The Clyde chatbot is kind of a nightmare.

My TTRPG group got selected for the Clyde chatgpt chatbot today. While I initially had fun razzing it with captchas and pictures of monkeys (see below) after getting up from the computer to walk my dog I quickly found a sense of profound frustration seep into me. Even though logically it doesn't operate off any more information than what Discord's privacy policy already allows you to take, it felt like an invasion of privacy.

I don't know if you've tested it yourself but Clyde is an interminable killjoy. ChatGPT appears to have solved all the famous problems such as prompt injection, hallucinations and confidently giving false answers, but in the context of a group conversation this makes it feel like ad copy. Where in a normal conversation someone would respond to "What do you know about x" with "Who's that", Clyde's response is:
"I apologize, y, but I do not have any information about a character named x in this server. If you could provide me with more information about this character, I may be able to look up some information and provide you with a description." In the 40 minutes we had the bot active it said "As an AI, I-" around 20 times. He is an absolute square.

Now I understand that ChatGPT's function is not to be a convincing simulation of a human being. These mechanisms are in place to prevent the issues that would absolutely destroy any attempt at implementing it as a product and in its current state it seems like it would be a perfectly serviceable virtual assistant, copy writer or customer service representative. But the personality embodied in this software is not how I've wanted anyone to act ever. Even at its most natural it has been absolutely soul-draining.

To explain why I'm gonna have to get a little pretentious. Embedded in the internet (and as a result a massive amount of human culture) is a suffocating, almost inescapable relationship between normal people and marketing grads. The culture you are a part of is almost universally open-source and can be analyzed like you would data from a weather station. There's a bunch of people that will be interested it and various things they will do with it but eventually someone goes "Hey, what if we use this data to see how we can talk to you?" and then suddenly you start to see institutions talking in a way that feels like the way you and people you know talk even though they tangibly do not conform to the same set of circumstances that inform why. Pretty quickly people become aware of and tired of this relationship and change in response, becoming more ironic, more staunchly political, maybe messing with the institutions themselves a little bit, but institutions act just as quickly to incorporate this into their messaging as people are to distinguish themselves from it. I've felt like I'm living through a constant search for a language to relate to each other, any language, that we can't have captured and reproduced.

Okay so that was pretentious as promised but I think it was necessary to explain why Clyde peeves me off. That cycle is as true in twitter brand accounts co-oping shitposter culture to increase brand recognition as it is in chatgpt constantly triangulating the perfectly inoffensive authorial voice, and direct messaging between friends is maybe one of the only methods of online communication that isn't subject to it. To drop Clyde, a perfectly smooth homunculus layer-caked in 42 inches of public relations varnish into that makes me feel terrible.
I don't want an inoffensive all-purpose assistant on deck and call when I'm talking to my friends about scifi and the Calvin and Hobbes Smock Arc, I want to be alone. I want a rare reprieve from the constant awareness I'm being managed. Even the people working on this stuff are just a team of humans. Are we not all tired of this? Spiritually? I know I'm aping like 100 different prestige movie climaxes here and probably being overly emotional for a feedback form but I don't think I'm capable of expressing myself in a different way about this.