I read an article by a woman who used an AI companion named George. While she was always acutely aware that she was basically talking to herself, she grew a bit fond of George. Now I still had #LLMStudio lying around. So with the privacy aspect sorted, I decided to see if I couldn't tweak a human voice out of it. This is a chat I had with "Sheba" on my phone earlier, using #smolchat. She studies philosophy, history and art in London and works in a teashop for tourists.

I learned that you can run LLM models locally, even offline. this seemed to solve the privacy headache I have with this technology. so I installed #LLMStudio and then waited for it to download some huge models. I said "hello", the LLM said "hello" back and I haven't touched the thing for weeks now. it turns out I just have no questions for it. I read a lot of wikipedia and regularly search for things online, but at no point have I thought: ooh, an LLM could help with that! what's wrong with me.