We will obviously have AGI before we figure out relationship therapy— and likely have AI value alignment before human value alignment.
Practically, though… try having a conversation with chatGPT asking it to translate what you each want to say into “nonviolent communication”. Wow. AI seems way better than humans at human relationship communication…
Honestly the worst thing about non confrontational communication is that someone named a very useful communication technique after something - violence - that is not the thing being avoided.
Yes, I have a colleague whose normal writing tone is confrontational. It’s something they’re aware of and work to remove during the editing process. ChatGPT has made this significantly easier for them:
1st draft
-> prompt “soften the language”
—-> review and revise output
———> prompt “soften again”
————> final draft
Enter more complex prompts or more iterations if you want editing for multiple things, e.g., prompt “suggest edits for clarity, brevity, grammar, and softer language”. You can iterate very quickly.
I'd be curious about what a normal short message from your colleague looks like unedited. I wonder if my writing style is confrontational. I think my client's is.
Practically, though… try having a conversation with chatGPT asking it to translate what you each want to say into “nonviolent communication”. Wow. AI seems way better than humans at human relationship communication…