TL;DR

GTC Keynote from Jensen Huang, CEO of NVIDIA

Jensen Huang, CEO of NVIDIA, at GTC March 2024 discusses the evolution of computing and the emergence of generative AI, a new kind of software that can be used to create other software. He also unveils a new line of products from NVIDIA, including NIM (NVIDIA Inference Microservice), and Omniverse, a digital twin platform that can be used to create and simulate robotic systems.

• Generative AI is a new kind of software that can be used to create other software. This is a major breakthrough that has the potential to revolutionize many industries.
• NVIDIA is a leader in the development of generative AI, and the company is committed to making this technology accessible to everyone. Nims is a new product from NVIDIA that makes it easy for companies to integrate generative AI into their workflows.
• Omniverse is another new product from NVIDIA that is designed to help companies create and simulate robotic systems. Omniverse is a digital twin platform that can be used to create virtual models of real-world systems. These models can then be used to train robots and to test new products and processes.
• NVIDIA believes that generative AI and Omniverse have the potential to transform many industries, including healthcare, manufacturing, and transportation. The company is committed to working with its partners to bring these technologies to market.

The Power of the Ecosystem

The one awesome part of GTC is that NVIDIA is bringing together partners and building a strong ecosystem. There have been so many blogs, press releases, and videos to capture all of the news, but one is the idea that it is going to take many developers, partners, and tech stacks to come together in order to deliver on the promise of AI. I was proud to see that MathWorks is in the ecosystem.

One thing that has captured my interest is the Omniverse Cloud APIs, which I believe are set to revolutionize autonomous system development through advanced sensor simulation. These APIs offer developers crucial tools for creating and testing realistic scenarios, thus enhancing the safety and effectiveness of autonomous technologies. NVIDIA’s initiative not only accelerates the development process but also aims to democratize access to cutting-edge technologies, fostering innovation in the autonomous sector.

By bringing together an expansive ecosystem of simulators, verification and validation (V&V) tools, content and sensor developers, the Omniverse Cloud APIs enable a universal environment for AI system development.

Developers and software vendors such as CARLA, MathWorks, MITRE, Foretellix, and Voxel51 underscore the broad appeal of these APIs for autonomous vehicles.

Here’s a video showing how Foretellix leverages NVIDIA Omniverse Cloud APIs to generate high-fidelity sensor simulations for autonomous vehicle development.

2024 is shaping up to be exciting. AI tech, platforms, partners, and developer interest are hitting at once.

So, what did I do first? Of course, I ChatGPT’d myself.

GPT-4 with browsing had much to say about me and my blog.

Hans Scharler writes on a variety of topics on his blog. His writing ranges from life experiences and tech to comedy, games, marketing, and community-building. Looking at the titles and categories of his recent posts, some specific subjects include:

• Tech: His content includes discussing his journey with different web browsers, tips for using the Windows Subsystem for Linux for full stack development, and the concept of “the final programming language is language”​.
• AI and Software: He has posts that relate to AI and software development, including “Master ChatGPT Prompts: Prompt Engineering for ChatGPT Plus and GPT-4” and “Chatting With Bard AI About Online Communities”​​.
• Community: He also explores community-related topics such as “Virtual Communities are Having Their Moment” and “Why Community Matters More Than Ever in the Age of AI”​​.
• Reviews: Scharler reviews products or ideas, as seen in his post “Build: An Unorthodox Guide to Making Things Worth Making”​.
• Life: There are also posts about everyday life, like “I Learned Everything That I Know About Snow Shoveling From My Neighbor”​​.

In terms of categories, the topics he covers include AI, games, life, comedy, technology, software, hardware, community, audio, BBQ, coffee, entertainment, marketing, projects, reviews, and writing, among others​. His writing is quite broad and eclectic, reflecting a wide range of interests.

This blog post needs a thumbnail. So, I asked ChatGPT to write a prompt for DALL-E.

“A fusion of life, technology, and community in a playful and inspiring setting, symbolizing the eclectic and innovative spirit of Hans Scharler’s blog.”

ChatGPT

I went to DALL-E and entered the prompt.

Whelp.

Midjourney 5 Update

Out of curiosity, Jake tried the same prompt using Midjourney 5. This is a spectacular result. Part of some of these options perfectly articulates what is going on in my head all day long.

What can you do with Web Browsing (GPT-4)?

ChatGPT, as it likes to tell you, is not connected to the internet, and its training data was cut off a couple of years ago. Generally, this does not matter much for writing and code generation, but if you want to know something like Modular Mojo that dropped last week, you are out of luck. With the “Web Browsing (GPT-4)” model update, you can get around this limitation by asking ChatGPT to read a webpage to help with the reply. I am still experimenting with what this all means, but I have found it very useful to summarize web pages and ask a website questions to get up-to-date responses.

For what it’s worth…

Here’s a summary of the upcoming features of the Mojo programming language:

Small Independent Features: Mojo is still lacking several key features that don’t depend strongly on other features. These include tuple support (partially implemented), keyword arguments in functions, improved package management, a more comprehensive standard library, support for top-level code at file scope, algebraic data types (like ‘enum’ in Swift/Rust), and pattern matching.

Ownership and Lifetimes: The ownership system is only partially implemented, with features like capture declarations in closures and a borrow checker already in place. The

Protocols / Traits: Mojo plans to add language support for Protocols, similar to those found in languages like Swift, Rust, Haskell, and C++. This will allow for the definition of requirements for types that conform to them and will dovetail into static and dynamic metaprogramming features​.

Classes: Mojo currently does not support classes, a core feature commonly used in Python. Once the core language semantics are more established, they plan to add full support for all the dynamic features in Python classes.

C/C++ Interop: Mojo intends to integrate transparently with Clang C/C++ modules, with the aim of generating a seamless foreign function interface between C/C++ and Mojo.

Full MLIR Decorator Reflection: The behavior of decorators in Mojo is currently hardcoded in the parser. In the future, decorators will be moved to compile-time metaprograms using MLIR integration.

Sharp Edges: There are several smaller features and behaviors that Mojo currently lacks or doesn’t handle as expected. These include the lack of Python list or dictionary comprehension expressions, no support for defining anonymous functions with the lambda keyword, no support for global variables, and the absence of parametric aliases. Additionally, Mojo’s exception type is called Error rather than Exception, and it doesn’t currently support stack unwinding or contain a stack trace when an error is raised.

Update: May 12, 2023

OpenAI changed to names of the new models on the May 12, 2023 update. ChatGPT PLUS subscribers now see a new menu to select the model that you want to interact with. Under GPT-4, you see Default, Browsing Beta, Code Interpreter Alpha, and Plugins Beta. You also get nifty icons.

This version of the On Air Light is cheaper and easier to build than my previous versions.

From Postcapes,

As a Software Developer at ioBridge and active speaker, and developer (including having over 800 people following his toaster on Twitter) in the Internet of Things space Hans Scharler is someone to watch as he sits on the interesting intersection of DIY’ers and corporate products.

Follow Hans and his latest on Twitter @scharler.