NVIDIA’s Conversation Starter

NVIDIA’s wants to make chatbots that can handle both work and play.

Photo by Galeanu Mihai on iStock.

Sign up to uncover the latest in emerging technology.

NVIDIA wants to build a chatbot for all seasons. 

The company wants to patent a conversational AI platform that combines both closed and open domain styles of dialogs. For reference: A closed domain chatbot can only understand certain topics, while an open domain one is built to understand any kind of conversation, each one coming with its own pros and cons. 

NVIDIA’s invention relies on bringing those pros and cons together, analyzing the textual input and routing it to whichever style of domain makes the most sense, housing both domains into a single “intelligent dialog management system” on the user-end.

For instance, say a city integrated a chatbot into its tourism website. If a person asks the chatbot a specific question, such as “Find the top-rated Italian restaurants in the city,” the question will be routed to a closed domain for the topic of restaurants. But if they asked a more open-ended request, like “Tell me about the local music scene” the input would route to an open domain, allowing the chatbot to respond in a more colloquial and casual way. (It’s unclear if it could answer hard-hitting questions like “How do you like spaghetti vongole?”)

NVIDIA said this method mitigates the problem of a chatbot or virtual assistant only being able to give broad responses, or very narrow ones. Another point from this filing: the company noted that this tech can apply both to speech and textual inputs, and that the device accepting these commands could be anything from a chatbot to an autonomous vehicle. 

“Users often rely on software applications—such as chatbots, virtual assistants, social bots, and other forms of AI— to submit queries and conduct conversations,” NVIDIA said in its filing. “However, these software applications generally use dialog systems configured to provide either broad or narrow responses, but not both.”

Photo via the U.S. Patent and Trademark Office.

NVIDIA is undoubtedly a leader in the AI sector, specifically in building the hardware and components that are fundamental to building AI. The company reportedly controls 80% of the market for the chips and GPUs that power AI at major tech firms, including OpenAI’s ever-growing ChatGPT. The company far outpaces competitors like AMD and Intel, which hold 20% and less than 1% of the market share, respectively. 

That said, NVIDIA’s isn’t usually the first company name that many conjure when thinking of AI software offerings themselves. Power players like Google, Microsoft and OpenAI continue to dominate the broader public conversation around AI, specifically chatbots. 

By patenting, developing and advancing its conversational AI the company may be trying to branch out and “build an ecosystem,” according to Romeo Alvarez, director and research analyst at William O’Neil.  He added that the enterprise AI software market represents a $150 billion opportunity for NVIDIA.

“It’s probably a factor that most people don’t know because they see NVIDIA as a GPU company,” said Alvarez. “They’re definitely more than that. By also developing AI software platforms, their (total addressable market) expands.” 

One other possibility: This tech could be wrapped up in NVIDIA’s autonomous vehicles unit. The company noted that this kind of conversational AI could be integrated into several kinds of tech, including a “vehicle display” that utilizes speech recognition. In any case, with the strong standing that NVIDIA has in the AI sector, the possibilities are endless. 

Have any comments, tips or suggestions? Drop us a line! Email at admin@patentdrop.xyz or shoot us a DM on Twitter @patentdrop. If you want to get Patent Drop in your inbox, click here to subscribe.