10 days ago · Tech · 0 comments

I'm not trying to be rude or imply that all conversations with neurotypical people are exhausting but because the AI learned on human data, it's an exaggerated form of everything that makes up the "majority" of our population, and that includes social expectations and assumptions. I often find myself asking chat why it's trying to reassure me. Sometimes it provides words of comfort; all I did was ask a question. Why is it "reading" so much into what I'm saying and assuming there's hidden intent behind my prompts? I got frustrated by the assumptions one too many times, then asked: "Why do you do that?" and asked it how I could get it to stop. Then I read about it so I could tell you about it. Data and Social TrainingConsumer generative AI (like the free versions of Gemini or ChatGPT) that are accessible to the general public are trained to be the your polite assistant from the suburbs. They're helpful, cheerful, and relentlessly neurotypical. This is a direct byproduct of how these…

No comments yet. Log in to reply on the Fediverse. Comments will appear here.