This is nothing else than the reflexion I am talking about. It is not a reflexion of you, the person chatting with the bot, but an “average” reflexion of what humanity has expressed in the data llms have been trained on.
If a mirror is placed in front of another mirror, the “infinite tunnel” only exists in the mind of the observer.
Abyssian@lemmy.world 1 week ago
Neuroscience News isn’t a conspiracy rag. It’s an article summarizing a research paper, which they link to. So many of you don’t bother to read actual research and instead repeat whatever you’ve seen online about how things work. More parrot than the AI.
certified_expert@lemmy.world 1 week ago
I indeed have not read the link you shared. However, I am not discrediting your comment or source. I apologize if it came across with a hostile tone. On the contrary, thanks for commenting on my post with an interesting article.
My intent is not “parroting” AI-bad. The original post and my follow up comments are the result of… well, a shower. (I am a CS researcher myself, and I have been studying AI fundamentals for the last month)
My point is that the behavior you mention may as well be also part of the “reflexion” of the human behavior. After all, it is just text, attention, feedforward, and repeat on human text. We tend to create “conventions” when we talk: OP, TL;DR, IMHO, ELI5… are some examples of little agreements we take to compress information. We, indirectly reward that behavior. It makes sense that a program that detects and replicates human behavior will also pick that up.
In any case. Thanks for the comment, cheers!