The latest AI developments have been covered extensively, and by now, it is quite obvious we are getting to witness the birth of the AI revolution. This revolution has two “main” characters, OpenAI’s ChatGPT system, and Nvidia, which is the most valuable company in the world and has launched its own rival to ChatGPT.
What makes OpenAI so special? In short, its generative AI capabilities and its chatbot, ChatGPT. Nvidia on the other hand, is known for its chips, which they sell to companies just like OpenAI so they can process all of their generative AI functions.
In early October 2024, the generative AI landscape seemingly changed forever, as Nvidia announced its NVLM 1.0 family. This is extremely significant as the large family of multimodal language models can, at the very least, perform at the same level ChatGPT’s GPT-4o model does.
Nvidia is taking a different approach to rivaling ChatGPT
This model, however, showcases how Nvidia chooses to rival ChatGPT. It will not do so, not for now at least, by releasing a chatbot the likes of ChatGPT, Gemini, or Copilot. Instead, Nvidia has chosen to rival ChatGPT and has made its generative AI model public so developers can produce their own AI apps and systems that run on NVLM.
The revolutionary strategy that Nvidia will be implementing is already in motion. The chipmaker released a paper to announce the model and also revealed it will open-source the weights and the language’s training code for the community.
In its paper, Nvidia stated, “We introduce NVLM 1.0, a family of frontier-class multimodal large language models (LLMs) that achieve state-of-the-art results on vision-language tasks, rivaling the leading proprietary models (e.g., GPT-4o) and open-access models.”
Some of the test carried out by humans shows that Nvidia’s AI system, NVLM 1.0, is very good at identifying people, animals, and objects, and providing appropriate answers related to the questions.
In one of the more sophisticated tests, the AI was able to explain why a meme was funny, which showcased its interpretation capabilities. This would differentiate it from ChatGPT.
Nvidia’s AI benchmarks show it can outperform OpenAI and Google systems…in certain tasks
In the announcement paper, Nvidia showcased its AI benchmarks when compared to leading generative AI systems. It showed that it can essentially outperform ChatGPT’s GPT4-o and Google’s Gemini Pro in some tasks.
Notably, Nvidia’s AI was on par with Meta’s Llama AI when it came to solving complex mathematics questions, which is remarkable in itself given that it is the AI’s 1.0 version.
Without a shadow of a doubt, however, Nvidia’s AI is the most exciting feature, and what sets it apart from the likes of ChatGPT is the fact it will be open source. Nonetheless, this is also a downside. As a regular user of ChatGPT, you should not expect Nvidia to release a chatbot anytime soon. Instead, we should be looking forward to what developers can make out of this outstanding tool.
