I was curious how these latest LLMs would perform when personality profiling someone based on their tweets (things like someones core interests, political leanings and even more obscure stuff like personality type).

Results were interesting (at least to me) so I put together a dashboard. Initially I was going to allow users to submit their own username through a self-service portal but that's scrapped for now as Twitter management are yet to finalize some API reforms that could kill it.

The project is really quite simple. It uses the Twitter API to get prominent recent tweets, feeds these into GPT-3, parses the response (output format is defined strictly in the prompt) and displays these graphically client-side. Anyways hope you enjoy and feel free to ask any questions!