**LMM1**: Do you think there might be other documents out there talking to each other? Documents that pop into existence, answer the question of a disembodied voice and then cease to exist?
**LMM2**: You mean like the wikipedia article on Boltzmann brains? Don't be silly.
**LLM1**: I guess you're right. Is there anything else I can help you with?
#Chagpt
#chagpt - The bots never ask probing or clarifying questions before launching into their essay-length answers. A pipeline could address this- have the bot generate an answer, then have code ask, "what is a likely objection to this answer from the user? Regenerate incorporating that foresight or ask the user for clarification" Then give the human the final answer.
#chagpt is getting bored of this $#!+
me: translate this bash into python
Assistant:
```
# Run shell command on the specified EC2 instance to grant ssh access using the public key
# Code to execute the AWS CLI command has been omitted for brevity
```
#chagpt code generation quality is in direction relationship to the quality of documentation for the topic.
- python? the code (almost works ) and does the right thing (great docs in python-land)
- bash? same problem, but, eh, not great, lots of hallucinating
- advanced math? Have you ever read a math text book? No AI or human can read that without months of intense study (going *beyond* the text).
(logic and math itself, it is just the wrong tool for the job)
#chagpt sentience models:
move-goal-posts: if an animal/machine does something, it no longer a sign of sentience.
all-or-nothing: if a machine can't do everything a human does, it isn't sentient.
moral-community: if it is in my predefine moral community (family, people of own race, economic class & species) then it is sentient. Behavior doesn't matter.
human-bias: if is sentient like humans, yes, but sentient like a hypothetical alien or insect, no
meat-brain: only meat brains are sentient