Imagine yourself seated at your desk, facing the frustrating challenge of locating a document buried deep within the archives, dating back to a considerable time ago. The document seems to have mysteriously vanished from its designated folder, and your quest involves an extensive and time-consuming search, with no guarantee of success.
Now, picture yourself effortlessly asking your computer, "Can you retrieve the signed contract from April of 2016 pertaining to [the relevant service]?" In a matter of seconds, your computer swiftly provides you with the precise information you sought.
This revolutionary capability is made possible by Retrieval-Augmented Generation (RAG), a groundbreaking technology highlighted by the tech giant NVIDIA at the recent Consumer Electronics Show.
Much like other artificial intelligence models, RAG operates based on the data it is fed. In the foreseeable future, many businesses are expected to adopt a closed-off version of this technology, specifically tailored to search through their servers and data repositories. For instance, it could seamlessly scour through your emails, answering queries like, "What was the name of the restaurant in Nashville that Jonathan recommended last May?" or "Which vendor did Veronica utilize last year?"
nVidia dubs this cutting-edge feature as "Chat with RTX," leveraging the popularity of their brand associated with video processing cards.
According to NVIDIA, “Almost any business can turn its technical or policy manuals, videos or logs into resources called knowledge bases that can enhance large language models. These sources can enable use cases such as customer or field support, employee training and developer productivity.”
Several prominent companies have already embraced Retrieval-Augmented Generation, including AWS, IBM, Glean, Google, Microsoft, and Oracle, as per NVIDIA's statements.
The evolution from traditional search engines like AskJeeves to this advanced technology marks a significant leap forward in the realm of information retrieval and utilization.