Generative AI was one of the great protagonists of GTC 2024, and as it could not be otherwise, this technology also became one of the highlights of this week’s NVIDIA AI Decoded. This technology is enjoying increasing adoption and can be used in more than 100 million computers and workstations which use NVIDIA RTX graphics cards.

When we talk about generative artificial intelligence, we mean a type of artificial intelligence is able to create diverse content, including images, text, music, sounds and even videos. This technology uses large learning models that are able to perform pre-configured tasks for which they have been previously trained. They work through a simple interface that allows you to use natural language directly.

NVIDIA is aware of the importance of generative artificial intelligence, which is why it celebrated recently competition focused on the mentioned technology of which we already know three winners: OutlookLLM, CLARA and Rocket League BotChat, three very interesting projects that are summarized in an application that allows the integration of private generative artificial intelligence in Outlook, a project that improves the PowerShell interface by offering a translation of instructions from English to executable commands and a chatbot for the well-known game RocketLeague.

NVIDIA Chat with RTX becomes ChatRTX

ChatRTX is an artificial intelligence chatbot that uses so-called extended invocation generation, plus NVIDIA TensorRT-LLM software and NVIDIA RTX acceleration. All of this together made it possible to create a platform with generative artificial intelligence capabilities running locally on Windows systems.

Those of you who read us daily already know how it works and know the potential it has. NVIDIA has confirmed that ChatRTX will soon expand its capabilities thanks to voice support, images and new models, an advancement that will make it one of the most advanced platforms in its class, and yes, it will still be fully functional locallywith all that that entails in terms of security and privacy.

When this support is incorporated, we will be able to communicate with ChatRTX using voice and also with photos in a very special and very intelligent way, because we can use search by words, phrases or expressions to find photos in private bookstores with which this chatbot is connected. Interesting, isn’t it? Finally, it is also confirmed that he will have ChatGLM2 support.

A more complete ecosystem that will make our lives easier

And as usual NVIDIA focused on developers. Blackmagic Design and Topaz Labs are confirmed to have integrated RTX AI acceleration into their software solutions, a new feature that will allow developers to take advantage of important benefits that will improve their productivity by reducing work time on each project.

For example, we can highlight the value represented by TensorRT, from double speed various AI effects such as rotoscoping, noise removal, super resolution and video stabilization in DaVinci Resolve and Topaz.

On the other hand, TensorRT-LLM is integrated into very popular developer frameworks and ecosystems such as LangChain, LlamaIndex, Oobabooga and Jan.AI. This will facilitate the creation and deployment of generative artificial intelligence applications both locally, via GPU, and in the cloud.

Developers can also bet on NIM microserviceswhich consist of pre-built AI containers and compatible with standard APIs, profiling them as an optimized solution that will help reduce implementation time from weeks to minutes. They are compatible with over a dozen popular models from companies such as Google, Meta, Microsoft, Shutterstock, NVIDIA, and Getty Images, among others.

We cannot end without talking about NVIDIA Blackwell, another of the big stars of GTC 2024. The new graphics architecture with which NVIDIA reaffirmed its leadership in the world of AI at the hardware level and opens a new horizon for the green giant, as it is the first A GPU with a multi-chip design introduced by NVIDIA (the B200 GPU consists of two interconnected chips that share a capsule).

AI generated cover image.