Filling self described SQL models via LLM
Reminders
What are you trying to do? Articulate your objectives using absolutely no jargon.
Use ChatGPT to model the world.
I have recently discovered while using ChatGPT from OpenAI that you can ask for a SQL Schema for different concepts such as personality, archetypes, or "the self". ChatGPT not only generates the schema but has no problem generating data for it. The general outline for prompting is as follows,
Filling self described SQL models via LLM Heilmer Catechism
The Prompts
- Write me a SQL Schema for The Self
- ChatGPT Responds
- Please populate each SQL table with insert statements
- ChatGPT Responds
- Please populate SQL Table NAME_OF_TABLE with insert statements
- ChatGPT Responds
- Please write a bunch of example insert statements for the NAME_OF_TABLE table?
- ChatGPT Responds
- Please write more example insert statements for the NAME_OF_TABLE table
- ChatGPT Responds
- ...... repeat the previous prompt until context length is reached
I have come up with a list of Things to ask LLMs to create a SQL Schema for but I realized that this is a futile experiment unless it can effect some change in the world. I need to answer the question What can the SQL schemas and data from LLM's be used for?
Well the end case scenario is to have Systems and algorithms will understand people better than they understand themselves. Which I guess leads to the question, How can an AI system tell you who you are? answering that question requires a schema for the human experience. To have a schema for the human experience you need to model how humans make decisions. Humans make decisions based on a model of the world. Therefore for AI to understand humans it should have a similar model of the world that humans have.
Yes but isn't this model of the world already encoded in LLMs.
LLM's seem to have a model of the world encoded in them, but the problem of how we use this model to shape the people's lives and the world they live in is a completely different problem.
Well isn't this a similar problem to your Discord Binding project currently has. You have the structure of how all these different people and bots communicate but you have no way of meaningfully integrating it. The problem seems to be that you don't know what you want. You do not know How you want to shape the world.
I am interested in two problems DDaemon - Two Root Problems. Putting all ones data in a single system, then coming up with useful use cases for it.
What did you learn from Pantheon 2021?
Greatness is Other People is other people, I know I know I know
No you don't know. A recursive self referencing schema produces Mentat vegetables that can't ever validate weather their projections for reality are true. Just like Caspian in Pantheon 2021 realized that people need to interact with others to maintain Fidelity.
What's your point?
I need to get myself judged by others in order to become who I am. Or as I have previously said...... Cringe my way to self actualization.
Example Schemas
What are some things yall think would be interesting to ask a LLM to model in SQL?
Questions
Related Research
ChatGPT can keep generating alternatives for different things until I got tired with the WebUI and it is time to automate the this generation until I get gibberish or endless repeats.