AI Weekly Digest: Unveiling Meta's AI Wizardry – From Natural Language to Breathtaking 3D Web Spaces, and Exploring the Economic Colossus Unleashed by Generative AI

In this world of tech, the language of artificial intelligence is everywhere, pushing the boundaries of human imagination, and there's a lot going down in this week's roundup, so buckle up!

First on our list is ElevenLabs, giving us a groundbreaking tool called the AI Speech Classifier. This genius system can take any piece of audio and, with a simple upload, can tell you if what you're hearing was generated by ElevenLabs' AI.

Now, let's talk Nvidia Research. These folks have gifted us SceneScape, a technology that paints immersive, virtual walkthroughs drawn solely from a simple text prompt. It's like reading a book, but instead of visualizing in your mind, you're walking right through the story.

Then, we have Meta AI. They're introducing the game-changing Image Joint Embedding Predictive Architecture (I-JEPA). This powerful AI model learns from the world in a human-like fashion, mastering computer vision tasks while maintaining computational efficiency. The best part? It builds an internal model of the world, comparing the abstract representations of images rather than just the pixels. Meta is also opening the doors to their code and model checkpoints, giving everyone a chance to experience the power of I-JEPA.

In more Meta news, they're gearing up to commercialize the next version of LLaMA, their open-source LLM, and have unleashed MusicGen, an open-source model that generates music based on text and melody prompts.

Adobe has also stepped up their AI game with the launch of Generative Recolor. This tool, powered by Adobe Firefly generative AI, lets you design custom color schemes simply by typing text prompts.

OpenAI has been busy too, with the unveiling of their function calling capability in the Chat Completions API and some hefty upgrades to gpt-4 and gpt-3.5-turbo, including a new 16k context version. And they've made it even more accessible with a 75% cost reduction on the embeddings model and a 25% cost cut on input tokens for gpt-3.5-turbo.

Brace yourself for a massive economic shift as per a report by McKinsey, which predicts that generative AI could boost the economy by a staggering $2.6 to $4.4 trillion annually across various fields, including customer operations, marketing and sales, software engineering, and R&D.

EU lawmakers are also catching up, passing AI regulations that require review of generative AI systems before commercial release, and moving towards restricting real-time facial recognition.

In Google-land, Google Lens is now equipped to identify skin conditions. What's more, Lens will be integrated with Bard, Google’s AI chatbot, so that Bard can understand images in user prompts.

AMD is dropping its most advanced GPU for artificial intelligence, the MI300X, soon to be in the hands of lucky customers later this year.

Vercel's unveiling its AI SDK - an open-source library to build conversational, streaming, and chat user interfaces with top-notch support for OpenAI, LangChain, and Hugging Face Inference. They've also announced 'Vercel AI Accelerator', a 6-week program that gives developers access to a whopping $850k in free credits from OpenAI and Replicate.

Salesforce is rolling out AI Cloud - generative AI designed for enterprise use. It includes the new Einstein Trust Layer to prevent large-language models from retaining sensitive customer data.

Meanwhile, Cohere and Oracle are joining forces to make it easier for enterprise customers to train their own specialized large language models while keeping their training data private.

Coda is stepping up its AI game with Coda AI - an AI-powered assistant to automate workflows. To sweeten the deal, they've also announced ‘Coda's AI at Work Challenge’ with a total of $40,000 up for grabs for creators of the most useful Coda AI templates.

Lastly, heavyweights OpenAI, Google DeepMind, and Anthropic have pledged to give the UK “early or priority access” to their AI models to bolster research into evaluation and safety.

There's a lot of movement in the AI world and it's fascinating to see how these developments will shape the future. Stay tuned for more updates and keep innovating!