predicted shortage of healthcare workers.

• OpenAI is preparing to release a new open-source AI model in response to the growing availability of open-source alternatives, but it is unlikely to match the capabilities of its proprietary model, GPT.

• Poe API is being launched to all developers, aiming to provide broad access to LLM-based services and enable the development of diverse application.

• Cloudflare brings AI with Constellation, allowing developers to run pre-trained machine learning models and perform inference tasks on Cloudflare Workers.

• Zoom announces partnership with Anthropic to bring Claude chatbot to Zoom products, starting with the Zoom Contact Center.

• Apple introduces new accessibility features, including custom text-to-speech voices and improved tools for cognitive disabilities, vision impairment, and speech assistance.

• Google’s high-performance computing (HPC) and generative AI, such as Med-PaLM 2, are expected to accelerate drug discovery and precision medicine.

Tools of the Trade ⚒️

Our Pick 👌

Vondy AI: 100+ generative AI tools all bundled into one API designed for developers. Join the waitlist now!

• DragGAN: Manipulate images with just a few clicks! Point and drag to control pose, shape, and expression.

• Recap: Open-source browser extension to summarize text on webpage with ChatGPT.

• Cognosys: Web-based AI agent that revolutionizes productivity, simplifies complex tasks, and prioritizes customer focus.

• Bizway: Turn your ideas into business plans in minutes, provides customizable roadmaps, auto-generating and completing tasks.

• KIIT: ChatGPT with live audio/video capabilities that can answer questions, take meeting notes, and act as a multilingual translator.

• DecisionMentor: AI-powered decision making app that provides chat-based suggestions, scientific approach, visualization, and social sharing.

• Facia: Fast and accurate face verification with AI, preventing fraud and offering a secure and frictionless customer experience.

• Humata: AI-powered tool for faster learning, research, and analysis, with instant question-answering and paper summarization capabilities.

• Dora: Powerful web design tool that enables code-free creation of 3D and animated sites, offering advanced animation capabilities.

• Llama Chat: Free and open-source chat platform to interact with LLaMa, Alpaca, and GPT4All models on Mac.

• Bricabarc: App generator to instantly creates apps based on descriptions, requiring no coding, allows easy design tweaks and feature additions.

• DapperGPT: Enhanced interface for ChatGPT, with features like customization, voice-text conversions, and AI-powered notes.

• PrintNanny: Provides constant quality control for 3D print farms, offering automated notifications and actions in case of any printing quality issues.

• ChatGPT Fund: A public experiment testing ChatGPT's ability to outperform human money managers in portfolio management.

Hot Takes 🔥

AI Meme of the Week 🤡

That’s all for this week!

Will see you next Saturday with more such content. Don’t forget to subscribe and give your feedback below.

BONUS 🎉

Share this newsletter with three other friends and stand a chance to win my book GPT-3: The Ultimate Guide to build NLP Products with OpenAI API. Winners will be selected on a monthly basis.