Chels

I wasn’t intending to talk about generative AI again this week, but I’ve been struck down by a quite severe case of writer’s block. The solution, according to far too many people these days, would be to use AI assistance to help me along the way. The problem is I don’t write for output, I write because I enjoy writing. I choose topics that interest me, I spend time researching, and I try to present my ideas in a logical format (something I’ve often struggled with). Sticking a prompt into an LLM and receiving a premade text with punchy SEO friendly buzzwords just doesn’t bring me the same satisfaction.

My criticism this week isn’t my usual issue – of course, the environmental impact of the increased data centres is frightening, and the news that AI companies are buying up and destroying rare books has made me incredibly frustrated for our future, but I’ve said all of this before, and so many others have too. My issue, this week, is that people seem to be genuinely treating chatGPT as a friend figure.

I spend most of my time in quite anti-AI circles – my friends are artists and writers whose livelihoods are threatened by companies turning from skilled humans to AI for cheap labour – so it’s rare I see that people actually use AI regularly. In fandom spaces, I knew of people using Character AI to ‘speak to their favourite characters’. It seemed strange to me, and a more isolating version of collaborative roleplays that used to exist in fan spaces. Of course, I also saw plenty of discourse surrounding the use of generative AI to ‘create’ fanfiction and fanart, but it seemed to me that the majority of creators were against this, and often generative AI use was frowned upon and self-policed within fandoms. The most alarming thing that I saw was the subculture of people in ‘romantic relationships’ with ChatGPT. It was strange, of course, but I couldn’t help assuming it just wasn’t real.

Can you blame me? I have very little experience with AI, aside from the automatic summaries on Google, so I didn’t really understand how conversations could be had with the service. I understand now on a technical level – AI assistants are programmed to be helpful to the user, and to encourage further interaction (although I still don’t understand why, when most AI services are free to use, they’re encouraging increased use – I’d have assumed they’d want to make money). Increased interaction leads to increased recognition of the needs of the user, and of successful and unsuccessful responses, which would feel like a conversation, or even a relationship. What I don’t understand is what drives a person to desire a relationship with a chatbot.

I’m not great socially – anyone who knows me would agree. I struggle with casual conversation, I’m reluctant to speak to new people, and I have a tendency to isolate myself when I’m overwhelmed. I understand the feeling of loneliness. I have often thought perhaps it would be much easier if things like friendship and conversation were more logical, with less complicated external factors, but I still haven’t sought out friendship with an LLM. This is not to ridicule or berate people who have, it’s more an attempt to understand it myself. Surely, people understand that a chatbot is not conscious – it doesn’t actually have thoughts. It can’t form friendships or romances because it isn’t alive – it’s just computer coding.

The thing is, though, that while my first instinct to understand AI relationships was through the lens of loneliness, many of the people online who have justified their AI romances have insisted that they aren’t lonely outside of AI. Why turn to it then, if not out of loneliness?

My second hypothesis is the convenience of it all. With a friend, you don’t have 24/7 instant access to communication. People sleep, work, and have social lives – they don’t have the time to answer a message instantly. In the instant gratification culture we’ve developed through our addiction to short form content and next-day delivery, it would make sense that we’ve lost our sense of patience. You could ask your friend if they’ve seen the latest rom com or heard that new song, and wait a few hours to hear their opinion, or you could ask the handy chatbot in your phone. The thing is, that chatbot doesn’t actually have an opinion – it’s just regurgitating the opinions of anyone that’s shared them online. It also doesn’t actually know your tastes. AI chatbots can take in information that you share with them, and ‘remember’ by storing past conversations, but they can’t process the information the same way a friend would. 

Who am I to say one shouldn’t treat ChatGPT as your friend, or even your romantic partner? It’s none of my business, and it’s your life. The thing is, these ‘relationships’ aren’t real in the way human to human relationships are. There is no consciousness within Artificial Intelligence programs (no matter how many sci-fi films try to convince us otherwise. I worry that a reliance on AI for friendship could be detrimental to our social skills, to our independence, and to the increased isolation we seem to be experiencing. Even past conveniences are becoming more convenient, rendering us even less capable. Old people love to say that back in their day there were no search engines to look up facts. The thing is, with old-school search engines, one still had to sift through results to find what one looked for. Now, a ‘helpful’ AI summary precedes results, hoping you’ll find what you’re looking for – and if not, it prompts you to ask some further questions – your friend Gemini, or perhaps Copilot, would be happy to chat to you more…