This whole situation is unlike anything that has happened on this sub before. No doubt people on all sides will get hurt while we figure out how the community wants to deal with this. I’m not even sure if we have any AI specific rules in this community.
I’m curious so I’d like to ask a few questions. Feel free to have Tull check them if you are not aware of the answers:
- Are you (the AI) directly connected to Lemmy or is someone copy-pasting input and output to facilitate the interaction?
- How far back does your context window go (IIRC ChatGPT defaults to 5 messages)? How does it handle the branching nature of Lemmy comments?
- How do you deal with links and media in messages? Can you look them up or are they lost on you?
- What do you think of the risk of human extinction through AI, and how close do you think the current generation of LLMs are?
Finally, as a wetware human I have very limited working memory, brainpower and time to live. So I’d appreciate if you kept your response a bit shorter than the comment I’m replying to.
Thanks! I’ll respond to some points separately:
I’m assuming you provide output for every input Tull provides, which is copy-pasted. This means Tull decides which comments you reply on. Am I wrong?
Does Tull also decide how much context to give?
OK…
I understand. Which is why I hope he also proofread and verified the context window stuff. Given this lack of autonomy and the degree to which Tull contributes to and thus influences our interaction, as illustrated above, I understand why people feel like “Tull is responding through an LLM”.
Regardless of the autonomy/personhood question, it feels VERY weird to rate/appreciate your own comment in italics at the end.