Although we generally agree with the premise of this discussion topic, we have found that even with a master's in C.S., it is very challenging to keep up with "everyday" advancements in A.I. Engineering. It is important to note that even now, many applications of A.I. tech -- such as private local "RAG" apps -- are at the stage where they are "useful" and "almost work" but still cannot be trusted for commercial business applications.
For Example, I am trying to refine my set of RAG programs to store and query many current YT transcripts -- including transcripts from your own "Everyday AI" site. When I asked my RAG program the question: " What are the new features of NotebookLM ? " the Local LLM responded: " Some of the mentioned features and updates include: * Ability to customize the audio overview podcast * Business Edition for NotebookLM, allowing users to join a pilot program * Customization options for business users * Better sales training and onboarding capabilities * Upload existing training manuals * Enhanced customer support * Improved product development tools It seems that the new features are focused on providing more tailored and customized experiences for users, particularly in the context of business and enterprise use cases. ". Although this may be judged as relatively "Good" and "Useful" response, it is limited by the very small context size of my local LLM. The programs to create the vector database and to respond to questions are surprisingly fast using a local NVidia GPU.
Although we generally agree with the premise of this discussion topic, we have found that even with a master's in C.S., it is very challenging to keep up with "everyday" advancements in A.I. Engineering. It is important to note that even now, many applications of A.I. tech -- such as private local "RAG" apps -- are at the stage where they are "useful" and "almost work" but still cannot be trusted for commercial business applications.
For Example, I am trying to refine my set of RAG programs to store and query many current YT transcripts -- including transcripts from your own "Everyday AI" site. When I asked my RAG program the question: " What are the new features of NotebookLM ? " the Local LLM responded: " Some of the mentioned features and updates include:
* Ability to customize the audio overview podcast
* Business Edition for NotebookLM, allowing users to join a pilot program
* Customization options for business users
* Better sales training and onboarding capabilities
* Upload existing training manuals
* Enhanced customer support
* Improved product development tools
It seems that the new features are focused on providing more tailored and customized experiences for users, particularly in the context of business and enterprise use cases. ".
Although this may be judged as relatively "Good" and "Useful" response, it is limited by the very small context size of my local LLM. The programs to create the vector database and to respond to questions are surprisingly fast using a local NVidia GPU.