NiGPT...?

We use ChatGPT and other large language models far too extensively in our modern lives. With tremendous demand comes rapidly increasing profits for the tech monopolies who run said models. Despite their elaborate, and somewhat deceptive, facades, AI chat bots are surprisingly simple to implement for small-scale non-commercial use.

That is what I attempted today. Using an automated WeChat framework, which, despite not being publically-shown, has previously been successfully utilized in El Servidor and NOGRM, it is possible to connect a desktop WeChat client, and its logged-in account, to a self-repeating script.

However, that is only stage one of technological revolution: from hand to machine. If that is riding a elevator, going from machine to circuit, stage two, is climbing a mountain barefoot.

The code, written in JavaScript as pip mysteriously malfunctioned, was relatively simple to write. If all went well, NiGPT should be live and ready to generate text as I type this sentence right now. Life, however, is not a hacked vidoe game, and the script, to nobody’s surprise, failed.

After reading online documentation and requesting, ironically, ChatGPT to review the code, the cause was found. It had nothing to do with the actual code, WeChat, my cloud server, or even AI itself, and was instead something literally too expensive to fix: billing. Turns out, the OpenAI API, despite its basic models being free to operate, requires a billing plan and five dollar deposit, both of which I did not have.

Thus, I have lost my first battle between broke tech geeks and filthy rich monopolies. My goal now is to find an alternative, and, hopefully, deploy said alternative before the end of this business week.

“If the trend does continue, we’ll eventually have to pay for the oxygen we breathe.”
– Cyberpunk 2077