In the movie ‘Her,’ Samantha, the AI operating system, tells him how other AI systems have created a ‘philosopher AI’ from learning everything about a person. This made me think that nowadays that’s possible. The idea also made sense because of Yuval Noah Harari’s book ‘Nexus,’ where he mentions a book being able to react back or answer back your questions in its tone and mood. His example included religious books and people getting answers from them in a new way of interaction. Many services already exist for this, including those utilizing local models.
RAG in LLMs can achieve this; am I understanding this correctly? This approach I am using is before any extensive research on LLMs, and I’m doing a one-shot query to gauge its first responses. I am choosing a NotebookLM approach for this
The person I really want to try first was Bhagat Singh. I really want to feed it the data of what he published in forms of letters, pamphlets, etc. I came to this approach after considering sources like Wikipedia, which clearly offer a third-person view of his life and can include tons of biases. So again, this is just an initial phase of an experiment without much research and fact-checking for now. The sources I’m using include the list from this link:
https://www.marxists.org/archive/bhagat-singh/index.htm
I have a few questions I’ve asked it already and will share once I am confident I am asking justifiable questions, considering my limited historical and political knowledge. Potential topics to explore include his sudden change in social ideology, especially around religions or God. Another topic is him being that young and able to focus in such a way. I’m also exploring external and internal factors and struggles, to see what insights an AI can glean from such limited data. Can this lead to a future historical discourse through the AI?
I’m hoping to improve the process as I go with better techniques and better data. Feel free to guide me. Do share some of your questions and topics to cover; it’s a very interesting time.
This post was written by me and fixed some phrasing and grammar via AI.
