🍔 Friday - AI Wrap-up #15

AuthorAuthor

Martin Musiol & Daniel Aguspina
August 02, 2024

We are closing in on the strongest shipping week in the last 20 months.

Google, Meta, OpenAI, and Midjourney have released better AI versions. Especially interesting:

  • The agenda for the most enabling course on AI is out ✨

  • Gemini is currently the most powerful LLM that ever existed! 🥇 

  • Runway ML’s video gen conveys emotion like a top actor 🎭

Reading time is precisely 167 sec; Let’s roll! 

own

The world’s most enabling course on AI: Everyone Can Code! ✨ - Agenda’s out.

We present to you the agenda of the world’s most enabling online course (currently, my opinion, soon reality … I am manifesting).

We put great effort into this course. Here’s the agenda.

After the course, you’ll learn how to use AI the right way, and you’ll be able to achieve the following in no time with our frameworks:

  • Build a web app hosted on the cloud, accessible for the whole world.

  • Build a mobile app.

  • Build a browser extension.

  • Build AI solutions, LLMs via API (e.g., OpenAI) and SLMs open-source offline.

  • Build an RAG system, including prompt engineering.

  • Build frontends with a professional look.

  • Build backend-only applications.

  • Build real-world applications and add features to larger projects.

  • Know what’s next, even though this is already the future way of building things.

  • Understand what can’t be built via our frameworks.

  • ⚠️ Side effect: Understand how developers build code professionally, including software architecture and DevOps, e.g., version control, CI/CD, and more.

Join the Waitlist for AI Software Development Course - Learn to Build Software with AI, Access Proven Frameworks, and Get Direct Support from the Instructor

Hope to see you in the course.

As I am finalizing the course, is there something specific you'd like me to cover?

Certain implementation, application, wishes ...?

Login or Subscribe to participate in polls.

Together with Growthschool 🤝 

AI & ChatGPT Mini Crash Course - Eliminate workplace burnout & save 16+ hours every week. Learn 20+ AI tools, prompting techniques & hacks for free.

For the first time, Google DeepMind’s Gemini is the best LLM that ever existed 🥇 

The new Gemini 1.5 Pro (Experimental 0801) delivers robust performance in technical areas like Math, Hard Prompts, and Coding.

🤔 Coding? We’ll make sure to include it in the course.

Runway ML’s Gen-3 is incredible in conveying emotions

Segment Anything 2 - Meta just launched SAM 2, and it's very powerful

(Source)

Why powerful? It is helpful not only for video editing but also for AI to understand the world better/ more in-depth. 

This is crucial for moving forward with AGI.

Further, it has huge implications for other fields. It improves cancer cell segmentation, precision in agriculture, and autonomous vehicle navigation.

Segment Anything Model (SAM): Overview: SAM is an AI model developed by Meta AI capable of "cutting out" any object in any image with a single click. It supports zero-shot generalization, meaning it can segment unfamiliar objects and images without additional training. Input Prompts: SAM uses interactive points, boxes, and other prompts to specify segmentation tasks. Capabilities: Automatically segments entire images. Generates multiple masks for ambiguous prompts. Integrates with other systems, such as AR/VR, for enhanced functionality. Outputs: The generated masks can be used for various applications, including video tracking, image editing, 3D lifting, and creative tasks. Training Data: Trained on over 11 million images and 1.1 billion masks using a model-in-the-loop data engine. Model Design: Features a decoupled architecture with a one-time image encoder and a lightweight mask decoder, enabling efficient processing.

Segment Anything!

OpenAI has released its advanced voice mode (not to all)

From story-telling to our kids, language teachers, and everyday problem solvers, interacting with AI will never be the same.

Interacting with AI will never be the same. The examples in the X-thread below are hyper-realistic (needs to breathe while counting 🫠 ).

AI will be pervasive.

Midjourney 6.1 is here, and it opens up new possibilities

AI-generated images are now officially indistinguishable from taken photographs.

Beautiful sunrise over a misty Swiss landscape, with golden light illuminating a lush green valley and a cascading waterfall surrounded by dense forest.

Generated via Midjourney v6.1

One can use it for app UI ideation.

My prompt …

Design a serene and nostalgic UI for a "Reconnect" app. The home dashboard features a gradient background transitioning from lavender to sky blue. A sleek, transparent navigation bar at the top has a minimalistic heart and friendship bracelet logo. In the center, display a large circular profile picture of the user, surrounded by softly glowing, animated particles. Below it, a cursive welcome message reads: "Welcome back! Let's reconnect." Include a rounded search bar with the placeholder "Find Friends." Beneath the search bar, showcase a card carousel with "Friend" each card displaying a profile picture, name, and shared history snippet, with "Reconnect" and "Skip" buttons at the bottom. The footer has minimalist icons for Home, Search, Messages, and Profile, with a gentle glow when selected. Add small, animated shooting stars in the background for a whimsical touch. —ar 2:1 —v 6.1 —q 2

… results in this great-looking vision of a frontend/ UI.

Beautiful Reconnect App UI: Gradient background with lavender to blue hues, featuring a central profile picture, welcome message, friend suggestions carousel, and minimalist action buttons for reconnecting with friends.

UI generated via Midjourney for an imaginary “Reconnect with old friends app“.

I love it! 😆 

own

It’s a wrap.

I wish you an awesome weekend.

-Martin