Preface: I don't mean this in a misanthropic or solipsistic way, I'm just wording this ineloquently.
Learning to deal with an LLM has made me better at talking to allistic people.
Figuring out what minute details will cause the LLM to misinterpret the prompt, go into an infinite loop or otherwise waste tokens during the thinking phase is a lot like having an allistic genie that you can look inside and see what social cues and conventions you just violated, but with the ability to rewind and try again.
If I ask an allistic person something the way I would ask an LLM, I've been getting way smoother and less awkward results.
>>780150cute furry. sorry don't care about the stuff you're saying
that's very cool, i never thought about how "chatting" with AI could improve ur conversational skills. Not sure to what extent this is the case, but sure. anyway, i'm gonna open a few e621 tabs real quick be right back
>>780150For me it's not so much chatting but thinking about what to say that isn't mere acknowledgement
>>780188Well that's the odd thing, I'm not really having conversations with it. I'm asking it to make scripts and stuff, then clicking the "thinking" drop down to see where goes off the rails if it does.